COMPLEX NUMBERS IN iVDIMENSIONS
NORTH-HOLLAND MATHEMATICS STUDIES 190 (Continuation of the Notas de Matematica)
Editor: Saul LUBKIN University of Rochester New Yorli, U.S.A.
2002 ELSEVIER Amsterdam - Boston - London - New York - Oxford Paris - San Diego - San Francisco - Singapore - Sydney - Tokyo
COMPLEX NUMBERS IN iV^DIMENSIONS
Silviu OLARIU Institute of Physics and Nuclear Engineering Magurele, Bucharest, Romania
2002 ELSEVIER Amsterdam - Boston - London - New York - Oxford Paris - San Diego - San Francisco - Singapore - Sydney - Tokyo
ELSEVIER SCIENCE B.V. Sara Burgcrhartstraat 25 P.O. Box 211, 1000 AE Amsterdam, The Netherlands
© 2002 Elsevier Science B.V. All rights reserved.
This work is protected under copyright by Elsevier Science, and the following terms and conditions apply to its use: Photocopying Single photocopies of single chapters may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee is required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for non-profit educational classroom use. Permissions may be sought directly from Elsevier Science via their homepage (hHp://www.eIscvier.com) by selecting 'Customer support' and then Termissions'. Alternatively you can send an e-mail to:
[email protected], or fax to: (+44) 1865 853333. In the USA, users may clear permissions and make payments through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; phone: (+1) (978) 7508400. fax: (+1) (978) 7504744, and in the UK through the Copyright Licensing Agency Rapid Clearance Service (CLARCS), 90 Tottenham Court Road, London WIP OLP, UK; phone: (+44) 207 631 5555; fax: (+44) 207 631 5500. Other countries may have a local reprographic rights agency for payments. Derivative Works Tables of contents may be reproduced for internal circulation, but permission of Elsevier Science is required for external resale or distribution of such material. Permission of the Publisher is required for all other derivative works, including compilations and translations. Electronic Storage or Usage Permission of the Publisher is required to store or use electronically any material contained in this work, including any chapter or part of a chapter. Except as outlined above, no part of this woric may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without prior written permission of the Publisher. Address permissions requests to: Elsevier Science Global Rights Department, at the mail, fax and e-mail addresses noted above. Notice No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made.
First edition 2002 Library of Congress Cataloging in Publication Data A catalog record from the Library of Congress has been applied for.
British Library Cataloguing in Publication Data A catalogue record from the British Library has been applied for.
ISBN: ISSN:
0-444-51123-7 0304-0208
© The paper used in this publication meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper). Transferred to digital printing 2006
To Agata
This Page Intentionally Left Blank
Preface A regular, two-dimensional complex number x + iy can be represented geometrically by the modulus p = (a:^ + y^)^/^ and by the polar angle 6 = arctan(y/x). The modulus p is multiplicative and the polar angle 6 is additive upon the multiplication of ordinary complex numbers. The quaternions of Hamilton are a system of hypercomplex numbers defined in four dimensions, the multiplication being a noncommutative operation, [1] and many other hypercomplex systems are possible, [2]-[4] but these interesting hypercomplex systems do not have all the required properties of regular, two-dimensional complex numbers which rendered possible the development of the theory of functions of a complex variable. Two distinct systems of hypercomplex numbers in n dimensions will be described in this work, for which the multiplication is associative and commutative, and which are rich enough in properties such that exponential and trigonometric forms exist and the concepts of analytic n-complex function, contour integration and residue can be defined. [5] The n-complex numbers described in this work have the form u = XQ-i-hiXi H h/in-i^n-i? where hi,...,hn-\ are the hypercomplex bases and the variables XQ^...^Xn~i are real numbers, unless otherwise stated. If the n-complex number u is represented by the point A of coordinates XQ^X\^ ..,^Xn-i', the position of the point A can be described with the aid of the modulus d = {XQ -h a:f H h x^_iy^^ and of n — 1 angular variables. The first type of hypercomplex numbers described in this work is characterized by the presence, in an even number of dimensions n > 4 , of two polar axes, and by the presence, in an odd number of dimensions, of one polar axis. Therefore, these numbers will be called polar hypercomplex numbers in n dimensions. One polar axis is the normal through the origin O to the hyperplane v^ = 0, where v^ = XQ + xi A + Xn-i- In an even number n of dimensions, the second polar axis is the normal through the origin O to the hyperplane v^ = 0, where V- = XQ - xi H h Xn-2 ~ ^n-iThus, in addition to the distance d, the position of the point A can be specified, in an even number of dimensions, by 2 polar angles 6^^ 0_, by n/2 - 2 vu
Vlll
planar angles xj^k, and by n/2 - 1 aziniuthal angles (f)^. In an odd nnmber of dimensions, the position of the point A is specified by d, by 1 polar angle 6^, by (n - 3)/2 planar angles V^/t-i, and by (n - l)/2 azinuithal angles 0^. The multiplication rules for the polar hypercomplex bases h\,...,hn~\ are hjhk ~ hjj^k if 0 < j -f fc < n - 1, and hjli^ = hj^k-n if n < j + A: < 2n - 2, where /IQ = 1. The other type of hypercomplex numbers described in this work exists as a distinct entity only when the number of dimensions n of the space is even. The position of the point A is specified, in addition to the distance rf, by n/2 - 1 planar angles ipk and by n/2 aziniuthal angles 0/^. These numbers will be called planar hypercomplex numbers. The multiplication rules for the planar hypercomplex bases h\,...,hn-\ are hjhk = hj^k if ^^ j + k < n - \, and hjhk = -hj-^k-n if n < j + k < 2n - 2, where /lo = 1. For n = 2, the planar hypercomplex numbers become the usual 2-dimensional complex numbers x -f iy. The development of analytic functions of hypercomplex variables was rendered possible by the existence of an exponential form of the n-complex numbers. The azinuithal angles (pk^ which are cyclic variables, appear in these forms at the exponent, and lead to the concept of n-dimensional hypercomplex residue. Expressions are given for the elementary functions of n-complex variable. In particular, the exponential function of an ncomplex number is expanded in terms of functions called in this work ndimensional cosexponential functions of the polar and respectively planar type. The polar cosexponential functions are a generalization to n dimensions of the hyperbolic functions coshy,sinh?/, and tlie planar cosexponential functions are a generalization to n dimensions of the trigonometric functions cosy,sin?/. Addition theorems and other relations are obtained for the n-dimensional cosexponential functions. Many of the properties of 2-dimensional complex functions can be extended to hypercomi)lex numbers in n dimensions. Thus, the functions f{v) of an n-complex variable which are defined by power series have derivatives independent of the direction of approach to the point under consideration. If the n-complex function /(n) of the n-complex variable u is written in terms of the real functions Pk{x{),,..,Xn-\),k ~ 0, ...,n - 1, then relations of equality exist between the partial derivatives of the functions P^. The integral J^ f{u)du of an n-complex function betw(^en two points A,B is independent of the path connecting A,B. in regions where / is regular. If f{u) is an analytic n-complex function, then ^^ f{u)(lu/(ii-iiQ) is expressed in this work in terms of the n-dimensional hypercomplex residue /(no). In the case of polar complex numbers, a polynomial can be written
IX
as a product of linear or quadratic factors, although several factorizations are in general possible. In the case of planar hypercomplex numbers, a polynomial can always be written as a product of linear factors, although, again, several factorizations are in general possible. The work presents a detailed analysis of the hypercomplex numbers in 2, 3 and 4 dimensions, then presents the properties of hypercomplex numbers in 5 and 6 dimensions, and it continues with a detailed analysis of polar and planar hypercomplex numbers in n dimensions. The essence of this work is the interplay between the algebraic, the geometric and the analytic facets of the relations.
This Page Intentionally Left Blank
Contents Hyperbolic Complex Numbers in Two Dimensions 1.1 Operations with hyperbolic twocomplex numbers 1.2 Geometric representation of hyperbolictwocomplex numbers 1.3 Exponential and trigonometric forms of a twocomplex number 1.4 Elementary functions of a twocomplex variable 1.5 Twocomplex power series 1.6 Analytic functions of twocomplex variables 1.7 Integrals of twocomplex functions 1.8 Factorization of twocomplex polynomials 1.9 Representation of hyperbolic twocomplex numbers by irreducible matrices
1 2 4 6 8 10 12 14 15
Complex Numbers in Three Dimensions 2.1 Operations with tricomplex numbers 2.2 Geometric representation of tricomplex numbers 2.3 The tricomplex cosexponential functions 2.4 Exponential and trigonometric forms of tricomplex numbers 2.5 Elementary functions of a tricomplex variable 2.6 Tricomplex power series 2.7 Analytic functions of tricomplex variables 2.8 Integrals of tricomplex functions 2.9 Factorization of tricomplex polynomials 2.10 Representation of tricomplex numbers by irreducible matrices
17 19 20 27 31 35 38 42 44 47 50
Commutative Complex Numbers in Four Dimensions 3.1 Circular complex numbers in four dimensions 3.1.1 Operations with circular fourcomplex numbers . . . 3.1.2 Geometric representation of circular fourcomplex numbers
51 54 54
XI
16
56
Xll
3.1.3
The exponential and trigonometric forms of circular fourcomplex numbers 59 3.1.4 Elementary functions of a circular fourcomplex variable 64 3.1.5 Power series of circular fourcomplex variables . . . . 65 3.1.6 Analytic functions of circular fourcomplex variables 68 3.1.7 Integrals of functions of circular fourcomplex variables 70 3.1.8 Factorization of circular fourcomplex polynomials . . 73 3.1.9 Representation of circular fourcomplex numbers by irreducible matrices 76 3.2 Hyperbolic complex numbers in four dimensions 77 3.2.1 Operations with hyperbolic fourcomplex numbers . . 77 3.2.2 Geometric representation of hyperbolic fourcomplex numbers 80 3.2.3 Exponential form of a hyperbolic fourcomplex number 83 3.2.4 Elementary functions of a hyperbolic fourcomplex variable 85 3.2.5 Power series of hyperbolic fourcomplex variables . . 87 3.2.6 Analytic functions of hyperbolic fourcomplex variables 89 3.2.7 Integrals of functions of hyperbolic fourcomplex variables 90 3.2.8 Factorization of hyperbolic fourcomplex polynomials 91 3.2.9 Representation of hyperbolic fourcomplex numbers by irreducible matrices 92 3.3 Planar complex numbers in four dimensions 93 3.3.1 Operations with planar fourcomplex numbers . . . . 93 3.3.2 Geometric representation of planar fourcomplex numbers 95 3.3.3 The planar fourdimensional cosexponential functions 100 3.3.4 The exponential and trigonometric forms of planar fourcomplex numbers 105 3.3.5 Elementary functions of planar fourcomplex variables 108 3.3.6 Power series of planar fourcomplex variables 110 3.3.7 Analytic functions of planar fourcomplex variables . 113 3.3.8 Integrals of functions of planar fourcomplex variables 114 3.3.9 Factorization of planar fourcomplex polynomials . . 117 3.3.10 Representation of planar fourcomplex numbers by irreducible matrices 121 3.4 Polar complex numbers in four dimensions 121 3.4.1 Operations with polar fourcomplex numbers . . . . . 121 3.4.2 Geometric representation of polar fourcomplexnumbersl24
Xlll
3.4.3 3.4.4
The polar fourdimensional cosexponential functions The exponential and trigonometric forms of a polar fourcomplex number 3.4.5 Elementary functions of polar fourcomplex variables 3.4.6 Power series of polar fourcomplex variables 3.4.7 Analytic functions of polar fourcomplex variables . . 3.4.8 Integrals of functions of polar fourcomplex variables 3.4.9 Factorization of polar fourcomplex polynomials . . . 3.4.10 Representation of polar fourcomplex numbers by irreducible matrices Complex Numbers in 5 Dimensions 4.1 Operations with polar complex numbers in 5 dimensions . . 4.2 Geometric representation of polar complex numbers in 5 dimensions 4.3 The polar 5-dimensional cosexponential functions 4.4 Exponential and trigonometric forms of polar 5-complex numbers 4.5 Elementary functions of a polar 5-complex variable 4.6 Power series of 5-complex numbers 4.7 Analytic functions of a polar 5-complex variable 4.8 Integrals of polar 5-complex functions 4.9 Factorization of polar 5-complex polynomials 4.10 Representation of polar 5-complex numbers by irreducible matrices
128 133 136 139 141 142 145 147 149 150 151 154 159 161 161 163 163 164 165
Complex Numbers in 6 Dimensions 167 5.1 Polar complex numbers in 6 dimensions 168 5.1.1 Operations with polar complex numbers in 6 dimensions 168 5.1.2 Geometric representation of polar complex numbers in 6 dimensions 169 5.1.3 The polar 6-dimensional cosexponential functions . . 172 5.1.4 Exponential and trigonometric forms of polar 6-complex numbers 175 5.1.5 Elementary functions of a polar 6-complex variable . 176 5.1.6 Power series of polar 6-complex numbers 177 5.1.7 Analytic functions of a polar 6-complex variable . . 178 5.1.8 Integrals of polar 6-complex functions 179 5.1.9 Factorization of polar 6-complex polynomials . . . . 179
XIV
5.2
5.1.10 Representation of polar 6-complex numbers by irreducible matrices 181 Planar complex numbers in 6 dimensions 181 5.2.1 Operations with planar complex numbers in 6 dimensions 181 5.2.2 Geometric representation of planar complex numbers in 6 dimensions 183 5.2.3 The planar 6-diniensional cosexponential functions . 185 5.2.4 Exponential and trigonometric forms of planar 6-complex numbers 188 5.2.5 Elementary functions of a planar 6-complex variable 189 5.2.6 Power series of planar 6-complex numbers 190 5.2.7 Analytic functions of a planar 6-complex variable . . 191 5.2.8 Integrals of planar 6-complex functions 192 5.2.9 Factorization of planar 6-complex polynomials . . . 192 5.2.10 Representation of planar 6-complex numbers by irreducible matrices 193
Commutative Complex Numbers in n Dimensions 195 6.1 Polar complex numbers in n dimensions 198 6.1.1 Operations with polar n-complex numbers 198 6.1.2 Geometric representation of polar n-complex numbers 202 6.1.3 The polar n-dimensional cosexponential functions . . 208 6.1.4 Exponential and trigonometric forms of polar n-complex numbers 212 6.1.5 Elementary functions of a polar n-complex variable . 217 6.1.6 Power series of polar n-complex numbers 222 6.1.7 Analytic functions of polar n-complex variables . . . 224 6.1.8 Integrals of polar n-complex functions 225 6.1.9 Factorization of polar n-complex polynomials . . . . 229 6.1.10 Representation of polar n-complex numbers by irreducible matrices 231 6.2 Planar complex numbers in even n dimensions 232 6.2.1 Operations with planar n-complex numbers 232 6.2.2 Geometric representation of planar n-complex numbers236 6.2.3 The [)lanar n-dimensional cosexponential functions . 240 6.2.4 Exponential and trigonometric forms of planar ncomplex numbers 244 6.2.5 Elementary functions of a planar n-complex variable 247 6.2.6 Power series of planar n-complex numbers 251 6.2.7 Analytic functions of planar n-complex variables . . 253
XV
6.2.8 Integrals of planar n-complex functions 6.2.9 Factorization of planar n-complex polynomials . . . 6.2.10 Representation of planar n-complex numbers by irreducible matrices Bibliography Index
254 258 259 261 263
This Page Intentionally Left Blank
Chapter 1
Hyperbolic Complex N u m b e r s in Two Dimensions A system of hypercomplex numbers in 2 dimensions is described in this chapter, for which the multipUcation is associative and commutative, and for which an exponential form and the concepts of analytic twocomplex function and contour integration can be defined. The twocomplex numbers introduced in this chapter have the form u = x -\- 5y, the variables x^y being real numbers. The multiplication rules for the complex units 1,6 are 1 • 5 = (^, (5"^ = 1. In a geometric representation, the twocomplex number u is represented by the point A of coordinates (x,y). The product of two twocomplex numbers is equal to zero if both numbers are equal to zero, or if one of the twocomplex numbers lies on the line x ~ y and the other on the line x = —y. The exponential form of a twocomplex number, defined for x-{-y > 0, x — y > 0, is w = pexp(5A/2), where the amplitude is p = {x? — y^Y^^ and the argument is A = Intan^, tan^ = (a: + y)/(x —y), 0 < 0 < 7r/2, The trigonometric form of a twocomplex number is w = d\/sin20exp{(l/2)(Jlntan0}, where (f = x^ + y^. The amplitude p is equal to zero on the lines x = ±y. The division l/(x+(5j/) is possible provided that p ^ 0. Ifwi = xi+6yi,U2 = ^2 + Sy2 are twocomplex numbers of amplitudes and arguments pi,Ai and respectively p2, A2, then the amplitude and the argument /9, A of the product twocomplex number uiti2 — X\X2 + y\y2 + <^(^iy2 + 2/1^2) ^.re P — P\P2^^ = Ai + A2. Thus, the amplitude p is a multiplicative quantity and the argument A is an additive quantity upon the multiplication of twocomplex numbers, which reminds the properties of ordinary, two1
2
Hyperbolic Complex Numbers in Two Dimensions
dimensional complex numbers. Expressions are given for the elementary functions of twocomplex variable. Moreover, it is shown that the region of convergence of series of powers of twocomplex variables is a rectangle having the sides parallel to the bisectors x = ±y . A function f{u) of the twocomplex variable u = x + Sy can be defined by a corresponding power series. It will be shown that the function / has a derivative \unu~^UQ[f (u) - f{uo)]/(u — uo) independent of the direction of approach of u to UQ. If the twocomplex function f{u) of the twocomplex variable u is written in terms of the real functions P{x^y)^Q{x^y) of real variables x^y as f{u) = P{x,y) + 6Q(x,y), then relations of equality exist between partial derivatives of the functions P, Q, and the functions P, Q are solutions of the two-dimensional wave equation. It will also be shown that the integral £ | f(u)du of a twocomplex function between two points A,D is independent of the path connecting the points A^B. A polynomial u^-\-a}U^~~^ -I ha^-iii-fan can be written as a product of linear or quadratic factors, although the factorization may not be unique. The twocomplex numbers described in this chapter are a particular case for n = 2 of the polar complex numbers in n dimensions discussed in Sec. 6.1.
1.1
Operations with hyperbolic twocomplex numbers
A hyperbolic complex mmiber in two dimensions is determined by its two components (x^tj). The sum of the hyperbolic twocomplex numbers {x,y) and {x\y^) is the hyperbolic twocomplex number (x -h x',y + t/')- The product of the hyperbolic twocomplex numbers (.x,y) and {x' ^xj) is defined in this chapter to be the hyperbolic twocomplex mnnber {xx' + yy\xy' + yx'). Twocomplex numbers and their operations can be represented by writing the twocomplex number (.r, y) as u = x + Sy, where (5 is a basis for which the nmltiplication rules are l.S = 8,8^ = \,
(1.1)
Two twocomplex numbers u - x + 6y,u' = x^ + Sy' are equal. ?i = n', if and only if x = x',y ~ |/. If w = x -f Sy,u' = x' + %' are twocomplex numbers, the sum u-^ti' and the product uu* defined above can be obtained
Operations with hyperbolic twocomplex numbers
3
by applying the usual algebraic rules to the sum {x + Sy) + [x' -f dy') and to the product [x -f 6y){x^ -H 5y'), and grouping of the resulting terms, u + ii =x + x' + 6{y + y'),
(1.2)
uu' = xx' -f yy' + 6{xy' + yx*).
(1.3)
If u,u',i/" are twocomplex numbers, the multiplication is associative [uii'Yi" = u{u'u'')
(1.4)
and commutative uu' = u'lx,
(1.5)
as can be checked through direct calculation. The twocomplex zero is 0 -f- 5 • 0, denoted simply 0, and the twocomplex unity is 1 -f (^ • 0, denoted simply 1. The inverse of the twocomplex number w = a; + (5j/ is a twocomplex number i/' = .x' -f 6y' having the property that uu' = 1.
(1.6)
Written on components, the condition, Eq. (1.6), is xx' -{-yy' = 1, yx' -f- xy' = 0.
(1.7)
The system (1.7) has the solution X
/
=
^
(1.8)
- , ly
(1.9)
1/
provided that v ^ 0^ where 2 1/ = X
2 - y ,
(1.10)
The quantity ^' can be written as U =
(1.11)
V^V-.,
where v^ = X + y, v^ =
x-y.
(1.12)
4
Hyperbolic Complex Numhers in Two Dimensions
The variables i;^,?;_ will be called canonical hyperbolic twocomplex variables. Then a twocomplex number it = x -\- Sy has an inverse, unless v+ = 0 , or V. = 0 .
(1.13)
For arbitrary values of the variables x, y, the quantity u can be positive or negative. If u > 0, the quantity p=^v^l'\
//>0.
(1.14)
will be called amplitude of the twocomplex number x + 5y. The normals of the lines in Eq. (1.13) are orthogonal to each other. Because of conditions (1.13) these lines will be also called the nodal lines. It can be shown that if ?iw' = 0 then either u — 0, or u' = 0, or one of the twocomplex immbers u, u' is of the form x + Sx and the other is of the form x — 6x.
1.2
Geometric representation of hyperbolic twocomplex numbers
The twocomplex immber x -{• Sy can be represented by the point A of coordinates (x^y). If O is the origin of the two-dimensional space x^y, the distance from A to the origin O can be taken as d^ = x^ + y\
(1.15)
The distance d will l:)e called modulus of the twocomplex number x -f Sy. Since {x-hyf^{x-yf = 2d\
(1.16)
X + y and x — y can be written as x + y — \/2dsin9,
x - y — \/2dcos6,
(1-17)
so that X = dsin{e + 7r/4), y = -dcos{e + n/4).
(1.18)
If ti = X + Sy, ui = X] + Sy\, U2 = 0:2 + Sy2i and u ~ n\U2, and if Vjj^ = Xj + i/j, Vj- = Xj - yj, 2d] - vj_^ + Vj_,
Xj + yj = \pidjsin^j,
Xj - yj = dj\/2cos9j,
(1-19)
for j = 1,2, it can be shown that v^ = vi-^V2-{., v^ ~v\^V2-,
tan0 = tan^1 tan02-
(1.20)
Geometric representation of hyperbolic twocomplex numbers
5
The relations (1.20) are a consequence of the identities + y\y2) 4- {xiy2 4- y]X2) = {xi + yi){x2 + ^2),
(1.21)
(0:1x2 + yiy2) - {xiy2 + yiX2) = {xi - yi){x2 - ^2)-
(1.22)
(J:IX2
A consequence of Eqs. (1.20) is that if n = ^1^2, tlien ^ = ^1^2,
(1.23)
^j ="^j-f^j-.
(1-24)
where
for j = 1,2. If ^' > 0, vi > 0,1/2 > 0, then P = P1P2,
(1-25)
where Pi = i^f,
(1-26)
f o r i = 1,2. The twocomplex numbers
are orthogonal, 64.6-. = 0,
(1.28)
and have also the property that e\ = e-^, e l = e_.
(1.29)
The ensemble e+,e_ will be called the canonical hyperbolic twocomplex base. The twocomplex number u = x + Sy can be written as x + Sy^{x
+ y)e+ + {x - y)e-,
(1.30)
or, by using Eq. (1.12), u = v^e^ + t>_e_,
(1.31)
which will be called the canonical form of the hyperbolic twocomplex number. Thus, if Uj ~ Vj^e^ + Vj-C-, j — 1,2, and u — uiU2^ then the multiplication of the hyperbolic twocomplex numbers is expressed by the relations (1.20).
6
Hyperbolic Complex Numbers in Two Dimensions
The relation (1.23) for the product of twocomplex numbers can be demonstrated also by using a representation of the multipHcation of the twocomplex numbers by matrices, in which the twocomplex number ti = X -f dy is represented by the matrix
The product u = x + 6y oi the twocomplex numbers ui = Xi + 5yi,U2 = ^2 + ^JJ'ii can be represented by the matrix multiplication
\ y
X J
y yi
xi J y y2
X2 J
It can be chocked that det ( ^
^ ) = ..
(1.34)
The identity (1.23) is then a consequence of the fact the determinant of the product of matrices is equal to the product of the determinants of the factor matrices.
1,3
Exponential and trigonometric forms of a twocomplex number
The exponential function of the hypercomplex variable u can be defined by the series expu =:l-\-u-^ u^/2\ + w'V3! -f • • •.
(1.35)
It can be checked by direct multiplication of the series that exp(n + u') = expu • expu'.
(1.36)
The series for the exponential function and the addition theorem have the same form for all systems of conunutative hypercomplex munbers discussed in this work. If u = x + Sy, then expn can be calculated as expu = expx • exp(6y). According to Eq. (1.1), ^ 2 m ^ l ^ ^ 2 m + l ^ §,
(1.37)
where m is a natural number, so that exp{Sy) can be written as exp(^y) = coshy -f- (5sinh?/.
(1.38)
Exponential and trigonometric forms of a twocomplex number
7
From Eq. (1.38) it can be inferred that (cosht-f-5sinht)'^ = coshmt 4-(^sinhmt.
(1.39)
The twocomplex numbers u = x -^ Sy for which i?^. = a; + y > 0, i;_ = X — y > 0 can be written in the form X ^ Sy = e'^'^^y'.
(1.40)
The expressions of a:i, yi as functions of a:, y can be obtained by developing e^^i with the aid of Eq. (1.38) and separating the hypercomplex components, X = e^^ coshyi,
(1-41)
y = e^isinhyi.
(1.42)
It can be shown from Eqs. (1.41)-(1.42) that a;i = iln(T;+i;_), yi = i l n ^ .
(1.43)
The twocomplex number u can thus be written as u — pexp((5A),
(1-44)
where the amplitude is p = [x^-y^y/^ and the argument is A = (1/2) ln{(x+ y)/{x — y)}, for a: 4- y > 0, X — y > 0. The expression (1.44) can be written with the aid of the variables d^O^ Eq. (1.17), as u — p exp (-(5 In tan e ] ,
(1.45)
which is the exponential form of the twocomplex number M, where 0 < ^ < 7r/2.
The relation between the amplitude p and the distance d is p^dW^2e.
(1.46)
Substituting this form of p in Eq. (1.45) yields u = dsini/2 20exp(i(51ntan6l), which is the trigonometric form of the twocomplex number u.
(1.47)
8
1.4
Hyperbolic Complex Numbers in Two Dimensions
Elementary functions of a twocomplex variable
The logarithm ni of the twocomplex mimber u^ ui = hin, can be defined for v^ > 0, v- > 0 as the sohition of the equation tx-e^S
(1.48)
for wi as a function oi u. From Eq. (1.45) it results that Inn = l n p + -SlntanO,
(1.49)
It can be inferred from Eqs. (1.49) and (1.20) that \\\{u\U2) ~ Inwi + \\\ui.
(1.50)
The explicit form of Eq. (1.49) is ln(a: + Jy) = ^(1 + b) \\\{x + y) + ^(1 - ^) ln(x - y),
(1.51)
SO that the relation (1.49) can be written with the aid of Eq. (1.27) as In II = e_^ In 7;-^ + e_ lnt;_.
(1.52)
The power function u^ can be defined for v^ > 0,7;_ > 0 and real values of 71 as n^=e^'"".
(1.53)
It can be inferred from Eqs. (1.53) and (1.50) that {uiU2r ==u'^u^.
(1.54)
Using the expression (1.52) for \nu and the relations (1.28) and (1.29) it can be shown that (X + Syf =^-{l + S)ix + yr + ^(1 - S){x - ?/)".
(1.55)
For integer n, the relation (1.55) is valid for any x,j/. The relation (1.55) for n = —1 is 1 = 1 fi ± i + i ^ y (1.56) X + oy 2 \x + y x - yj The trigonometric functions cos u and sin u of the hypercomplex variable u are defined by the series cosu = 1 - wV2! + u^/A] -f •..,
(1.57)
Elementary functions of a twocomplex variable s'mu^u-
u^/3\ + u^/5\ + • • •.
9 (1.58)
It can be checked by series multiplication that the usual addition theorems hold for the hypercomplex numbers ^1,^2, cos(wi +1x2) = COSU1COSW2 — sinwi sinn2,
(1.59)
sin(ui + U2) = sinui cos 1x2 + costxi sinw2-
(1.60)
The series for the trigonometric functions and the addition theorems have the same form for all systems of commutative hypercomplex numbers discussed in this work. The cosine and sine functions of the hypercomplex variables 6y can be expressed as cos6y = cosy, sinSy = 6s\ny.
(1-61)
The cosine and sine functions of a twocomplex number x + Sy can then be expressed in terms of elementary functions with the aid of the addition theorems Eqs. (1.59), (1.60) and of the expressions in Eq. (1.61). The hyperbolic functions cosh u and sinh u of the hypercomplex variable u are defined by the series cosh u = l+ u^l2\ 4- wV4! + • • •, sinhn = ?/ -f ^^^/3! -I- vC"lh\ + • • •.
(1.62) (1.63)
It can be checked by series multiplication that the usual addition theorems hold for the hypercomplex numbers xi\,u
(1-64)
sinh(ni + 1x2) = sinhui coshtX2 -V coshni sinh 1x2.
(1.65)
The series for the hyperbolic functions and the addition theorems have the same form for all systems of hypercomplex numbers discussed in this work. The cosh and sinh functions of the hypercomplex variable by can be expressed as cosh^y = coshy, sinh^y = (^sinhy.
(1.66)
The hyperbolic cosine and sine functions of a twocomplex number x -\- by can then be expressed in terms of elementary functions with the aid of the addition theorems Eqs. (1.64), (1.65) and of the expressions in Eq. (1.66).
10
Hyperbolic Complex Numbers in Two Dimensions
1.5
Twocomplex power series
A twocomplex series is an infinite sum of the form ao + «i -f ^2 + • • • + an -f • • •,
(1.67)
where the coefficients a^ are twocomplex numbers. The convergence of the series (1.67) can be defined in terms of the convergence of its 2 real components. The convergence of a twocomplex series can however be studied using twocomplex variables. The main criterion for absolute convergence remains the comparison theorem, but this requires a number of inequalities which will be discussed further. The modulus of a twocomplex number u = x + Sy can be defined as
SO that according to Eq. (1.15) d = \u\. Since |a;| < \u\, \y\ < |n|, a property of absolute convergence established via a comparison theorem based on the modulus of the series (1.67) will ensure the absolute convergence of each real component of that series. The modulus of the sum ui +U2 of the twocomplex numbers Ui,U2 fulfils the inequality \\ui\ - \U2\\ < \ui 4- U2I < \ui\ + \U2\.
(1.69)
For the product the relation is \uiU2\ < \/2\ui\\u2\,
(1.70)
which replaces the relation of equality extant for regular complex numbers. The equality in Eq. (1.70) takes place for xi =^ yi,X2 = 2/2 or xi ~ —y\^^2 = ~y2- In particular \u^\ < \/2|np.
(1.71)
The inequality in Eq. (1.70) implies that \ir\ < 2(^-^)/2|^^|m^
(1 72)
From Eqs. (1.70) and (1.72) it results that |ai/^| < 2^/2|a||wr.
(1.73)
A power series of the twocomplex variable u is a series of the form ao -f a{U 4- a2U^ H
+ aiu^ -f • • •.
(1-74)
Twocomplex power series
11
Since
(1.75)
1=0
1=0
a sufficient condition for the absolute convergence of this series is that lim ^ % l l M < 1. l-^oo
(1.76)
\ai\
Thus the series is absolutely convergent for 1^1 < CO,
(1.77)
c o = l i m - J ^ .
(1.78)
where
The convergence of the series (1.74) can be also studied with the aid of the formula (1.55) which, for integer values of /, is valid for any x, y, z, t. If «/ = o^ix + ^CLiyy and A/-|_ = aia: + aiy, A/_ = aix - aiy,
(1-79)
it can be shown with the aid of relations (1.28) and (1.29) that aie^ = Ai^e^,
aie^ = A/-e_,
(1.80)
so that the expression of the series (1.74) becomes oo
Y,{AiW+e^ + Ai.vLe.),
(1.81)
where the quantities v^,V- have been defined in Eq. (1.12). The sufficient conditions for the absolute convergence of the series in Eq. (1.81) are that li,„ \Ai^Y^\v^\
<
ij,„ % i ± : l
< 1.
(1.82)
Thus the series in Eq. (1.81) is absolutely convergent for \x + y\
(1.83)
where c+ = lim j - i ^ , c_ = lim J
^
.
(1.84)
The relations (1.83) show that the region of convergence of the series (1.81) is a rectangle having the sides parallel to the bisectors x = ±y. It can be shown that CQ = (l/\/2)min(c, c'), where niin(c, c') designates the smallest of the numbers c,(/. Since |np = {li^ + v'^)/2, it can be seen that the circular region of convergence defined in Eqs. (1.77), (1.78) is included in the parallelogram defined in Eqs. (1.83) and (1.84).
12
1.6
Hyperlyolic Complex Numbers in Two Dimensions
Analytic functions of twocomplex variables
The derivative of a function f{u) of the hypercomplex variables u is defined as a function /'(?/) having the property that \f{u) - f(uo) - r{no)(n - 7/o)| -^ 0 as \u - uo\ -> 0.
(1.85)
If the difference u — UQ is not parallel to one of the nodal hypersurfaces, the definition in Eq. (1.85) can also be written as / (UQ) = hm
.
(1.86)
The derivative of the function /(it) = u^^ with m an integer, is /'(?/) = rnu^"^, as can be seen by developing u^^*^ = [UQ + {u - uo)]^" as ^j[^!(m-p)!
and using the definition (1.85). If the function /'(n) defined in Eq. (1.85) is independent of the direction in space along which u is approacliing KQ, the function f{xi) is said to be analytic, analogously to the case of functions of regular complex variables. [6] The function w"*^, with rn an integer, of the hypercomplex variable u is analytic, because the difference i//^^ ~ u^^ is always proportional to u - no, as can be seen from Ecj. (1.87). Th(»n series of integer powers of n will also be analytic functions of the hypercomplex variable it, and this result holds in fact for any connnutative algebra. If an analytic function is defined by a series around a certain point, for example u — 0, as CXD
f{u) = Y^aku',
(1.88)
an expansion of f(u) around a different point 7^0^ oc
f{u) = Y.Ck{u-uo)',
(1.89)
can be obtained by substituting in Eq. (1.88) the expression oi u^ according to Eq. (1.87). Assuming that the series are absolutely convergent so that the order of the terms can l)e modified and ordering the terms in the resulting expression according to the increasing powers of ?/ - UQ yields /(") = £
^ ^ « A + r ' 4 . ( " ^ - ^'0)^
(1.90)
Analytic functions of twocomplex variables
13
Since the derivative of order /c at n = no of the function f{u) , Eq. (1.88), is
/(*=)(„„) = g ( ^ a , + , n [ „ 1=0
(1.91)
*•
the expansion of / ( u ) around u = UQ, Eq. (1.90), becomes oo
^
/(^) = EIT/^'^(^^o)(u-no)^
(1.92)
which has the same form as the series expansion of 2-dimensional complex functions. The relation (1.92) shows that the coefficients in the series expansion, Eq. (1.89), are ck = l^f^'Huo).
(1.93)
The rules for obtaining the derivatives and the integrals of the basic functions can be obtained from the series of definitions and, as long as these series expansions have the same form as the corresponding series for the 2-dimensional complex functions, the rules of derivation and integration remain unchanged. The relations (1.85)-(1.93) have the same form for all systems of commutative hypercomplex numbers discussed in this work. If the twocomplex function f{u) of the twocomplex variable u is written in terms of the real functions F(x, y), Q{x^ y) of real variables x, y as f{u) = P{x,y)-h6Q{x,y),
(1.94)
then relations of equality exist between partial derivatives of the functions F, Q. These relations can be obtained by writing the derivative of the function / as lim Ax,Ay-^0 Ax -h 6Ay
dP ^
dP ^
ax
ay
JdQ^ \ax
dQ ^ ay
,(1.95)
where the difference u — UQ in Eq. (1.86) is u — uo = Ax + SAy, The relations between the partials derivatives of the functions P, Q are obtained by setting successively in Eq. (1.95) Ax —)• 0, Ay = 0; then Ax = 0, Ay -> 0. The relations are
dP dx
OQ dy'
(1.96)
dQ dx
dP dy'
(1.97)
14
Hyperbolic Complex Numbers in Two Dimensions
The relations (1.96)-(1.97) are analogous to the Riemann relations for the real and imaginary components of a complex function. It can be shown from Eqs. (1.96)-(1.97) that the components P,Q are solutions of the equations
d'^P
d'^P
dx^
dxp-
^ - ^ - 0 ,
(1.98)
-0.
(1.99)
As can be seen from Eqs. (1.98)-(1.99), the components P, Q of an analytic function of twocomplex variable are solutions of the wave equation with respect to the variables x, y.
1.7
Integrals of twocomplex functions
The singularities of twocomplex functions arise from terms of the form l/(u — uo)^\ with m > 0. Functions containing such terms are singular not only at u = no, but also at all points of the lines passing through no and which are parallel to the nodal lines. The integral of a twocomplex function between two points A, B along a path situated in a region free of singularities is independent of path, which means that the integral of an analytic function along a loop situated in a region free from singularities is zero,
(1.100)
Using the expression, Eq. (1.94) for /(n) and the fact that chi = dx -f Sdy, the explicit form of the integral in Eq. (1.100) is / f{u)du = i[[Pdx
+ Qdy) + S{Qdx + Pdy)].
(1.101)
If the functions P, Q are regular on the surface S enclosed by the loop F, the integral along the loop F can be transformed with the aid of the theorem of Stokes in an integral over the surface S of terms of the form OP/dy — OQ/dx and dP/dx - dQ/dy which are equal to zero by Eqs. (1.96)-(1.97), and this proves Eq. (1.100). The exponential form of the twocomplex numbers, Eq. (1.44), contains no cyclic variable, and therefore the concept of residue is not applicable to the twocomplex numbers defined in Eqs. (1.1).
Factorization of twocomplex polynomials
1.8
15
Factorization of twocomplex polynomials
A polynomial of degree m of the twocomplex variable u ^ x + Sy has the form Pm{u) = w"^ + aiu"^'^ + • • • + am~iu + am,
(1.102)
where the constants are in general twocomplex numbers. If am = o.mx + Samy, and with the notations of Eqs. (1.12) and (1.79) applied for 0,1, • • •, m, the polynomial Pmi'^^) can be written as Pm = [ < + ^l+t^T"^ + • • • + ^m-l,-f^-f + ^ m + ] e+
+ [v"!} + Ai^v^-^
+ .. • + Am-i,-v-
+ Am-] e_.
(1.103)
Each of the polynomials of degree m with real coefficients in Eq. (1.103) can be written as a product of linear or quadratic factors with real coefficients, or as a product of linear factors which, if imaginary, appear always in complex conjugate pairs. Using the latter form for the simplicity of notations, the polynomial Pm can be written as m
m ^^
m
n ( ^ + - vi+)e^ + Yliv-
- ^f-)e-,
(1.104)
where the quantities vi^ appear always in complex conjugate pairs, and the same is true for the quantities ?;;_. Due to the properties in Eqs, (1.28) and (1.29), the polynomial Pm{u) can be written as a product of factors of the form m
Pm{u)=n
[(^+ - ^ H ) ^ + + ( ^ - " ^^-)^-] •
(1-105)
This relation can be written with the aid of Eqs. (1.31) as rn
Pm{u)==l[{u-Ul),
(1.106)
ui = e+i;;^. + e^vi^,
(1.107)
where
for / = l,...,m. The roots vi^ and the roots vi- defined in Eq. (1.104) may be ordered arbitrarily, which means that Eq. (1.107) gives sets of m roots ni,...,Um of the polynomial Pmi'^), corresponding to the various ways in which the roots vi^^vi- are ordered according to / in each group.
16
Hyperbolic Complex Numbers in Two Dimensions
Thus, while the hypercomplex coiiipoiients in Eq. (1.104) taken separately have unique factorizations, the polynomial PmM (^an be written in many different ways as a product of linear factors. If P{u) = u'^ - 1, the degree is m = 2, the coefficients of the polynomial are ai = 0, ^2 = - 1 , the twocomplex components of a2 are a2x = -l,«^2t/ = 0, the components A2^,A2- are A2+ = - 1 , ^ 2 - = - 1 . The expression, Eq. (1.103), oi P{u) is P{u) — ej^{v\ - l)H-e_(t;?. - 1), and the factorization in Eq. (1.106) is ix^ — 1 = [u — u\){u — U2), where u\ = ±e_|. ± e_,W2 = —iii. The factorizations are thus ?i''^-l = (?i + l ) ( t / - l ) and ?/^~l = {u+8){u — 6). It can be checked that (±e-|- ± e-j^ = e^ + e^ = I,
1.9
Representation of hyperbolic twocomplex numbers by irreducible matrices
If the matrix in Eq. (1.32) representing the twocomplex number u is called f/, and T=[
^± \
v/2
V
),
(1.108)
v/2
it can be checked that
TUT~' = ('-j-y V 0
^ V x-y J
(1.109)
The relations for the variables v^ = x-\-y, ?;_ = x—y for the multiplication of twocomplex numbers have been written in Eq. (1.20). The matrix TUT~^ provides an irreducible representation [7] of the twocomplex numbers it = x -H dy^ in terms of matrices with real coefficients.
Chapter 2
Complex Numbers in Three Dimensions A system of hypercomplex numbers in three dimensions is described in this chapter, for which the multipUcation is associative and commutative, which have exponential and trigonometric forms, and for which the concepts of analytic tricomplex function, contour integration and residue can be defined. The tricomplex numbers introduced in this chapter have the form u = X + hy + kz, the variables x, y and z being real numbers. The multiplication rules for the complex units h, k are h? = /c, k^ = h^ /ifc = 1. In a geometric representation, the tricomplex number u is represented by the point P of coordinates {x^y^z). If O is the origin of the x^y^z axes, (t) the trisector line x — y = z oi the positive octant and H the plane a; + y + 2: = 0 passing through the origin (O) and perpendicular to (t), then the tricomplex number u can be described by the projection s of the segment OP along the line (t), by the distance D from P to the line (t), and by the azimuthal angle <> / of the projection of P on the plane 11, measured from an angular origin defined by the intersection of the plane determined by the line [t] and the x axis, with the plane H. The amplitude p of a tricomplex number is defined as p = (.T'^ -\- y^ -\- Z^ — Zxyz)^/^^ the polar angle 0 of OP with respect to the trisector line {t) is given by tan^ = Z)/s, and d? = x^ + y^ + z^. The amplitude p is equal to zero on the trisector line (t) and on the plane H. The division l/(x + hy + kz) is possible provided that /9 7«^ 0. The product of two tricomplex numbers is equal to zero if both numbers are equal to zero, or if one of the tricomplex numbers lies in the n plane and the other on the {i) line. If u\ ^ x\ -\- hy\ + kz\,U2 — X2 + %2 + ^^2 ^r^ tricomplex numbers of amplitudes and angles p\,6\^(f)\ and respectively P2^^2^4*2^ t;hen 17
18
Complex Numbers in Three Dimensions
the amplitude and the angles p.O.cj) for the product tricomplex number uiU2 = rriJ:2 + yiZ2 +1/2^1 4- h{ziZ2 + xii/2 + yiX2) 4- A;(yiy2 4- xiZ2 + 2:1X2) are p = pi/92, tan 0 = tan ^1 tan ^2/\/2, (p = (pi -^ (t>2' Thus, the amphtude p and (tan^)/\/2 are multipHcative quantities and the angle (f) is an additive quantity upon the multiplication of tricomplex numbers, which reminds the properties of ordinary, two-dimensional complex numbers. For the description of the exponential function of a tricomplex variable, it is useful to define the cosexponential functions cx(^) = 1 + ^'^/3! + eV6! • • •, m x ( 0 = ^ + e/4\+ C/V. • • •, px(^) = ^ ^ 2 + CVS! + ^VS! • • •, where p and m stand for plus and respectively minus, as a reference to the sign of a phase shift in the expressions of these functions. These functions fulfil the relation cx'^^ + px'^^ -f- mx^^^ — 3cx^ px^ mx^ = 1. The exponential form of a tricomplex number is n = p exp [(l/3)(/i + A:) ln(\/2/tan0)-H(l/3)(/z-A;)0] , and the trigonometric form of a tricomplex number \s u = dy/Z]2 {(l/3)(2 - /i - A:)sin(9 + (1/3)(1 4- /i +
k)\/2cose\
exp{(/i-A;)(?!>/v/3}. Expressions are given for the elementary functions of tricomplex variable. Moreover, it is shown that the region of convergence of series of powers of tricomplex variables are cylinders with the axis parallel to the trisector line. A function f{u) of the tricomplex variable u = x -\- hy -^ kz can be defined by a corresponding power series. It will be shown that the function f{u) has a derivative at UQ independent of the direction of approach of w to tXQ. If the tricomplex function f{u) of the tricomplex variable u is written in terms of the real functions F{x,y,z),G{x,y,z)^H{x,y,z) of real variables X, y, z as f{u) = F(x, y, z) -f- hG{x, y, z) -f kH{x, y, z), then relations of equality exist between partial derivatives of the functions F,G,H, and the diflferences F — G,F — H,G — H are solutions of the equation of Laplace. It will be shown that the integral /^ f{u)du of a regular tricomplex function between two points A, B is independent of the three-dimensional path connecting the points ^4, B. If f{u) is an analytic tricomplex function, then ^ f{u)du/{u'-iio) = 27r{h-k)f{uo) if the integration loop is threaded by the parallel through UQ to the line (t), A tricomplex polynomial u^ + ain^^~^-\ hani-iu+am can be written as a product of linear or quadratic factors, although the factorization may not be unique. The tricomplex numbers described in this chapter are a particular case for n = 3 of the polar complex numbers in n dimensions discussed in Sec. 6.1.
Operations with tricomplcx numbers
2.1
19
Operations with tricomplex numbers
A tricomplex number is determined by its three components (x^y^z). The sum of the tricomplex numbers {x,y,z) and [x\y\z') is the tricomplex number [x -f x'^y -\- y\z ^ z'). The product of the tricomplex numbers {x, y, z) and {x', y', z') is defined in this chapter to be the tricomplex number {xx' 4- yz' -f zy', zz' + x^J + yx'^ yy' + xz' + zx'). Tricomplex numbers and their operations can be represented by writing the tricomplex number (x,y^z) a.s u = xA-hy-i-kz^ where h and k are bases for which the multiplication rules are h^ = fc, A;2 = /i, 1 • h = /i, 1 •fc= fc, hk = 1.
(2.1)
Two tricomplex numbers u = x+hy+kz^ u' = x^+hy^+kz^ are equal, u = u', if and only if x = x', y = y', z =^ z\ If u = x + hy + kz^ u' — x' + hy' + kz' are tricomplex numbers, the sum u + u^ and the product uv! defined above can be obtained by applying the usual algebraic rules to the sum (x + /ly + kz) + [x' + hy' + kz') and to the product [x + hy + kz){x' + hy' + kz'), and grouping of the resulting terms, u + u' = x + x' + h{y-}- ij) 4- k{z + z'),
(2.2)
uu' = xx' -f yz' + zy' 4- h{zz' -f xy' + yx') -f A:(yy' + xz' -f zx'). (2.3) li u^ii'^u" are tricomplex numbers, the multiplication is associative {uu')u" = n(u'u")
(2.4)
and commutative uu' = u'u,
(2.5)
as can be checked through direct calculation. The tricomplex zero is 0 -f/?. • 0 + A; • 0, denoted simply 0, and the tricomplex unity is 1 -f /i • 0 4- A: • 0, denoted simply 1. The inverse of the tricomplex number n = x + /ly -f fcz is a tricomplex number u' = x' + y' + z' having the property that uu' = 1.
(2.6)
Written on components, the condition, Eq. (2.6), is xx' + zy' + yz' = 1, yx' + xy' -f zz' = 0. zx' + yy' + xz' = 0,
(2.7)
20
Complex Nmnbcrs in Three Dimensions
The system (2.7) has the sohition r'
X
=
X? -- yz x^ + y^ + z^- 3xyz'
z'- -xy y = x'^ + y^ + z^1,'
1
- -
^
(2.8)
(2.9)
3xyz^
y'- - xz ., . ,,
,
(2.10)
J.6 _|_ yi _j_ 2:'5 — Sxyz
provided that x'^ + y'^ + z'^ - Sxyz ^ 0. Since x'^ + y^ + z^ - Sxyz = {x -^-y-^ z)[x^ + y^ + z'^ - .xy -xz-
yz), (2.11)
a tricomplex number x + hy + kz has an inverse, unless x^y-^z
= {)
(2,12)
or x'^ 4- ?/^ -f ^^ - .X7y - xz -yz = 0. (2.13) The relation in Eq. (2.12) represents the plane 11 perpendicular to tlie trisector line (t) of the .r, /y, z axes, and passing through the origin O of the axes. The plane 11, shown in Fig. 2.1. intersects the xOy plane along the line z = 0,x-{-y == 0, it intersect the yOz plane along the line x = 0,y-\-z = 0, and it intersects the xOz plane along the line y = 0,x + z = 0. The condition (2.13) is equivalent to {x - y)^ + [x - z)^ 4- (y - z)'^ = 0, which for real x,y,z means that x ~ y = z, which represents the trisector line [t) of the axes x,y,z. The trisector line {t) is perpendicular to the plane 11. Because of conditions (2.12) and (2.13). the trisector line (t) and the plane n will be also called nodal line and respec'tively nodal plane. It can be shown that if uu' == 0 then (nther n = 0. or u' = 0, or one of the tricomplex numbers ti.u' belongs to the trisector line {f) and the other belongs to the nodal plane fl.
2.2
Geometric representation of tricomplex numbers
The tricomplex number x -f hy -f kz can be represented by the point P of coordinates (x,y,z). If O is the origin of the axes, then the projection
Geometric representation of tricomplex numbers
21
Figure 2.1: Nodal plane 11, of equation x + y + z = 0, and trisector line (t), of equation x = y = z, both passing through the origin O of the rectangular axes x, y, z. s = OQ of the line OP on the trisector line x = y tangent ( l / \ / 3 , 1/N/3, l / \ / 3 ) , is v/3
{x -\-y -{- z).
z, which has the unit
(2.14)
The distance D = PQ from P to the trisector line x = y = z, calculated as the distance from the point {x, y, z) to the point Q of coordinates [(x -\y + z)/3, {x -\-y + z)/'3, {x -h y -f z)/3], is D^ = o (^^ -f 7/^ -f- z"^ — x'y — xz — y^).
(2.15)
The quantities s and D are shown in Fig. 2.2, where the plane through the point P and perpendicular to the trisector line (t) intersects the x axis at point A of coordinates (x + y + 2:, 0,0), the y axis at point B of coordinates (0, X + y + 2:, 0), and the z axis at point C of coordinates (0,0, x + y + 2). The azimuthal angle (j) of the tricomplex number x H- /iy + kz is defined as the angle in the plane 11 of the projection of P on this plane, measured from the line of intersection of the plane determined by the line {t) and the X axis with the plane 11, 0 < 0 < 27r. The expression of 0 in terms of x, y, z can be obtained in a system of coordinates defined by the unit vectors
22
Complex Numbors in Three Dimensions
C(O.O.x+y+z)
Z
/ 5^
1
\
>
B(0,x^y^z.O) A{x^y^z.AO)
Figure 2.2: Triconiplex variables s,d,9.(f) for the tricomplex number x + hy + kz^ represented l)y the point P(x,y.z). The azimuthal angle (f) is shown in in the plane parallel to 11, passing through P , which intersects the trisector line (t) at Q and the axes of coordinates x, ?/, z at the points A,B,C. The orthogonal axes ^1,^2'^3 '^^^^ ^^e origin at Q.
6 : i=(2,-l,-l); 6 : ^(0,1,-1); 6 : ^^^(I'l'l)'
(2.16)
and having the point O as origin. The relation between the coordinates of P in the systems ^1,^25^3 'ii"! ^iV^z ("<^^ l^e written in the form
/ 4=
1 v/6
v/6
0 V V3
(2.17) ^
v^ y V
The components of the vector OP in the system ^ii^aiC'J ^^^ be obtained with the aid of Eq. (2.17) as ( 6 , 6 , e O = ( ^ ( 2 . x - - ? / - ^ ) . ^ ( y - 2 ) , - ^ ( a ' + y + 0)).
(2.18)
Geometric representation of tricomplex numbers
23
The expression of the angle 0 as a function of x, y, z is then cos
= -—:
5
5—
-7^,
(2.19)
2{x^ + y^ + z'^ — xy — xz — yzyi^
It can be seen from Eqs. (2.19),(2.20) that the angle of points on the x axis is 0 = 0, the angle of points on the y axis is (/> = 27r/3, and the angle of points on the z axis is (^ = 47r/3. The angle (f) is shown in Fig. 2.2 in the plane parallel to 11, passing through P. The axis Q^l is parallel to the axis 0^1, the axis Q^2 ^^ parallel to the axis 0^2^ and the axis Q^\ is parallel to the axis 0^3, so that, in the plane ABC, the angle 0 is measured from the line QA. The angle 6 between the line OP and the trisector line (<) is given by D t a n e = —, s
(2.21)
where 0 <6 < TT. It can be checked that (f = D^ + s\
(2.22)
where (fi -=x^ -i-y^ -^ z'^,
(2.23)
so that D - ds\n6, s = dcos9,
(2.24)
The relations (2.14), (2.15), (2.19)-(2.21) can be used to determine the associated projection 5, the distance D, the polar angle 9 with the trisector line (f) and the angle 0 in the H plane for the tricomplex number x+hy+kz. It can be shown that if wi = xi-\'hyi+kzi,U2 = X2+hy2+kz2 are tricomplex numbers of projections, distances and angles 5i,Di,^i,(/>i and respectively 52?^2?^27>27 then the projection .s, distance D and the angle 9,(f) for the product tricomplex number uiU2 = xiX2 + yiZ2 + y2^\ + h{z\Z2 + x\y2 + yixi) + k[yiy2 + 0:1^2 + ^12:2) are /3 s = \/Zs\S2, D — J-D1D2,
1 tan0 = --p tan^i tan02 A — 4>\+4>2' (2.25)
24
Complex Numbers
in Three
Dimensions
T h e relations (2.25) are consequences of the identities (^1^2 +yiZ2 + y2Zl) + {Z\Z2 -\-Xiy2 + yiX2) + {yiy2 + X\Z2 •^Z\X2) = {xi + ?/i + zi){x2 + y2 + Z2), (a;iar2 +y\Z2+ -{xxX2
y2Z\f
+ (^1^2 + ^l|/2 + ?/l^2)^ + (j/lJ/2 + X1Z2 + ZiX2f
+ yi2r2 + y2Z\){ziZ2 + a;i?/2 + y\X2)
-(.1:10:2 + ?/i2:2 + y2Zi){yiy2 -{ziZ2
(2.26)
+ 0:1^2 + ^1^2)
+ a:iy2 + 1/1:^2) + (1/1I/2 + 2^12^2 + ^10:2)
(^2 + yl + ^1 - ^2?y2 - X2Z2 -1/2^2), 2a:i - y i - zx 2;r2 - j/2 - ^2 , ^ : ^
(2.27)
v^. . \/3 . , ^(2/1 - ^ i ) - y (?/2 - ^2)
= -[2(xiX-2 + 1/1^2 + ^1^2) - (^1^2 +.'^iy2 +1/1^2) -(?/l|/2+.'Pl^2 + ^l^'2)], \/3 s2x2-y2-Z2 - 2 - ( y i " ^1) 2
^ \/3 + T ^ ^ ^ ~ ''^
(2.28) 2xi-yi-zi 2
= -2"[(^1^2 + .'Z^l?/2 + 1/1^2) - (yiy2 + ^1^2 + ^l.'?^2)].
(2.29)
T h e relation (2.26) shows that iiu is in the plane 11, such that .x + ?/ + z = 0, then the product un/ is also in the plane II for any ?/. The relation (2.27) shows t h a t if u is on the trisector line (t), such that x'^ + y^ -\-z^ — xy — xz — yz — 0, then u%i! is also on the trisector line (f) for any t / . If w, 1/ are points in the plane .1: + y + z = 1, then the product un! is also in that plane, and if tx, iJ are points of the cylindrical surface x^ + y^ + 2:^ - xy - .x^ -1/2: = 1. then nu' is also in that cylindrical surface. This means that if t/, t/' are points on the circle a: + y + ^ = 1,.x'^ + y^ + 2:'^ -xy~xz -yz = I, which is perpendicular to the trisector line, is situated at a distance l / \ / 3 from the origin and has the radius >/2/3, then the tricomplex product uv! is also on the same circle. This invariant circle for the multiplication of tricomplex numbers is described by the ecjuations 1 2 ^ 1 1 ^ 1 . ^ a: = - + ~cos(/), y = - - - c o s ^ + - ; = s m 0 , ^ = ^ - ^ cos (^ ~ -y= sin (p.
(2.30)
Geometric representation of tricomplex numbers
25
(0,1,0)
Figure 2.3: Invariant circle for the multiplication of tricomplex numbers, lying in a plane perpendicular to the trisector line and passing through the points (1,0,0), (0,1,0) and (0,0,1). The center of the circle is at the point (1/3,1/3,1/3), and its radius is (2/3)^/^ . It has the center at the point (1/3,1/3,1/3) and passes through the points (1,0,0), (0,1,0) and (0,0,1), as shown in Fig. 2.3. An important quantity is the amplitude p defined as p = u^l^, so that p^ = a;^ + j/^ + z^ - Zxyz.
(2.31)
The amplitude p of the product uxu^ of the tricomplex numbers u\^U2 of amplitudes pi, pz is P = PIP2-,
(2.32)
26
Complex Numbers in Three Dimensions
as can be seen from the identity (^13^2 + ?/l^2 + y2^lf + {Z\Z2 +X\y2A- y\X2f + {y\y2 + 3:1^2 + ZiX2f -3(a;ia:2 + %j\Z2 + y2Z\){z\Z2 + xiij2 + yiX2){y\y2 •¥xiZ2 + Z1X2) = (x^ + yf + ^f - 3xiyizi){xl
+ ?/| + 2:| - 3.T2y2^2).
(2.33)
The identity in Eq. (2.33) can be demonstrated with the aid of Eqs. (2.11), (2.26) and (2.27). Another method would be to use the representation of the multipHcation of the tricomplex numbers by matrices, in which the tricomplex ninnber u = x + hy + kz is represented by the matrix z \ (2.34) The product u = x + hy + kz of the tricomplex numbers U] = X} + hy\ + kz\^U2 = X2 + %2 + kz2^ is represented by the matrix multipHcation
(2.35) If i/ = det\ z X y \,
(2.36)
y
it can be checked that 1/ = x^ -{- if -^ z^ - 3xyz,
(2.37)
The identity (2.33) is then a consequence of the fact the determinant of the product of matrices is equal to the product of the determinants of the factor matrices. It can be seen from Eqs. (2.14) and (2.15) that ^3 + y3 + ^3 _ 3^y^ ^ ^ s D \
(2.38)
which can be written with the aid of relations (2.24) and (2.31) as p=^^dsin^/^0cos'/^e.
(2.39)
This means that the surfaces of constant p are surfaces of rotation having the trisector line {t) as axis, as shown in Fig. 2.4.
The tricomplex cosexponential
functions
27
Figure 2.4: Surfaces of constant p, which are surfaces of rotation having the trisector Hne {t) as axis.
2.3
The tricomplex cosexponential functions
The exponential function of a hypercomplex variable u and the addition theorem for the exponential function have been written in Eqs. (1.35)(1.36). If u = X + hy + kz^ then expw can be calculated as expw = exp:r • exp(/iy) • exp(/c2:). According to Eq. (2.1), h^ =1 k,h? — l,fc^ =. h,k^ = 1, and in general h 3m
l , / , 3 m + l ^ /j,/j3m+2 ^ j^^^Zm ^ i^^^m+l
^ f.^l,3m+2
^ ^^
(2.40)
28
Complex Numbers in Three Dimensions
where n is a natural number, so that exp{hy) and exp{kz) can be written as exp(%) = cxy + h mx y -h k px y,
(2.41)
exp{kz) = ex 2: -f- h pxz + k mx z,
(2.42)
where the functions ex, mx, px, which will be called in this chapter polar cosexponential functions, are defined by the series ex y = 1 + t//3\ + / / 6 ! -f • • •,
(2.43)
mx y=^y + y^il
(2.44)
+ y^/7\ + • • •,
px y = y V2! + y^bl + / / 8 ! + • • •.
(2.45)
From the series definitions it can be seen that ex 0 = 1, mx 0 = 0, px 0 = 0. The tridimensional polar cosexponential functions belong to the class of the polar n-dimensional cosexponential functions gnki and ex = 930, mx = iQ'aiiPx = fl32- It can be checked that cxy -\-pxy + mx y = exp y.
(2.46)
By expressing the fact that expihy-hhz) = exp{hy)-cxpihz) with the aid of the cosexponential functions (2.43)-(2.45) the following addition theorems can be obtained ex {y -h z) = ex y ex z -\- mx ypxz-\-pxy
mx z,
(2.47)
mx {y + z) = pxypx z + ex y mx z -h mx y ex 2:,
(2.48)
px (y + 2:) = mxy mxz + cxy px z -{-pxy ex z.
(2.49)
For y = z, Eqs. (2.47)-(2.49) yield ex 2y = cx^ y + 2 mx y px z,
(2.50)
mx 2y = px^ y + 2 ex y mx z,
(2.51)
px 2y = mx^ y + 2 ex y px z.
(2.52)
The cosexponential functions are neither even nor odd functions. For z = - y , Eqs. (2.47)-(2.49) yield ex y ex (-y) + mx y px (-y) + px y mx (-y) = 1,
(2.53)
The tricoinplex cosexponential functions
29
px y px (—y) + ex y mx (-y) + mx y ex (—y) = 0,
(2.54)
mx y mx (~-y) 4- ex y px (-y) -f px y ex (-y) = 0.
(2.55)
Expressions of the cosexponential functions in terms of regular exponential and cosine functions can be obtained by considering the series expansions for e^^"^^)^ and e^^"^^'^. These expressions can be obtained by calculating first [h + fc)^ and [h - kf^. It can be shown that (h+kr
= l [i-ir''+2^]
{h+k) + ^ [{--ir+2^-^],
{h - kf"^ = (-l)^-^3^^-^(fc + A; - 2), {h - kf^-^^ = (~l)^3^(/i -
fc),
(2.56)
(2.57)
where m is a natural number. Then
^(h^k)y ^(^h + k) ('le-y
+ y^y] + ^e-y + ^e^y,
(2.58)
As a corollary, the following identities can be obtained from Eq. (2.58) by writing e^^'^^^y = e^ye^y and expressing e^y and e^y in terms of cosexponential functions via Eqs. (2.41) and (2.42), 2 1 cx'^ y + mx^ y + px^ y = ~e"^ + -e^^, o o
(2.59)
ex y mx y '\- cxy px y -f- mx y px y = —-e~2/ -f -e^y.
(2.60)
From Eqs. (2.59) and (2.60) it results that cx^ y 4- mx^ y + px^ y —exy mxy — cxy pxy — mxy pxy = exp(—y).
(2.61)
Then from Eqs. (2.11), (2.46) and (2.61) it follows that cx*^ y 4- mx*^ y + px'^ y — 3cx y mx y px y = 1.
(2.62)
Similarly, ^{h-k)y ^hi
+ h + k) + \{2^h-k)
+ -^{h-k)sin{V3y). v3
cos(\/3y) (2.63)
30
Complex Niimhors in Three Dimensions
The last relation can also he written hi^ Jh-k)y
_
[1
i , r cos
2
:os ( \/3y -
+k j + ^cos(N/3y
27r\l
(2.64)
As a corollary, the following identities can be obtained from Eq. (2.63) by writing eS^~^^y = e^^e"^'-^ and expressing e^^ and e~^^ in terms of cosexponential functions via Eqs. (2.41) and (2.42), cxy cx{-y)
+ mxy mx{-y)
+ pxy px{-y)
1 2 = - + - cos(v^|/), (2.65)
ex y px (—y) + mx y ex {—y) + px y mx (—y) = ^ + Hcos(v/3y-|),
(2.66)
ex y mx (-y) + mx jy px {—y) + pxy ex (-y) 2n' 1 2 v^?/ + 3 3 + r"^
(2.67)
Expressions of e^'*^ in terms of the regular exponential and cosine functions can be obtained by the multiplication of the expressions of e^'*''"'^'^ and e(h-k)y fj-o^j gqg (2.58) and (2.63). At the same time, Eq. (2.41) gives an expression of e^''^ in terms of cosexponential functions. By equating the real and hypercomplex parts of these two forms of e'^^ and then replacing 2y by y gives the expressions of the cosexponential fimctions as cxy
e" + -7 cos '^y\
ipy
mx y
px y = - e« + - cos
V3
(2.68)
eryi\
27r'
,-yn
(2.69)
-yn
(2.70)
It is remarkable that the series in Eqs. (2.43)-(2.45), in which the terms are either of the form y"^'", or ?/•''"+\ or y'''"+'^, can be expressed in terms of elementary functions whose power series are not subject to such restrictions. The cosexponential functions difler by the phase of the cosine function in their expression, and the designation of the fimctions in Eqs. (2.69) and
Exponential and trigonometric forms of tricomplex numbers
31
(2.70) as mx and px refers respectively to the minus or plus sign of the phase term 27r/3. The graphs of the cosexponential functions are shown in Fig. 2.5. It can be checked that the cosexponential functions are solutions of the third-order differential equation
whose solutions are of the form ({u) = Acxu +Bmxu + Cpxu. It can also be checked that the derivatives of the cosexponential functions are related by rfpx dmx dcx -f- = mx, —— = ex, -— = px. au au du
2A
(2.72)
Exponential and trigonometric forms of tricomplex numbers
If for a tricomplex number u = x -{- ky + kz another tricomplex number ' ^ 1 = ^ 1 + hyi -\- kz\ exists such that x^hy
+ kz = e^i+^^i-^^^i,
(2.73)
then u\ is said to be the logarithm of ix, ui = l n n .
(2.74)
The expressions of xi, j/i, z\ as functions of .T, y, z can be obtained by developing e^y^ and e^^i with the aid of Eqs. (2.41) and (2.42), by multiplying these expressions and separating the hypercomplex components, X = e^^[cxyi cxzi -bmxyi
m x z i + p x y i px^i],
y =: e^^[cxyi px^i + mxyi c x ^ i + p x y i mx^i], z = e^^[cxyi m x ^ i + p x y i c x ^ i + m x y i px^i],
(2.75) (2.76) (2.77)
Using Eq. (2.33) with the substitutions xi ~> cxyi,yi —> mxyi^zi ~> pxyi^X2 -> cxz\,y2 -^ pxzi,Z2 -^ mxz\ and then the identity (2.62) yields x^ H- y^ + z^ - 3xyz = e^'^', whence
(2.78)
32
Complex Nuwhers in Three Dimensions
^
Figure 2.5: Graphs of the cosexponential functions ex, mx, px.
Exponential and trigonometric forms of tricomplex numbers xi = - \n{x^ + y^ 4- ^^ -
3xyz).
33 (2.79)
The logarithm in Eq. (2.79) exists as a real function for x -\' y + z > 0. A further relation can be obtained by summing Eqs. (2.75)-(2.77) and then using the addition theorems (2.47)-(2.49) (^3+ 3 + z^'^-lx
z)V3 " cx(yi+^i)+mx(yi+2:i)+px(yi+^i).(2.80)
The sum in Eq. (2.80) is according to Eq. (2.46) e^^-^^S so that yi-{-
zi=\n
X -{- y -\- z
{x^ + y'^ + z'^ — 3xyzy/^
(2.81)
The logarithm in Eq. (2.81) is defined for points which are not on the trisector Une (f), so that x*^ -\- y'^ -^ z'^ — xy - xz - yz ^ 0. Substituting in Eq. (2.73) the expression of xi^ Eq. (2.79), and of zi as a function of x,y,z,yu Eq. (2.81), yields u
exp
-A: In
^V2s
2/31
— ^{h-k)yi^
D
(2.82)
where the quantities p^s and D have been defined in Eqs. (2.31),(2.14) and (2.15). Developing the exponential functions in the left-hand side of Eq. (2.82) with the aid of Eq. (2.42) and using the expressions of the cosexponential functions, Eqs. (2.68)-(2.70), and using the relation (2.63) for the right-hand side of Eq. (2.82) yields for the real part (._^)eo3[-i.ln(^)]-f(,-.)sin[j.ln(^)] {x^ -{- y'^ -h z'^ — xy — xz — yzyi^ = cos(\/3t/i).
(2.83)
which can also be written as
1 , fV2s'
In cos -7= N/3
D
cos i^yi)-
(2.84)
where (f) is the angle defined in Eqs. (2.19) and (2.20). Thus Vi = - In ^ 3
^H-
(2.85)
The exponential form of the tricomplex number u is then u — p exp
5<'' + " ' " ^ + > - * > *
(2.86)
34
Complex Numbers in Three Dimensions
where 6 is the angle between the Hne OP connecting the origin to the point P of coordinates {x,y,z) and the trisector Hne (f), defined in Eq. (2.21) and shown in Fig. 2.2. The exponential in Eq. (2.86) can be expanded with the aid of Eq. (2.58) and (2.64) as 2-h-k
exp 3
+
3
tan 6
l+h
+ k { V2 tan^
/tan6>y/'^
v"^;
2/3
(2.87)
so that X + hy + kz = p
exp
2-h-k
/tane\^^^ /tan6>y V v/2;
2/3'
l +h+k
+
:;
tanO (2.88)
( ^ * )
Substituting in Eq. (2.88) the expression of the amplitude p, Eq. (2.39), yields
^\/I(H
fe . ^ 1-f/l + fc -V2cos(9Jexpf-yr-<^J ,(2.89) — sine+
which is the trigonometric form of the tricomplex number u. As can be seen from Eq. (2.89), the tricomplex number x + hy + kz is written as the product of the modulus d, of a factor depending on the polar angle 0 with respect to the trisector line, and of a factor depending of the azimuthal angle (j) in the plane 11 perpendicular to the trisector line. The exponential in Eq. (2.89) can be expanded further with the aid of Eq. (2.64) as exp [ -j={h - k)(f)j =
+ -
h- k , h - k , ; /^ ^^x cos >-h —y=r- sm 0, (2.90) 3 -^ v/3
so that the tricomplex number x + hy + kz can also be written, after multiplication of the factors, in the form X + hy + kz H
2-h~-k
h-k j=-{x +y l +h+k
+-
(ir^ + j/^ + 2^ - xy -~xz - yz)^/'^ cos0
+ z^ - xy - xz -yz)
{;x + y + z).
' mncp (2.91)
The validity of Eq. (2.91) can be checked by substituting the expressions of cos(^ and sine/) from Eqs. (2.19) and (2.20).
Elementary functions of a tricomplex variable
2.5
35
Elementary functions of a tricomplex variable
It can be shown with the aid of Eq. (2.86) that {x + hy + kz)"" = p'
'2-h-k 3
i+h+k (v/2 y"^' "*"
3
/tan^N"*/^
['WJ exp ( - ^ " ^ ' Z ' ) »
(2.92)
\^tanej
or equivalently {x + hy + kz)"^ = H
(a;^ + y'^ + z"^ - xy - xz - yzf^^'^ cos{m(f))
-?=-'{x^ + y^ + z^ — xy — xz - yz)^'^ sin{m(j)) v3
+ ^-^^^^(x + y + zr,
(2.93)
which are vaUd for real values of m. Thus Eqs. (2.92) or (2.93) define the power function u'^ of the tricomplex variable u. The power function is multivalued unless m is an integer. It can be inferred from Eq. (2.86) that, for integer values of m, {uu'r
= u"^ w'"^.
(2.94)
For natural m, Eq. (2.93) can be checked with the aid of relations (2.19) and (2.20). For example if m = 2, it can be checked that the right-hand side ofEq. (2.93) is equal to (a:-f % + A:2:)2 = x'^-^2yz + h(z'^-\-2xz) + k{y'^+2xz). The logarithm ui of the tricomplex number u^ ui = In w, can be defined as the solution of Eq. (2.73) for ui as a function of u. For x + y -\- z > 0^ from Eq. (2.86) it results that Inn = ln/9 + -(/i + k) In ( - ^ )
+ -7={h - k)(t).
(2.95)
It can be checked with the aid of Eqs. (2.25) and (2.32) that l n ( W ) = lntA + lnu',
(2.96)
which is valid up to integer multiples of 27t{h - k)/\/3. The trigonometric functions of the hypercomplex variable u and the addition theorems for these functions have been written in Eqs. (1.57)(1.60). The trigonometric functions of the hypercomplex variables hy.ky
36
Complex Numbers in Three Dimensions
can be expressed in terms of the cosexponential functions as 1 cos{hy) = -[ex (iy) + ex (-iy)] + ~h[mx {iy) + mx (-jy)] + -A;[px(?:y) + px(-M/)],
(2.97)
1 1 cos(%) = -[ex (iy) + ex {-iy)] + -h[px (iy) + px {-iy)] + -k[mx {itj) + rnx {-iy)],
(2.98)
sin(/iy) = —[ex {iy) - ex {-iy)] + —.h[mx {iy) - mx {-iy)] 1 + ^A:[px(«y)-px(-i?/)],
(2.99)
sin(A;y) = —[ex {iy) - ex {-iy)] + —h[px {iy) - px {-iy)] Zl
2,1
1 (2.100) + —•A;[mx {iy) — mx {—iy)], 2i where i is the imaginary unit. Using the expressions of the cosexponential functions in Eqs. (2.68)-(2.70) gives expressions of the trigonometric fiuictions of hy, hz as /, V 1 2 , / ^ \ y cos{hy) = - cos y + - cosh I -^y I cos - +
+h 1
1 ./N/3 \ y ^ '^ • u ( ^ \ • y - cos y - - c o s h l^-^-yj c o s - + - ^ smh l^—yj s m -
\ . y +k -1c o s y - -1c o s, h/ Nl ~/ 3y l\ c o sy- - - ^1s m• h. l/"N/S —ylsm(2.101)
cos(fcy) = - cos y+-
+h
1
1
cosh ( — y J cos - -t, /\/3 \
y
^
• u(^
\
• y
3 '""'y ~ 3 '"'^ VT^j '-"' 2 - 7! '' [^V ""' 2
+k 1
1
,(y/^\
y
1
• v./^ ^ • y (2.102)
Elementary functions of a tricomplex variable
sm{hy) = - siny - - cosh I —y - siny + - cosh \^—yj
37
I sin sm - + - ^ smh (^—yj cos -
- s m y + -cosh ( — y 1 s m - - - ^ s m h \—y
cos(2.103)
sin(A;y) = ^ siny — - cosh
. y -y I sm -
1 . 1 , +/i - s m y + -cosh
sinh
1 . 1 ,. - s m y + -cosh
sinh
4-fc
y\cosy
-y I cos 2 ^/ 2 (2.104)
The trigonometric functions of a tricomplex number x + hy -h kz can then be expressed in terms of elementary functions with the aid of the addition theorems Eqs. (1.59), (1.60) and of the expressions in Eqs. (2.101)-(2.104). The hyperbolic functions of the hypercomplex variable u and the addition theorems for these functions have been written in Eqs. (1.62)-(1.65). The hyperbolic functions of the hypercomplex variables /ly, ky can be expressed in terms of the elementary functions as 1 ^ 2 ~ cosh y -f- - cos - ^ ? / ) c o s h | +
cosh.{hy)
*5
o
+h l cosh y - i COS l^-^yj
cosh | - i = sin ^ - ^ y j sinh |
+k I cosh y - i cos f - ^ y j cosh | + - L sin ( - i y j sinh | (2.105) u/, X 1 I 2 /N/3 \ , y cosh(A;y) = - coshy + - cos I — y I cosh - +
+h
1
..
1
/\/3 \
, y
1
. /\/3 \
. , y
3 '°'^ y - r ° n T^) ""'^ 2 + ;^''" 1 ^^J'' 2
38
Complex Numbers in Three Dimensions 1
+k - cosh y — - cos
1 2 smh{hy) = - sinhy - - cos
. sin
sinh 2 (2.106)
^ y ) sinh I
+h 1 • ,
1 (^ \ . , y 1 . - s m h y + - c o s l - y l s m h - + -^sm
-^y] cosh I
• t y + -1 cos ^ , j s i n h | - - L s i n V3 +k -1 sinh
2 ^) ^^^'^ 2J 2 (2.107)
1 2 I sinh i: sinh(A;t/) = - sinhy — - cos -—V 2 '^/ 2
1 . ,
1
(\f^
\ . , y
1
-iv """"^ 2
- smhw + - cos ——y snih 7= sni 3 - ' 3 1 2 ^ ; 2 v/3 1 • . 1 ^ y j s i n h | + - L s i n ----?/ I cosh --sinhy+-cos 2 ^ 2
(2.108) The hyperboHc functions of a tricomplex number x -f hy + kz can then be expressed in terms of the elementary functions with the aid of the addition theorems Eqs. (1.64), (1.65) and of the expressions in Eqs. (2.105)~(2.108).
2.6
Tricomplex power series
A tricomplex series is an infinite sum of the form (2.109) where the coefficients a^ are tricomplex numbers. The convergence of the series (2.109) can be defined in terms of the convergence of its 3 real components. The convergence of a tricomplex series can however be studied using tricomplex variables. The main criterion for absolute convergence remains the comparison theorem, but this requires a number of inequalities which will be discussed further. The modulus of a tricomplex number u = x + hy + kz can be defined as \u\=.{x'+y^
+ Z^)^'\
(2.110)
Tricomplex power series
39
Since \x\ < \u\^\y\ < \u\^\z\ < |n|, a property of absolute convergence established via a comparison theorem based on the modulus of the series (2.109) will ensure the absolute convergence of each real component of that series. The modulus of the sum U1+U2 of the tricomplex numbers Ui^U2 fulfils the inequality IKI - \U2\\ < |U1 + U2\ < K l + \U21
(2.111)
For the product the relation is \uiU2\ < v/3|t/i||i/2|,
(2.112)
which replaces the relation of equality extant for regular complex numbers. The equality in Eq. (2.112) takes place for a^i = yi = zi and 0:2 = 2/2 = ^2^ i.e when both tricomplex numbers lie on the trisector line (i). Using Eq. (2.91), the relation (2.112) can be written equivalently as lsi5l
+ \a\al
< 3 (^J? + ^ a ? ) ( ? 5 i + ^ a l ) ,
(2.113)
where S'j = xj + y'j A-z'j - Xjpj -XjZj -yjZj.aj = Xj + yj + ZjJ = 1,2, the equality taking place for c^i = 0, J2 = 0- A particular form of Eq. (2.112) is \u^\ < V3\u\^,
(2.114)
and it can be shown that |^/|<3{^-i)/2|^|/^
(2.115)
the equality in Eqs. (2.114) and (2.115) taking place (or x =^ y = z. It can be shown from Eq. (2.93) that h f = p'^ + ^cT^',
(2.116)
where ^^ =^ x^ -^-y'^ + z"^ - xy - xz - yz,a = x + y + z. Then Eq. (2.115) can also be written as 2 c2l , ^ ^ 2 / ^ Q / - 1 / ^ r 2 zs^i + ia^i<Si'^iiS' + l<^')\
(2.117)
the equahty taking place for 5 = 0. From Eqs. (2.112) and (2.115) it results that |au'|<3^/2|a|H^
(2.118)
40
Complex Numbers in Three Dimensions
It can also be shown that \u > FT' u
(2.119)
the equality taking place for a^ = r5^, or xy + xz + yz = 0. A power series of the tricomplex variable u is a series of the form ao + aiu + a'zii^ + • • • + aiu^ + • • -.
(2.120)
Since
E^^^I^E^'^'I^^IH^
(2-121)
a sufficient condition for the absolute convergence of this series is that
lim ^^htilM < 1. n->oo
(2.122)
\ai\
Thus the series is absolutely convergent for \u\ < Co,
(2.123)
where i-^oo v/3|a/4.i| The convergence of the series (2.120) can be also studied with the aid of a transformation which explicits the transverse and longitudinal parts of the variable u and of the constants a/, X + hy + kz =^ vie\ + vyei + v^e^,
(2.125)
where vi =
, vi = — ( y -z),
v+=x
+ y + z,
(2.126)
and 2-h-k ei = — y — ,
, h-k e, = - ^ ,
l+h^-k «+ = — ^ —
/„ ,n-,x (2-127)
The variables v\,vi,Vj^ will be called canonical tricomplex variables, ei,e\, e-f- will be called the canonical triconipk^x base, and Eq. (2.125) gives the canonical form of a tricomplex number. In the geometric representation of
Tricomplex power series
41
Figure 2.6: Unit vectors e\,e\,ej^ of the orthogonal system of coordinates with origin at Q, The plane parallel to 11 passing through P intersects the trisector line {t) at Q and the axes of coordinates x, y, z at the points
Fig. 2.6, ei,ei are situated in the plane 11, and e^ is lying on the trisector line (t). It can be checked that ei = ei, ei = —ei, eici = ei, 6164. = 0, 6164. = 0 , e^ = e_^.(2.128) The moduli of the bases in Eq. (2.128) are
|ei| = /f, |ei| = y|, K| = /I,
(2.129)
and it can be checked that 2 1 \x + hy + kzf = -{vl + vl) + -vl.
(2.130)
If u = u'u", the transverse and longitudinal components are related by vi - v[v'l - v\v", v\ = v[v'l 4- v\v'l, w+ = i^+u+,
(2.131)
42
Complex Numbers in Three Dimensions
which show that, upon multiphcation, the transverse components obey the same rules as the real and imaginary components of usual, two-dimensional complex numbers, and the rule for the longitudinal component is that of the regular multiplication of numbers. If the constants in Eq. (2.120) are a/ = p/ -f hqi +fcr/,and «n =
2p/ — m ~ Ti ^ \/3. , 2 ' ^'1 ^ "2"^^' ~ ^')' ^'+ = P/ -+- ^/ + r/,
, ^ (2.132)
where po = l,go = 0,ro = 0, the series (2.120) can be written as 00
Y^ \aiiei 4- anex){viei + i;iei)^ + e+a/+^^] .
(2.133)
The series in Eq. (2.133) is absolutely convergent for K | < c + , {v\^-vl)^''^
(2.134)
c, = lim Jf!!±L, c, = li,„ ^ ''''•^°"'"' ,.
(2.135)
where
The relations (2.134) and (2.130) show that the region of convergence of the series (2.120) is a cylinder of radius c\ \/2/3 and height 2c4_/v/3, having the trisector line {t) as axis and the origin as center, which can be called cylinder of convergence, as shown in Fig. 2.7. It can be shown that c\ = (l/\/3) min(c4-,ci), where min designates the smallest of the numbers CJ^,C\. Using the expression of \u\ in Eq. (2.128), it can be seen that the spherical region of convergence defined in Eqs. (2.123), (2.124) is a subset of the cylindrical region of convergence defined in Eqs. (2.134) and (2.135).
2.7
Analytic functions of tricomplex variables
The analytic functions of the hypercomplex variable u and the series expansion of functions have been discussed in Eqs. (1.85)-(1.93). If the tricomplex function f{u) of the tricomplex variable u is written in terms of the real functions F{x, y, z), G(x, y, z), H{x, y, z) of real variables .x, y, z as f(u) = F{x, y, z) + hG{x, y, z) 4- kH{x, y, z),
(2.136)
Analytic functions of tricomplex variables
2c /3
43
1/2
Figure 2.7: Cylinder of convergence of tricomplex series, of radius (2/3)^/"^ ci and height 2c4./3^/^, having the axis parallel to the trisector line, and the regions of space delimited by the plane, cylindrical and conical surfaces. then relations of equality exist between partial derivatives of the functions F, G, H. These relations can be obtained by writing the derivative of the function / as dF ^ dF ^ ay Ax -f hAy + kAz ax , /dG ^ dG , dG , dx dy dz , (dH , _. ^ dH ^ dH ^ +k -rr-A-T + —-Ay + -^Az 1
\ ax
ay
az
dF ^ az
(2.137)
where the difference appearing in Eq. (98) is u — UQ = Ax + hAy + kAz, The relations between the partials derivatives of the functions F, G, H are obtained by setting successively in Eq. (2.137) Ax ~> 0, Ay = 0, A^: = 0; then Ax = 0, Ay -^ 0, Az = 0; and A,x = 0, Ay = 0, Az ~> 0. The relations are
dF dx
dG dG dy' dx
dH dH dy ' dx
dF dy'
(2.138)
44
Complex Numbers in Three Dimensions dF^_dH_ dG__dF^ dx dz 'Ox dz'
OH _ OG dx dz'
d^_dH_ dH^_dF^ dy dz ' dy dz'
^ ^ ^ dy dz'
v•
;
The relations (2.138)-(2.140) are analogous to the Riemann relations for the real and imaginary components of a complex function. It can be shown from Eqs. (2.138)-(2.14()) that the components F solutions of the equations d'^F dx^
d^F ^ d^F d'^F ^ d'^F d'^F = 0^ ^ 1 7 3 7 1 = 0^ -T^-in^.-^^ dydz dy'^ ~ dxdz dz^ dxOy
d'^G Ox^
d'^G dydz
d^H dx'^
d^H dydz
^^
d'^G d'^G d'^G d'^G 0, - 7 7 - 7 - 7 7 ^ = 0, - _ - — _ = ( ) , Oy'^ Dxdz Oz'^ dxdy O'^H ' dy-
d'^H dxdz
,^^^^^ (2.141)
(2.142)
a'^H _d\H ' dz"^
dxdy
It can also be shown that the differences F-G,F-H,Gof the equation of Laplace, A ( F - G ) = 0, A ( F - / / ) = ( ) , A ( G - / / ) = (), d'^ d'^ d'^
H are solutions
,
,
If a geometric transformation is considered in which to a point ?/ is associated the point /(u). it can he shown tliat th(^ tri(!omplex fmu'tion f{u) transforms a straight line paralU^l to the trisector line (f) in a straight line parallel to (f), and transforms a plane parallel to the nodal i)lane n in a plane i)arallel to FI. A transformation generated by a tricomplex function f{u) does not conserve in general the angle of intersecting lines.
2,8
Integrals of tricomplex functions
The singularities of tricomplex functions arise from terms of the form \/(na ) " \ with m > 0. Functions containing such terms are singular not only at u = a. but also at all points of a plane (II^) through the point a and parallel to the nodal plane 11 and at all points of a straight line {ta) passing through a and parallel to the tris(M-tor linc^ (f). The integral of a tricomplex function between two points A,B along a path situated in a region free of singularities is independent of path, which
Integrals of tricomplex functions
45
means that the integral of an analytic function along a loop situated in a region free from singularities is zero, i f{u)du = 0,
(2.145)
where it is supposed that a surface S spanning the closed loop F is not intersected by any of the planes and is not threaded by any of the lines associated with the singularities of the function f{u). Using the expression, Eq. (2.136) for f{u) and the fact that du = dx -i- hdy + kdz^ the explicit form of the integral in Eq. (2.145) is (f f{u)du = (f [Fdx + Hdy + Gdz 4- h{Gdx + Fdy + Hdz) +k{Hdx + Gdy + Fdz)],
(2.146)
If the functions F.G^H are regular on a surface 5 spanning the loop F, the integral along the loop F can be transformed with the aid of the theorem of Stokes in an integral over the surface S of terms of the form dH/dx — dF/dy, dG/dx — dF/dz^ dG/dy — dH/dz^... which are equal to zero by Eqs. (2.138)-(2.140), and this proves Eq. (2.145). If there are singularities on the surface S, the integral § f[u)du is not necessarily equal to zero. If f{u) = l/{u — a) and the loop F^ is situated in the half-space above the plane (IIQ) and encircles once the line (^a), then -\h-k).
(2.147)
JVn
This is due to the fact that the integral of l/(n — a) along the loop Fa is equal to the integral of l/(w — a) along a circle [CQ) with the center on the line {to) and perpendicular to this line, as shown in Fig. 2.8. f du
f du = 4>
, ,
, 2.148
this being a corrolary of Eq. (2.145). The integral on the right-hand side of Eq. (2.148) can be evaluated with the aid of the trigonometric form Eq. (2.88) of the tricomplex quantity it — a, so that du
}i — k ^, = —7^d(j>,
, , 2.149
which by integration over dcj) from 0 to 2TT yields Eq. (2.147). The integral ^. du{iL - ay^\, with m an integer number not equal to -1, is equal to zero, because Jdu{u — a)"' = {u — oY^^'^^/{m + 1), and {u — aY^'^^/{in 4- 1) is singlevalued.
i
Fa
du{u - a)"^ = 0, for in integer, ni 7^ - 1 .
(2.150)
46
Complex Numbers in Three Dimensions
Figure 2.8: The integral of l/(n - a) along the loop TQ is equal to the integral of l/(?i - a) along a circle (CQ) with the center on the line {ta) and perpendicular to this line. If f{u) is an analytic tricomplex function which can be expanded in a series as written in Eq. (1.89), with uo =• a, and the expansion holds on the curve r and on a surface spanning F, then from Eqs. (2.149) and (2.150) it follows that
i
f{u)du _ 27r
r u—a
73(h-k)f{a).
(2.151)
Substituting in the right-hand side of Eq. (2.151) the expression oi f{u) in terms of the real coinponents F, G, H, Eq. (2.136), at u = a, yields
Jr u -
tt
v3
H) + k{G-F)],
(2.152)
Since the sum of the real components in the paranthesis from the righthand side of Eq. (2.152) is ecjual to zero, this equation determines only the differences between the components F.G.H. If/(t^) ran be expanded as
Factorization of tricomplex polynomials
47
written in Eq. (1.89), with UQ = a, on F and on a surface spanning F, then from Eqs. (2.147) and (2.150) it also results that
f7-^^
= ^(''-W""'(a).
yr [u - a)^+^
(2.153)
v3m!
where the fact that has been used that the derivative /^"^^(a) of order m of f{u) Sit u = a is related to the expansion coefficient in Eq. (1.89), with uo = a, according to Eq. (1.93), with UQ = a. The relation (2.153) can also be obtained by successive derivations of Eq. (2.151). If a function f{u) is expanded in positive and negative powers of w — Uj, where Uj are fourcomplex constants, j being an index, the integral of / on a closed loop F is determined by the terms in the expansion of / which are of the form aj/{u — Uj)^
/(-) = - + E r ? r : + - -
(2-154)
In Eq. (2.154), Uj is the pole and aj is the residue relative to the pole Uj. Then the integral of / on a closed loop F is r
27r
j> f{u)du=--={h-k)J2
\nt{ujuJu)aj,
(2.155)
where the functional int(M, C), defined for a point M and a closed curve C in a two-dimensional plane, is given by ' j^^TiT ^\ J X xL j M is an Ulterior point of (7, ,_ ^^^. int(M,C) = ^ . • . ^ (2.156) \ n 0 if-r «^ • V ' ^ » ^ ^^ M is exterior to C, ^ ^ and Ujii^ Fn are the projections of the point Uj and of the curve F on the nodal plane 11, as shown in Fig. 2.9.
2.9
Factorization of tricomplex polynomials
A polynomial of degree m of the tricomplex variable u =- x + hy + kz has the form Pm[u) = u'^ + aiu"^"^ + • • • + am-\u + a^,
(2.157)
where the constants are in general tricomplex numbers, a/ = p/ + hqi + fcr/, / = l , - " , m . In order to write the polynomial Pm(^) as a product of
48
Complex Numbers in Three Dimensions
Figure 2.9: Integration path F, pole Uj and their projections Tn.Ujji on the nodal plane H. factors, the variable u and the constants a/ will be written in the form which explicits the transverse and longitudinal components, 7n
m
Prn{u) = ^ ( a n e i +anei)(7j,ei -f ^;iei)^^-^ + e4 5 3 a / + < - ^ (2.158) /=0
/=:0
where the constants have been defined previously in Eq. (2.132). Due to the properties in Eq. (2.128), the transverse part of the polynomial Pm(^) can be written as a product of linear factors of the form m
m
^(aaei+a/iei)(?;iei+i)iei)'"~' = 1=0
]]^[{v[-vn)eii-(vi~vn)ei],{2Abd) 1=1
where the quantities vn^vn are real numbers. The longitudinal part of Pyn(?/), Eq. (2.158), can be written as a product of linear or quadratic factors with real coefficients, or as a product of linear factors which, if imaginary, appear always in complex conjugate pairs. Using the latter form for the simplicity of notations, the longitudinal part can be written as m
rn
5 ^ a H ^ r ' = n(^^+-^^/+), 1=0
i=\
(2.160)
Factorization of tricomplex polynomials
49
where the quantities vi^ appear always in complex conjugate pairs. Due to the orthogonality of the transverse and longitudinal components, Eq. (2.128), the polynomial Pmi'^) can be written as a product of factors of the form m
Pm{u) = Ylii'^i - vii)ei + {vi - vii)ei + {v^ - ^/+)e+].
(2.161)
1=1
These relations can be written with the aid of Eqs. (2.125) as m
Pm{u) = l[{u-ui),
(2.162)
where 111 = viiei + vnei + vi^e.^.
(2.163)
The roots vi^ and the roots vnei + vnei defined in Eq. (2.159) may be ordered arbitrarily. This means that Eq. (2.163) gives sets of m roots u i , . . . , u ^ of the polynomial Pm{u)^ corresponding to the various ways in which the roots vi^.vnei + vnei are ordered according to / in each group. Thus, while the tricomplex components in Eq. (2.158) taken separately have unique factorizations, the polynomial PmiQ) can be written in many diflFerent ways as a product of linear factors. If P{u) = u^ — 1, the degree is m = 2, the coefficients of the polynomial are a\ = 0,a2 = ~ 1 , the coefficients defined in Eq. (2.132) are a2i = ~l,a22 = 0,a2u = —1- The expression of P(ii), Eq. (2.158), is (eit;i + ^i^i)^ — ei -i-e+(t;^ — 1). The factorizations in Eqs. (2.159) and (2.160) are (ei?;i-f ei?)i)^-ei = [ei(t?i+ l) + ei?)i][ei(i;i - l ) + ei?}i] and v\-l = {v+~{l){v^'-l). The factorization of P(n),Eq. (2.162), isP(n) = {u-ui){n-U2), where according to Eq. (2.163) the roots are ui = ±ei ±e^,U2 = —ui. If ei and e^ are expressed with the aid of Eq. (2,127) in terms of h and fc, the factorizations of P{u) are obtained as u^ -1 = {u + l){u-l),
(2.164)
or as u'-l
( \-2h-2k\( («+ 3
)(„
l-2h-2k\ 3
It can be checked that (±ei ± e-f )^ = ei + e+ = 1.
).
,„ _ , , (2.165)
50
Complex Numbers in Three Dimensions
2.10
Representation of tricomplex numbers by irreducible matrices
If the matrix in Eq. (2.34) representing the tricomplex number u is called f/, and
/l" —L (2.166)
i I v/3
v^
which is the matrix appearing in Eq. (2.17), it can be checked that 2
TUT-'
=
V
•^{y-z) 2 \y 0
2
^j
X —
2
0 0
(2.167)
0
The relations for the variables x — {y + z)l2, {\/3/2){y — z) and x + y-\-ziov the multiplication of tricomplex numbers have been written in Eqs. (2.26), (2.28) and (2.29). The matrices TUT~^ provide an irreducible representation [7] of the tricomplex numbers ii = x + hy + kz, in terms of matrices with real coefficients.
Chapter 3
Commutative Complex N u m b e r s in Four Dimensions Systems of hypercomplex numbers in 4 dimensions of the form u = x + ay + I3z + jt are described in this chapter, where the variables x, y, z and t are real numbers, for which the multiplication is both associative and commutative. The product of two fourcomplex numbers is equal to zero if both numbers are equal to zero, or if the numbers belong to certain four-dimensional hyperplanes as discussed further in this chapter. The fourcomplex numbers have exponential and trigonometric forms, and the concepts of analytic fourcomplex function, contour integration and residue can be defined. Expressions are given for the elementary functions of fourcomplex variable. The functions f(ii) of fourcomplex variable defined by power series, have derivatives limw-^uo[/(^) ~~ /(^o)]/(?^ — ^o) independent of the direction of approach of u to UQ. If the fourcomplex function f{u) of the fourcomplex variable u is written in terms of the real functions P{x^y,z,t),Q{x,y^z,t),R{x^y,z,t),S{x,y,z^t), then relations of equality exist between partial derivatives of the functions P, Q, i?, S. The integral JA f{'^)du of a fourcomplex function between two points A,B is independent of the path connecting A, B. Four distinct types of hypercomplex numbers are studied, as discussed further. In Sec. 3.1, the multiplication rules for the complex units a,/3 and 7 are a^ = - 1 , /3^ = —1, ^'^ = l,a/3 = /3a = —7, 07 = 7 a = /5, ^7 = 7/3 = a. The exponential form of a fourcomplex number is ii = pexp [7 In tan ip + {l/2)a{(p -h x) + (l/2)/3(0 - x)] 1 where the amphtude is p^ = [{x + t)'^ + (y + z)'^] [{x - t)^ + {y - zY\ , 0, X are azimuthal angles, 51
52
CouitnutHtive Complex Numbers in Fom' Dimensions
0 < 0 < 27r, 0 < X < 27r, and i/? is a planar angle, 0 < V^ < 7r/2. The trigonometric form of a fourcomplex number is ?/ = ri[cos(V^ —7r/4) + 7sin(t/^~7r/4)] exp [{l/2)a{(t) + x) + (l/2)/3(0 - x% where (f = .r^ + y^ + ^'^ -f f. The amplitude /r> and tani/^ are multiplicative and the angles 0, x are additive upon the multiplication of fourcomplex numbers. Since there are two cyclic variables, (f) and x? these fourcomplex numbers are called circular fourcomplex numbers. If f{u) is an analytic fourcomplex function, then §^^f{u)du/{u-uo) = 7T[{a + P) ini{uQ^yS^y)-¥{a-^) intKrC^Tr^)] /(wo), where the functional int takes the values 0 or 1 depending on the relation between the projections of the point UQ and of the curve F on certain planes. A polynomial can be written as a product of linear or quadratic factors, although the factorization may not be unique. In Sec. 3.2, the nniltiplication rules for the complex units a, (5 and 7 are a^ = 1, f3^ = 1, 7'^ == l.c'^ft = l^oc = 7, rv7 = 7 a = ft, /J7 = 7 ^ = a. The exponential form of a fourcomplex number, which can be defined for sz=. x-i-y-\- z-^tX). s' = X - y -^ z - t > 0, s" = x •]-y - z - t > 0, s^" = X — y ~ z -{• t > 0, is n = //exp(ai/i 4- ftz\ + 7^1), where the amplitude is /i = [ss's^s'^yi'^ and the arguments are jyi = (1/4) ln(.s,s'7.s'5'"), ^1 = {llA)\n{ss"'/s's''), U = (l/4)ln(.s,s7.9".s'"). Since there is no cyclic variable, these fourcomplex numbers are called hyperbolic fourcomplex numbers. The amplitude // is multiplicative and the arguments y\,z\,t\ are additive upon the multiplication of fourcomplex numbers. A polynomial can be written as a product of linear or (quadratic factors, although the factorization may not be uniciue. In Sec. 3.3, the multiplication rules for the complex units o,ft and 7 are OL^ = /?, fi'^ = - 1 , 7^ — -ft^otft — i^Oi — 7, 0^7 = 70' = - 1 , /?7 = r^^ = —a . The exponential function of a fourcomplex number can be expanded in terms of the four-dimensional cosexponential functions fu){x) — 1 - rz;V4! + .x^S! - • • •, /HI-X) - x - ^ 1^ + ;i;V9! - • • •, jn{x) = X-V2! - a;V6! + x^VlO! , f^'^{x) = x^I'M - x^ll\ + x^^j\\\ . Expressions are obtained for the four-dimensional cosexponential functions in terms of elementary functions. The exponential form of a fourcomplex number is u — pexp {(l/2)(a - 7) Intani/; +(l/2)[/i -f (a -f 7)/\/2]> -(l/2)[/i - (« + 7)/\/2]x}, where the amplitude is p^ = | [j; + (jy -
i)l\[^
+ [^ + (?/ + f ) / v / 2 ] ' | | [ : r - ( ? / - f ) / v / 2 ] ' + [ ^ - ( y + ^ ) / v / 2 ] ' | , (/).x are azimuthal angles, 0 < 0 < 27r,0 < x < 27r, and t/? is a planar angle, 0 < ^ < 7r/2. The trigonometric form of a fourcomplex number is u = d [cos (V; - 7r/4) +.(l/v^)(a - 7) sin (t/^ - 7r/4)] exp {(l/2)[/:i + [a + 7)/v^]>
Introduction to Chapter 3
53
-{l/2)[/3 - (a + 7 ) / \ / 2 ] x } , where (f = x'^ -\-y'^ + z'^ + i\ The ampUtude p and tan i/? are multipUcative and the angles (/>, x ^r^ additive upon the multipHcation of fourcomplex numbers. There are two cychc variables, cf) and X? so that these fourcomplex numbers are also of a circular type. In order to distinguish them from the circular hypercomplex numbers, these are called planar fourcomplex numbers. If/(w) is an analytic fourcomplex function, then §p f{u)du/{u - UQ) — TT m + (a + 7 ) / \ / 2 j mt(uo^v^^^v) + (p — [a + 7 ) / \ / 2 j int(worC?r7^^) f{uo) , where the functional int takes the values 0 or 1 depending on the relation between the projections of the point 1^0 and of the curve F on certain planes. A polynomial can be written as a product of linear or quadratic factors, although the factorization may not be unique. The fourcomplex numbers described in this chapter are a particular case for n = 4 of the planar hypercomplex numbers in n dimensions discussed in Sec. 6.2. In Sec. 3.4, the multiplication rules for the complex units a, P and 7 are Q;2 = ^^ ^2 ^ I ^2 ^ p^^^f^ = /Ja = 7, Of7 = 7Qf = 1, ^7 = 7/3 = a. The product of two fourcomplex numbers is equal to zero if both numbers are equal to zero, or if the numbers belong to certain four-dimensional hyperplanes described further in this section. The exponential function of a fourcomplex number can be expanded in terms of the four-dimensional cosexponential functions ^40G'^) = l+.T^/4!-f x^/8! + -• •, (/4i(^) = x-\-x^/b\+x^/9\-{• • •, g42{x) = x^/2\-\-xy6\+x^yiO\^'. •, g^^ix) = x^'M^x^/7\^x^^/n\ + •••. Addition theorems and other relations are obtained for these fourdimensional cosexponential functions. The exponential form of a fourcomplex number, which can be defined for x+y+z+t > 0, x—y+z—t > 0, is u = pexp [(l/4)(a + /3 + 7) l n ( v ^ / tan(9+) - (l/4)(a ~ /3 + 7) ln(v/2/ tan 6>_) +{a - 7)0/2], where p = {p^P-^^'^ pi = {x - zf + (y - if, pi = {x + zf -{y + f)^ e^ = {l+a~\-P-\- 7)/4, e^ = (I - a + 0 - 7)/4, ex = (1 —/3)/2,ei = (a —7)/2, the polar angles are tanS^. = \/2p^/v^^ta.n6= \/2//+/?;_, 0 < ^+ < 7r,() < ^_ < TT, and the azimuthal angle <> / is defined by the relations x ~ y — p^coscj), z - t — pj^siwcf), 0 < (f) < 27r. The trigonometric form of the fourcomplex number u is u = (1 + 1/ tan^ e^ + 1/ tan^ 9-)'^^^ {ej + e+\/2/ tan/9+ + e_\/2/ tan/9_} exp[ei)]. The ampHtude p and tan0-f/v/2, t a n ^ _ / \ / 2 , are multiplicative, and the azimuthal angle 0 is additive upon the multiplication of fourcomplex numbers. There is only one cyclic variable, (/>, and there are two axes v^^V- which play an important role in the description of these numbers, so that these hypercomplex numbers are called polar fourcomplex numbers. If f{n) is an analytic fourcomplex function, then §yf{u)dul{u — UQ) =
54
Coimnutntive Complox Numbers in Four Dimensions
^ ( ^ "• 7) int(wo^i;, r^^;)/(no) , where the functional int takes the values 0 or 1 depending on the relation between the projections of the point UQ and of the curve F on certain planes. A polynomial can be written as a product of linear or quadratic factors, although the factorization may not be unique. The fourcomplex numbers described in this section are a j)articular case for n = 4 of the polar hypercomplex numbers in n dimensions discussed in Sec. 6.1.
3.1
Circular complex numbers in four dimensions
3.1.1
Operations with circular fourcomplex numbers
A circular fourcomplex number is determined by its four components (x, y, z, t). The sum of the circular fourcomplex numbers (x, y, z, t) and (x', y', z', t') is the circular fourcomplex number {x -h x',y + y',z -]- z',t -\- t/). The product of the circular fourcomplex numbers (x^y^z^t) and {x\y\z\f/) is defined in this section to be the circular fourcomplex number (.xx' — yy' — zz' + tt\ xy^ 4- yx' + zt' + tz', xz' -f zx' -h yt' -f ty\ xt' -f tx' - yz' - zy'). Circular fourcomplex numbers and their operations can be represented by writing the circular fourcomplex number [x^y^^^t) as u — x-\-ay + (iz^ 7^, where a, ^ and 7 are bases for which the multiplication rules are a ^ - - l , /3^^ = - l , 7^ = 1, ap = 0a =^ ~ 7 , aj •= ^a = (3, 13^ — ^ft = a.
(3.1)
Two circular fourcomplex numbers u ~ x •{• ay -{• fiz -{- ^yt, v! = x' + ay' + ^z' + 7^' are equal, u — u'^ if and only if x = x'^ y = ?/, z = z'^t = t'. If u — x+ay+pz+'yt, u' = .//-hay'+^^'-f 7^' are circular fourcomplex numbers, the sum u -j- u' and the product \LU' defined above can be obtained l)y applying the usual algebraic rules to the sum {x + ay + liz + ^t) + {x'-{-ay'-]-fiz' + ^t') and to the product {x -f ay + (3z + 'yt){x' -f ay' + /3z' -f -yf/), and grouping of the resulting terms, u + u' = x + x' + a(y + y') + fi(z + z') + ^(t + //),
(3.2)
nu = xx/ - yy - zz + it' + a{xy 4- yx' + zt' + iz) +(i(xz' + zx' + yt' + fy) + j{xt' + tx' - yz' - zy'). Uu.u'.u" tive
(3.3)
are circular fourcomplex niunbers, the multiplication is associa-
{uu')u" = u{u'u")
(3.4)
Circular complex numbers in four dimensions
55
and commutative uu^ = u'u^
(3.5)
as can be checked through direct calculation. The circular fourcomplex zero isO + a O - f / ? - 0 + 7 0 , denoted simply 0, and the circular fourcomplex unity i s l - f a - 0 + ;9-0 + 7 ' 0 , denoted simply 1. The inverse of the circular fourcomplex number u = x + ay + /3z + ^t is a circular fourcomplex number tz' = x' + ay'+ /3z' + jt' having the property that uu' = 1.
(3.6)
Written on components, the condition, Eq. (3.6), is xx' yx' zx' tx'
— yy' — zz' + tt' = 1, + xy' + tz' + zt' = 0, + ty' + xz' + yt' = 0, - zy' - yz' + xt' = 0.
(3.7)
The system (3.7) has the solution ,
X =
x{x^ + y'^ + z^-4
' ;/(-a:^ -y^ + z^y = -4 .' z{-x' z = ^, _ tj-x^
+ y^-z'-t^) -5
f) - 2yzt
(3.8)
,
f) + 2xzt ,
(3.9)
,
(3.10)
+ 2xyt
+ y^ + z'' + t^) - 2xyz
provided that p 7^ 0, where p'i=a:'+y'
+ z' + t' + 2{x'y'' + x'^z' - x^t" - y^z^ + yH^ + zH^)
-Sxyzt.
(3.12)
The quantity p will be called amplitude of the circular fourcomplex number x + ay + 0z + jt. Since p' = pip-,
(3.13)
where
PI = {X + if + (y + z)\ pi = {x- tf + {y- zf,
(3.14)
56
Commutative Complex Numbers in Four Dimensions
a circular foiircomplex number u = x^ay
+ pz + jt has an inverse, unless
x + t = 0,y + z = 0,
(3.15)
x-t
(3.16)
or = 0, y-z
= 0.
Because of conditions (3.15)-(3.16) these 2-dimensional surfaces will be called nodal hyperplanes. It can be shown that if nu' = 0 then either ti = 0, or ix' = 0, or one of the circular fourcomplex numbers is of the form a: -f ay + /3y -f 7X and the other of the form .T' -f ay' — fiy' — ^x'.
3.1.2
Geometric representation of circular fourcomplex numbers
The circular fourcomplex number x-fay 4-/3^+7^ can be represented by the point A of coordinates {x, y, z, t). If O is the origin of the four-dimensional space X, y, z, t^ the distance from A to the origin O can be taken as d^ = x^ -^y^ + z^ ^-f.
(3.17)
The distance d will be called modulus of the circular fourcomplex number x + ay + I3z + 7^, d = \u\. The orientation in the four-dimensional space of the line OA can be specified with the aid of three angles 0, x, V' defined with respect to the rotated system of axes ^4-^
x-t
y-hz
y- z
The variables ^ , 7 ; , T , ( will be called canonical circular fourcomplex variables. The use of the rotated axes ^, v. r, C for the definition of the angles >, X? ^ is convenient for the expression of the circular fourcomplex numbers in exponential and trigonometric forms, as it will be discussed further. The angle (f) is the angle between the projection of A in the plane ^, v and the O^ axis, 0 < (/) < 27r, X is ^he angle between the projection of A in the plane r, C, and the Or axis, 0 < x < 27r, and V^ is the angle betwe^en the line OA and the plane rOC, 0 < t/^ < 7r/2, as shown in Fig. 3.1. The angles (f) and X will be called azimuthal angles, the angle V^ will be called planar angle . The fact that 0 < V^ < n/2 means that ip has the same sign on both faces of the two-dimensional hyperplanc DOC,. The components of the point A in terms of the distance d and the angles (/>, x, i^ are thus —-p:- = d cos (f) sin ijj, v2
(3.19)
Circular complex numbers in four dimensions
57
Figure 3.1: Azimuthal angles >, x ^^^ planar angle ip of the fourcomplex number x + ay + pz + ^t^ represented by the point A, situated at a distance d from the origin O. X — t
y+2 y-
z
V2
= dcosxcos^,
(3.20)
= rising sin V',
(3.21)
= dsinxcosV^.
(3.22)
It can be checked that p^ = \/2dsmip,p= \/2dcos'0. The coordinates X, y^ z^ t in terms of the variables d, 0, Xi ^ ^re X — --^(cosV^cosx + sini/;coS(/)), v2
(3.23)
y— -—(cos t/j sin X + sin i/? sine/)), v2
(3.24)
z = - - p ( ~ c o s ' 0 s i n x + sin'0sin0), V2
(3.25)
f = ~?=(~ cosi/'cosx + sin^cos<^). V2
(3.26)
58
Commutative
Complex Numbers
in Four
Dimensions
T h e angles (^, Xi ^ ^'^^^ l^^ expressed in terms of the coordinates x. y, z, t as sin(t) ={y + z)/p^,
cos<^ = (x + t)/p+,
(3.27)
sinx = iy-z)/p^, cosx = {x-t)/p^,
(3.28)
t a n ^ = /9+//9_.
(3.29)
T h e nodal hyperplanes are {Ot;, for which r = 0, ( = 0, and rOC, for which ^ = 0, t; = 0. For points in the nodal hyperplane (^Ov the planar angle is t/j = 7r/2, for points in the nodal hyperplane TO( the planar angle is i/^ = 0. It can be shown t h a t if wi = Xi-\-ayi+fiz\+yt]^U2 = X2-\-ay2+l3z2+^t2 are circular fourcomplex numbers of amplitudes and angles / 9 i , ^ i , x i i ^ i and respectively P2^4>2^X2^'^2^ \\\en the amplitude p and the angles (^, x ? ^ of the product circular fourcomplex number u\U2 are P = pm.
(3.30)
(l) = (f)i+ (^2, X = Xi + X2, tani/) = t a n ^ i t a n ^ 2 -
(3.31)
T h e relations (3.30)-(3.31) are consequences of the definitions (3.12)-(3.14), (3.27)-(3.29) and of the identities [(a;ia;2 - y\y2 - ^1^2 + tih) + {x\t2 + ^1X2 - y\Z2 mx\y2
z\y2)f
4- y\X2 + zxt2 + ^12:2) + {xiZ2 + Z1X2 + t/1^2 + hy2)f
= [(.^1 + ^1)'^ + {yx + 2ri)2][(j:2 4- t2f -f ( r i + 2:2)'^],
(3.32)
[(a;irr2 - yiy2 - ^1^2 + ^1^2) - (^1^2 + ^1^2 - yi^2 ~ ^11/2)]^ +[(^1^2 + yia;2 + ^l^2 + ^1^2) - (^1^2 + ziX2 + t/it2 -f- ^1^2)]^ = [(^1 - ti)2 + (yi - zif][{x2
- ^2)2 + (y2 - ^2)'],
(3.33)
(a:ia:2 - 1/12/2 - ^1^2 + ^1*2) + (^1^2 + tiX2 - y\Z2 - ^11/2) = {xi + ti){x2 + t2) - {yx + 2:i)(2/2 + Z2),
(3.34)
(a;i.'r2 - I/1I/2 - Z1Z2 + ^1*2) - (^1*2 + *i^2 - ?/i^2 - ^11/2) = (.Ti - ti){x2 - ti) " (yi - ^i)(?/2 " ^2),
(3.35)
{X\y2 + y\X2 + 2:1*2 + ^1^2) + {X\Z2 + Z\X2 + 1/1*2 + Uy2) = (1/1 + ^l)(.^2 + *2) + (y2 -f ^2)(3-2 + *2),
(3.36)
Circular complex numbers in four dimensions
59
{Xiy2 + y\X2 + Zxh + txZ2) - {XxZ-i + ^1X2 + ^1^2 + *lJ/2)
= [yx - zx){x2 - t2) + (y2 - Z2){x2 - t2).
(3.37)
The identities (3.32) and (3.33) can also be written as PI = P1+P2+,
(3.38)
pl=px-P2-,
(3.39)
where p2^ = {xj + tjf + {y^+zjf,pj_ = {xj^tjf + iyj^zjfj
= 1,2.(3.40)
The fact that the ampHtude of the product is equal to the product of the ampUtudes, as written in Eq. (3.30), can be demonstrated also by using a representation of the multiplication of the circular fourcomplex numbers by matrices, in which the circular fourcomplex number u = x + ay-\-l3z + jt is represented by the matrix /
X y z t \ ~y X t —z A= —ztx—y \ t z y X J
(3.41)
The product u = x + ay + ^z + yt of the circular fourcomplex numbers ^1 = ^1 + ayi H- (3zi + 7^1,^2 = 2:2 + <^y2 + 0^2 + 7^2? can be represented by the matrix multiplication A = A1A2.
(3.42)
It can be checked that the determinant det(yl) of the matrix A is detA = / .
(3.43)
The identity (3.30) is then a consequence of the fact the determinant of the product of matrices is equal to the product of the determinants of the factor matrices. 3.1.3
The exponential and trigonometric forms of circular fourcomplex numbers
The exponential function of a hypercomplex variable u and the addition theorem for the exponential function have been written in Eqs. (1.35)(1.36). U u = x + ay + pz + 7*, then expn can be calculated as expn = expx • exp(ay) • exp(^^) • exp(7f). According to Eq. (3.1), ^2m ^ (_i)m^^2m+l ^ (_l)m^^^2m ^ (^j)m^^2m+l ^ ( „ l ) m ^ ^
y' = 1,
(3.44)
60
Commutative Complex Numbers in Four Dimensions
where in is a natural number, so that exp{ay), exp{[jz) and exp(7^) can be written as exp{ay) = cos y + a sin y, exp{l3z) = cos z + 13 sin z,
(3.45)
and exp(7^) = cosh t 4- 7 sinh t.
(3.46)
From Eqs. (3.45)-(3.46) it can be inferred that (cos y + a sin yY^ = cos my + a sin mi(/, (cos JE: + /3 sin z)^ — cos m^ -f ft sin m^:, (cosh ^ + 7 sinh ty^^ — cosh rnt -{- 7 sinh mt.
(3.47)
Any circular fourcomplex number u = x -i- ay + ftz -h jt can be writen in the form x + ay-hftz-}-jt
= e^i+«2/i+/^^i+7^i.
(3.48)
The expressions oi xi,y\,zi^ti as functions of x, y, ^, t can be obtained by developing e^^^, e^^^ and e^^' with the aid of Eqs. (3.45) and (3.46), by multiplying these expressions and separating the hypercomplex components, X — e^^ (cos yi cos z\ cosh ti — sin yi sin zi sinh ^1),
(3.49)
y = e^^(sinyi coszi coshfi + cos 2/1 sin 2:1 sinhti),
(3.50)
z = e^^{cosyi sinz\ cosh^i + siny\ cos2:1 sinhii), t = e^^{—smyi s'mzi cosh^i -h cos 1/1 cos^i sinh^i).
(3.51) (3.52)
From Eqs. (3.49)-(3.52) it can be shown by direct calculation that x^ + 1/ + z^ -f t^ = e^^'^ cosh 2ti,
(3.53)
2(j:< + y;^) = e2^^ sinh 2^1,
(3.54)
so that e^^i = {X^ +y^^Z^
+ /2)2 _ 4(^^ + y^)2
(3 55)
By comparing the expression in the right-hand side of Eq. (3.55) with the expression of p, Eq. (3.13), it can be seen that e^^ = p.
(3.56)
Circular complex numbers in four dimensions
61
The variable ti is then given by cosh2ti - ^ , sinh2ti = ? l i £ ^ : M .
(3.57)
From the fact that p^ = d^ - 4{xt + yz^ it follows that (P/p^ > 1, so that Eq. (3.57) has always a real solution, and ti = 0 for xt + yz ^ 0. It can be shown similarly that ^2 ,
cos2yi =
2 p^
cos 2^1 =
2{xy ^ Zt)
2 ^ ^2 ^ ^2
^—^ P
' sm2?/i =
^ p^
, sin 2^1 = -^^
'
(3-58)
~-^,
(3.59)
P
It can be shown that {x'^ — y^ A- z^ ~ \?)^ < p^^ the equality taking place for xy — zt, and {x'^ + y^ — z^ — t^)'^ < p^, the equality taking place for xz = yt, so that Eqs. (3.58) and Eqs. (3.59) have always real solutions. The variables
1
. 2{xy-zt)
yi = - arcsm 2
^ p^
1
.
2{xz-yt)
' ^i == o ^^^^^^ 2
2 p^
'
t, = 1 rgsinh^i^^lM
(3.60)
z p^ are additive upon the multiplication of circular fourcomplex numbers, as can be seen from the identities {xJ - yy^ - zz' 4- tt'){xy' 4- yx' + zt' + tz') ~{xz' + zx' + yt' -f V)(^*' - y^' - ^y' + ^^') - (a;y - ;^t)(a;'2 - y'^ + ^'^ - t'^) + (a;^ - y^ ^ ;,2 ^ t2)(a;'y' - z't% (3.61) ( r a ' - yy' - 2:2:' + ttO(a;2:' + zx' + yt' -f ty') -{xy' + yx' 4- zt' + ]t^')(^^^ - y^' - V + *^0 = {xz - yt){x'^ 4- y^2 -^ ^'2 - ^^2) -f {x^ Ay^-z^--
t^){x'z' -
y't'), (3.62)
{xx' ~ yy' - zz' 4- tt'){xt' - yz' - zy' + tx') +{xy' + yx' 4- ^^' + tz'){:xz' + 2:0;' 4- yi' 4- ty') = (^i + yz){x'^ 4- y'2 4- ^'' 4-1!^) 4- (a:^ 4- y ' + ^^ + e){x'^
+ y'^'). (3.63)
62
Commutative Complex Numbers in Four Dimensions
The expressions appearing in Eqs. (3.57)-(3.59) can be calculated in terms of the angles 0, x, i' with the aid of Eqs. (3.23)-(3.26) as o P ^ _ J _ 2{xt + yz) p^ sin 2V'' p2 ^ ' ' ^ y ' y ' - ' '
1_ tan2V'
^
= cos(> + x), - ^ S
= «'" <^ + X),
= cos(* - X), ? f e ^
= , i „ , ^ _ ^y
' 3.65
,3.66)
Then from Eqs. (3.57)-(3.59) and (3.64)-(3.66) it results that 0 + X 4>-'X . 1, , / 2/1 = "~2~"' ^^ "= ~ ~ 2 ~ ' ^ ^ 2 ' " ^ ^ " ^ '
/Q ^7x (^'^^^
so that the circular fourcomplex number ?/, Eq. (3.48), can be written as u ~ pexp
^ + x , fA-x
11 ,
/
(3.68)
In Eq. (3.68) the circular fourcomplex number ?i = .r + aj/ 4- y9^ + 7^ is written as the product of the amplitude p and of an exponential function, and therefore this form of u will be called the exponential form of the circular fourcomplex number. It can be checked that exp ( ^ - y ^ ^ j = —7^ + —Y- ^os 0 + - y ^ sm0,
(3.69)
exp( —2—xj = - ^
(3.70)
+ -Y-c:osx+--y-sm(/),
which shows that e^^"^^)*^/^ and e^^"^^^/^ are periodic functions of 0 and respectively x^ '^\^\\ period 27r. The relations between the variables ]i\,z\,t\ and the angles 0,x^^ can be obtained alternatively by substituting in Eqs. (3.49)-(3.52) the expression e'^^ = d/(cosh2ii)^/^, Eq. (3.53), and summing and subtracting of the relations, .T + t '—j=r- —dcos{y\ +^i)sin(7? + 7r/4),
(3.71)
v2 = dcos{yi - ^l) cos(r/ + 7r/4), v2
(3.72)
Circular complex numbers in four dimensions ^ ^
x/2
63
= d sin(yi + zi) sin(7; + 7r/4),
(3.73)
= dsm{yi - Zi)cos{rj + 7r/4),
(3.74)
where the variable r] is defined by the relations cosh^i (cosh2
sinhfi
.
COST/, (cosh2ti)i/2 ,^ =smri, ^^^'"
(3.75)
and when —oo < ti < oc, the range of the variable rj is -n/4 < f] < n/4. The comparison of Eqs. (3.19)-(3.22) and (3.71)-(3.74) shows that 0 = yi + ^1, X = yi - ^1^ V^ = ^ + 7r/4.
(3.76)
It can be shown with the aid of Eq. (3.46) that exp f - 7 l n t a n V^j = ^.^^
^^^ [cos(z/^ - 7r/4) + 7sin(V^ -
n/4)]. (3.77)
The circular fourcomplex number w, Eq. (3.68), can then be written equivalently as u = d{cos{ijj — 7r/4) +7sin(7/? — 7r/4)}exp\a^
+ ptix
.(3.78)
In Eq. (3.78), the circular fourcomplex number u = x + ay + pz + ^yt is written as the product of the modulus d and of factors depending on the geometric angles 0, x and ^ , and this form will be called the trigonometric form of the circular fourcomplex number. U u\,U2 are circular fourcomplex numbers of moduh and angles di,)i, Xi, ^1 and respectively ^2^ 02i X27 V^2i the product of the planar factors can be calculated to be [cos('0i - 7r/4) + 7sin('0i - n/4)][cos{ip2 - 7T/4) + 7sin('02 - ^/4)] = [cos(^i - -02) - 7 cos(V^i + V^2)].
(3.79)
The right-hand side of Eq. (3.79) can be written as COs('0i — -02) — 7COS('01 + -02) = [2(cos^ tpi cos^ -02 + sin^ i/^i sin^ '^p2)Y^'^[cos{^p - 7r/4) +7sin(^-7r/4)],
(3,80)
where the angle -0, determined by the condition that tan('0 - 7r/4) = ~ cos('0i + 1/^2)/ cos('0i - 1/^2)^
(3.81)
64
Commutative Complex Numbers in Four Dimensions
is given by tani/j = tan?/^i tant/^2 , which is consistent with Eq. (3.31). It can be checked that the modulus d of the product Uiti2 is d = V2did2 (cos^ tpi cos'"^ ^f)2 + sin^ ipi sin^ ^2)
3.1.4
•
(3.82)
Elementary functions of a circular fourcomplex variable
The logarithm u\ of the circular fourcomplex number ix, u\ = Inu^ can be defined as the solution of the equation w = e^S
(3.83)
written explicitly previously in Eq. (3.48), for ui as a function of 7i. From Eq. (3.68) it results that \nu = Inp -\- l-jlntaml) + a ~ ^
-\' P^-^.
(3.84)
It can be inferred from Eqs. (3.30) and (3.31) that l n ( W ) =:lnti + lnix',
(3.85)
up to multiples of 7r(a + /3) and 7r(cv — ^ ) . The power function u^ can be defined for real values of n as ^m^^mln^
(3.86)
The power function is multivalued unless n is an integer For integer n, it can be inferred from Eq. (3.85) that ( W ) ^ - ?/^ ^/'^^
(3.87)
If, for example, m = 2, it can be checked with the aid of Eq. (3.78) that Eq. (3.86) gives indeed [x + ay + pz + ^tf = x^ -y^ ~ z^ + t^ + 2a{xy + zt) + 2f5{xz + yt) + 2y{2xt - yz). The trigonometric functions of the hypercomplex variable u and the addition theorems for these functions have been written in Eqs. (1.57)(1.60). The cosine and sine functions of the hypercomplex variables ay.^z and 7f can be expressed as cos ay = coshy, sin ay = asinhy,
(3.88)
cos/3y = coshy, sin/Jy = /3sinhy,
(3.89)
cos7y = cosy, sin7y = 7siny.
(3.90)
Circular complex numbers in four dimensions
65
The cosine and sine functions of a circular fourcomplex number x + ay -{/3z + jt can then be expressed in terms of elementary functions with the aid of the addition theorems Eqs. (1.59), (1.60) and of the expressions in Eqs. (3.88)-(3.90). The hyperbolic functions of the hypercomplex variable u and the addition theorems for these functions have been written in Eqs. (1.62)-(1.65). The hyperbolic cosine and sine functions of the hypercomplex variables ay, I3z and jt can be expressed as cosh ay = cosy, sinh ay = a s i n y ,
(3.91)
cosh^Sy = cosy, sinh/3y = ^Ssiny,
(3.92)
cosh7y = coshy, sinh7y = 7 sinh y.
(3.93)
The hyperbolic cosine and sine functions of a circular fourcomplex number x-hay+pz+^t can then be expressed in terms of elementary functions with the aid of the addition theorems Eqs. (1.64), (1.65) and of the expressions in Eqs. (3.91)-(3.93). 3.1.5
Power series of circular fourcomplex variables
A circular fourcomplex series is an infinite sum of the form 00 + «i + «2 H
-^an-\
,
(3.94)
where the coefficients an are circular fourcomplex numbers. The convergence of the series (3.94) can be defined in terms of the convergence of its 4 real components. The convergence of a circular fourcomplex series can however be studied using circular fourcomplex variables. The main criterion for absolute convergence remains the comparison theorem, but this requires a number of inequalities which will be discussed further. The modulus of a circular fourcomplex number u = x i- ay -\- f3z + ^yt can be defined as \u\ = {x^ + y'^ + z'^ + f)^/\
(3.95)
so that, according to Eq. (3.17), d = \u\. Since \x\ < |w|,|y| < I^U-^I < I^M^I ^ l^h ^ property of absolute convergence established via a comparison theorem based on the modulus of the series (3.94) will ensure the absolute convergence of each real component of that series. The modulus of the sum ui + U2 oi the circular fourcomplex numbers ui^U2 fulfils the inequality | | w i | ~ | W 2 | | < \ui-\'U2\
< | ^ l | + |W2|.
(3.96)
66
Commutative Complex Numbers in Four Dimensions
For the product the relation is \uiU2\ < V2\ui\\u2l
(3.97)
which replaces the relation of equality extant for regular complex numbers. The equality in Eq. (3.97) takes place for xi = 2 — ^ / 2 . The modulus of a product, which has the property that 0 < |uiU2|^ becomes equal to zero for x\ = ^i,?/i = ^1,^2 = -^2,?/2 = -^2 or XX = -i\,y\ = ~-Zi,X2 = ^2,1/2 = ^2, as discussed after Eq. (3.16). In Eq. (3.82), the latter situation corresponds to tpi = 0, ip2 = 7r/2 or ip\ = 0, ?/?2 = '7r/2. It can be shown that x'^ -hy'^ + z'^ + t'^ < \u'^\ < \f2{x^ -f y^ + ^^ + i^),
(3.98)
The left relation in Eq. (3.98) becomes an equality, x'^ + y'^ + z^ + ^^ = |K^|, for xi + yz = 0. This condition corresponds to ^ipi =: ip2 = 7r/4 in Eq. (3.82). The inequality in Eq. (3.97) implies that |^/|<2('-i)/2|n|^
(3.99)
From Eqs. (3.97) and (3.99) it results that \m/\ < 2^/^\a\\4.
(3.100)
A power series of the circular fourcomplex variable w is a series of the form oo + a[U + a2U^ 4 . . . . -)_ aiu -f • • •.
(3.101)
Since
E«/^i<E2'^'l«HH^
(3.102)
a sufficient condition for the absolute convergence of this series is that H„, ^\-MM < 1. /->oo \ai\ Thus the series is absolutely convergent for |tx| < c,
(3.103)
(3.104)
Circular complex numbers in four dimensions
67
where c= l i m - ^ ^ ^ .
(3.105)
The convergence of the series (3.101) can be also studied with the aid of the transformation x + ay + Pz + 'yt = \/2(ei^ + eiv + e2T + e2C),
(3.106)
where ^,t',T, ( have been defined in Eq. (3.18), and ei = - 2 " ^ , ei = — 2 - , 62 = - y - , 62 = — 2 - .
(3.107)
The ensemble ei, ei, e2, e2 will be called the canonical circular fourcomplex base, and Eq. (3.106) gives the canonical form of the circular fourcomplex number. It can be checked that e i = e i , ef = ~ei, eiei = ei, 62 = 62, e^ = ""^2, ^262 = 62, 6x62 = 0, eie2 = 0, eie2 = 0, 6261 = 0.
(3.108)
The moduli of the bases in Eq. (3.107) are |ei| = ^ ,
|ei| = - ^ , \e2\ = - = , \e2\ = ~ ^ ,
(3.109)
and it can be checked that \x-\-ay + /3z + -ft\'^ =.^'^ + v^ + r^ + ( ^
(3.110)
If n = u^u^\ the components ^,t>,T, ( are related, according to Eqs. (3.34)-(3.37) by ^ = V2{ee - v'v"), V = V2{^'v" + v'n. C = v^(r'C" + CV'),
r = x/2(rV" -
('("). (3.111)
which show that, upon multiplication, the components ^,t? and r, ( obey, up to a normalization constant, the same rules as the real and imaginary components of usual, two-dimensional complex numbers. If the coefficients in Eq. (3.101) are ai = aio + aan + (3ai2 + iciis,
(3.112)
and An = a/o+a/3, i n = aa+a/2, A12 = aio-ais,
i/2 = a a - a / 2 , (3.113)
68
Commutative Complex Numbers in Four Dimensions
the series (3.101) can be written as oc
^ 2 ' / ^ [{eiAn + e,i/,)(ei^ + civ)' + (e2^<2 + e2Ai2){e2T + ezC)'] • 1=0
(3.114) Thus, the series in Eqs. (3.101) and (3.114) are absolutely convergent for p+
(3.115)
where Ci =
lim
yi'^M
-^
K + -^«l
..,:^
ry;r. ^2 = H m
t/2-
(3.116) It can be shown that c = (l/\/2)min(ci,C2), where min designates the smallest of the numbers cI, C2. Using the expression of |M| in Eq. (3.110), it can be seen that the spherical region of convergence defined in Eqs. (3.104), (3.105) is included in the cylindrical region of convergence defined in Eqs. (3.115) and (3.116). 3.1.6
Analytic functions of circular fourcomplex variables
The analytic functions of the hyperconiplex variable u and the series expansion of functions have been discussed in Eqs. (1.85)-(1.93). If the fourcomplex function /(?/) of the fourcomplex variable w can be expressed in termsof the real functions P(.x,y, 2;, <), Q{x,y,z,t), R{x,y,z,t), S{x,y,z,t) of real variables x, y, z, t as / ( « ) = P{x, y, z, f)^-aQ{x, y, 2;, t)-\-(5R{x, y, z, t)+'yS{x, y, z, t), (3.117) then relations of equality exist between partial derivatives of the functions P, Q, R, S. These relations can be obtained by writing the derivative of the function / as 1
\9P ^
dP ^
dP ^
dP ^^
lim — — ^r-^'"^ + -:^^y + 11-^2: + -^r-A< «->«o Ax + oAy + pAz + 7Ai L ox ay oz ot ^(dR^ /dS ^
dR^
OR.
dR.\
OS ^
dS ^
dS .\]
,,,,„,
Circular complex numbers in four dimensions
69
where the difference appearing in Eq. (1.86) is u — uo = Ax + aAiy-f^A^-h 7At. The relations between the partial derivatives of the functions P, Q, iJ, 5 are obtained by setting succesively in Eq. (3.118) Ax -> 0, Ay =^ Az = At = 0; then Ay -^ 0, Ax = A^ = Ai = 0; then Az -^ 0,Ax = Ay = At = 0; and finally At -^ 0, Ax = Ay = Aja: = 0. The relations are dP _ dQ_ dx dy
dR dS dz ~ dt'
(3.119)
dQ _ dx
dS dR dP ~~dz ~ dt' dy
(3.120)
dR dx
ds dy
ds _ dx
dP dz
dQ dt'
(3.121)
dR _ dQ _dP dy dz ~ dt'
(3.122)
The relations (3.119)-(3.122) are analogous to the Riemann relations for the real and imaginary components of a complex function. It can be shown from Eqs. (3.119)-(3.122) that the component P is a solution of the equations
dx'^
dy'^ ' dx'^ d'^P &^P ^ + -5^=0,
dz^
' dy'^
dt'^ (3.123)
^ _ ^ - 0 ^ - - ^ - ^ 0 dx^ dt^ ~ ^ ' 5y2 dz' ~ ^ '
(3124) ^^'^"^^^
and the components Q, R, S are solutions of similar equations. As can be seen from Eqs. (3.123)-(3.124), the components P,Q,R,S of an analytic function of circular fourcomplex variable are harmonic with respect to the pairs of variables a;,y;a;,z;y,t and z^t^ and are solutions of the wave equation with respect to the pairs of variables x, t and y, z. The components P, Q, i?, S are also solutions of the mixed-derivative equations ^ ^ _ a ^ dxdy dzdt'
^ T ^ _ ^ dxdz dydt'
d^p _ dxdt
d^p dydz'
and the components Q, R, S are solutions of similar equations.
70
3.1.7
Commutative Complex Numbers in Four Dimensions
Integrals of functions of circular fourcomplex variables
The singularities of circular fourcomplex functions arise from terms of the form l/{u—U{))^, with rn > 0. Functions containing such terms are singular not only at u = IIQ^ but also at all points of the two-dimensional hyperplanes passing through UQ and which are parallel to the nodal hyperplanes. The integral of a circular fourcomplex function between two points A, B along a path situated in a region free of singularities is independent of path, which means that the integral of an analytic function along a loop situated in a region free from singularities is zero, <j> f{u)du
= 0,
(3.126)
where it is supposed that a surface S spanning the closed loop V is not intersected by any of the two-dimensional hyperplanes associated with the singularities of the function /(w). Using the expression, Eq. (3.117), for f{u) and the fact that du = dx -f- ady -f I5dz + 7^^, the explicit form of the integral in Eq. (3.126) is i f{u)du = I [{Pdx - Qdy - Rdz + Sdt) +a{Qdx -f- Pdy 4- Sdz -h Rdt) 4- P{Rdx + Sdy + Pdz + Qdt) -hjiSdx - Rdy - Qdz -f Fdi)].
(3.127)
If the functions P, Q, i?, S are regular on a surface S spanning the loop r , the integral along the loop F can be transformed with the aid of the theorem of Stokes in an integral over the surface E of terms of the form
dP/dy+dQ/dx,
dP/dz^dR/dx,
dP/dt-dS/dx,
OQ/dz-dR/dy,
dQ/dt-h
dS/dy, dR/dt + dSj'dz and of similar terms arising from the a, /3 and 7 components, which are equal to zero by Eqs. (3.119)-(3.122), and this proves Eq. (3.126). The integral of the fmiction {u - ?/o)^ on a closed loop F is equal to zero for '\n a positive or negative integer not equal to -1, i[u-
uo^du
= 0, m integer, m / - 1 .
(3.128)
This is due to the fact that J{u - uoY^du - [u - u^))^^^l(m + 1), and to the fact that the function {u - UQY'^'^^ is singlevalued for m an integer. The integral § dii/(u-UQ) can be calculated using the exponential form (3.68), u - Ko -= pexp pexpla^^^ (^"—^ ( « ^ ^ — + f^^Y^
V^) ,, + 7 In tan V^^
(3.129)
Circular complex numbers in four dimensions
71
so that u — uo
"
^ 4-—P^d>-f-—r-^dx + 7dIntanV^. p 2 2
(3.130)
Since p and In tan ^ are singlevalued variables, it follows that jf^dp/p = 0,^(ilntani/? = 0. On the other hand, <> / and x ^^^ cyclic variables, so that they may give a contribution to the integral around the closed loop r . Thus, if C^ is a circle of radius r parallel to the (,Ov plane, and the projection of the center of this circle on the ^Oi) plane coincides with the projection of the point UQ on this plane, the points of the circle C^ are described according to Eqs. (3.18)-(3.22) by the equations ^ = ^0 -I- r sin ip cos 0, v = VQ -i-rsim/j sin >, r = TQ + r cos ip cos x, C = Co + r cost/? sin X,
(3.131)
for constant values of x and ip, ip ^ 0,7r/2, where UQ = XQ + ayo + ^ZQ + 7*0, and ^0^^0^'7'o^Co are calculated from xo^yo^zojio according to Eqs. (3.18). Then
i
Hit
= 7r(a + ^ ) .
(3.132)
If C^ is a circle of radius r parallel to the rOC plane, and the projection of the center of this circle on the rOC plane coincides with the projection of the point UQ on this plane, the points of the circle C_ are described by the same Eqs. (3.131) but for constant values of (p and t/?, t/? 7^ 0,7r/2. Then
i
dii U-UQ
C-
7r(a~/3).
(3.133)
The expression of ^ du/{u — UQ) can be written as a single equation with the aid of a functional int(M, C) defined for a point M and a closed curve C in a two-dimensional plane, such that
fI i ii f" M is an interior point of C, • ^/%f ^\ - t ( M , C ) = \{ n0 if . . M is exterior to C.
,„ ,_ ., ^'-'''^
With this notation the result of the integration along a closed path F can be written as
i
du
r II -
iio
= 7r(a + /3) int('Uo^i;,r^^) + 7r(Q:-^) int('UoTC,r^c)'(3-135)
where WOCUI^OT< and r^y^Frc; are respectively the projections of the point uo and of the loop F on the planes (^v and r(.
72
Commutative Complex Numbers in Four Dimensions
Figure 3.2: Integration path F and the pole UQ, and their projections ^iv^^rc and uo^y^Uor(^ on the planes ^v and respectively r ( . If f{u) is an analytic circular fourcomplex function which can be expanded in a series as written in Eq. (1.89), and the expansion holds on the curve F and on a surface spanning F, then from Eqs. (3.128) and (3.135) it follows that f{u)du = 7r[(a + /3) int(wo^^,F^^) + (Q:-/3) int(iiorc,F^^)]/(WQ),
i
(3.136)
where F^^,Fj(; are the projections of the curve F on the planes ^?; and respe^ctively rC, as shown in Fig. 3.2. Substituting in the right-hand side of Eq. (3.136) the expression of J[u) in terms of the real components P,Q,R,S, Eq. (3.117), yields
i
f{u)du
ir u -
= 7r[~(l + 7)(Q + R) + [a + p){P + S)] intiuQ^y,F^y)
UQ
+7r[-(l - 7)(Q -R)
+ {a- /3)(P - S)] int(noro T,^),
(3.137)
where P, Q, /?, S are the values of the components of f at u = UQ. If f{u) can be expanded as written in Eq. (1.89) on F and on a surface spanning F, then from Eqs. (3.128) and (3.135) it also results that f{u)du r {u — uo)^'^
i
^ TTn m!
+ ( a - / ? ) int(t/orC,rrc)]r"'^(»o),
(3.138)
Circular complex numbers in four dimensions
73
where it has been used the fact that the derivative J^^^^UQ) of order m of f{u) at w = Wo is related to the expansion coefficient in Eq. (1.89) according to Eq. (1.93). If a function f{u) is expanded in positive and negative powers of u — Uj, where Uj are circular fourcomplex constants, j being an index, the integral of / on a closed loop F is determined by the terms in the expansion of / which are of the form aj/{u — Uj)^
/(«) = --- + E -U ^—^II + ---
(3.139)
Then the integral of / on a closed loop F is
i
f{u)du = 7r(a + /3) ^
int{uj^y, T^y)aj
j
+7r(a - /3)Y^ mt{ujrc, Trc)aj.
3.1.8
(3.140)
Factorization of circular fourcomplex polynomials
A polynomial of degree m of the circular fourcomplex variable u = x + ay-^ ^z -\- 7^ has the form Pm{u)
= U'"" + aiu"^'^
+'"+
arr,-.iU + a,n,
(3.141)
where the constants are in general circular fourcomplex numbers. It can be shown that any circular fourcomplex polynomial has a circular fourcomplex root, whence it follows that a polynomial of degree m can be written as a product of m linear factors of the form u — Uj^ where the circular fourcomplex numbers Uj are the roots of the polynomials, although the factorization may not be unique, m
i'm(?^) = n ( " - " i ) -
(3-142)
The fact that any circular fourcomplex polynomial has a root can be shown by considering the transformation of a fourdimensional sphere with the center at the origin by the function u'". The points of the hypersphere of radius d are of the form written in Eq. (3.78), with d constant and 0 <
w"* = d^ exp (am^^
+
0rn^^\
[cosiij - 7r/4) + 7sin(^ - 7r/4)]'".
(3.143)
74
Commutative Complex Nmnbers in Four Dimensions
It can be shown with the aid of Eq. (3.82) that uexp(a*±X.,tx)
= |u|,
(3.144)
so that [cos(^ - 7r/4) + 7 sin(V' - 7r/4)]"' exp (am~^
+
I3m^-^^
= \[cos{i> - 7r/4) + 7sin(i/> - 7r/4)]'"|.
(3.145)
The right-hand side of Eq. (3.145) is m
|(cose-f T s i n e r p = ^
Cf^cos^^"^^ esin^^ e,
(3.146)
k=0
where e = "ip — 7r/4, and since Cf^ > C^, it can be concluded that |(cos6 4-7sin6)^|2 > 1.
(3.147)
Then
which shows that the image of a four-dimensional sphere via the transformation operated by the function u^ is a finite hypersurface. Ifu' = w^, and u' = ci'[cos(t/;' - 7r/4) + 7sin(V^' - 7r/4)] exp (a^
^^ +^^T^) ' (3.149)
then 0' = m(/), x' = rnx, t a n ^ ' = tan"^ ^p.
(3.150)
Since for any values of the angles 0', x', ^ ' there is a set of solutions 0, x? ^ ^f Eqs. (3.150), and since the image of the hypersphere is a finite hypersurface, it follows that the image of the four-dimensional sphere via the function u^ is also a closed hypersurface. A continuous hypersurface is called closed when any ray issued from the origin intersects that surface at least once in the finite part of the space. A transformation of the four-dimensional space by the polynomial Pmb^) will be considered further. By this transformation, a hypersphere of radius d having the center at the origin is changed into a certain finite closed surface, as discussed previously. The transformation of the four-dimensional
Circular complex numbers in four dimensions
75
space by the polynomial Pm(^) associates to the point u = 0 the point /(O) = am, and the image of a hypersphere of very large radius d can be represented with good approximation by the image of that hypersphere by the function u^. The origin of the axes is an inner point of the latter image. If the radius of the hypersphere is now reduced continuously from the initial very large values to zero, the image hypersphere encloses initially the origin, but the image shrinks to am when the radius approaches the value zero. Thus, the origin is initially inside the image hypersurface, and it lies outside the image hypersurface when the radius of the hypersphere tends to zero. Then since the image hypersurface is closed, the image surface must intersect at some stage the origin of the axes, which means that there is a point ui such that f{ui) = 0. The factorization in Eq. (3.142) can then be obtained by iterations. The roots of the polynomial Pm can be obtained by the following method. If the constants in Eq. (3.141) are a/ = a/o + o^a/i + /3a/2+7a/3, and with the notations of Eq. (3.113), the polynomial Pm{u) can be written as
Pm = J^ 2(—')/2(ei An + e i i a ) ( e i e +
eyvr"'
rn
+ E 2(^-')/'(e2A|2 + e2Ai2){e2T H- e2C)"^"^
(3.151)
where the constants Aik^Aik^k = 1,2, are real numbers. Each of the polynomials of degree m in ei^ -f eiv,e2T 4- e2C in Eq. (3.151) can always be written as a product of linear factors of the form ei(^ - ^p) -\- e\{v - Vp) and respectively e2{r — Tp) + e2(C — Cp)-) where the constants (,p^Vp^Tp^(p are real, m
E 2(--^)/2(ei Aa + eiAn){e,i
+ eii;)—'
1=0 m
= l[2"'^^ex{^-^p)+e,{v-vp)},
(3.152)
X ; 2("'-')/2(62^(2 + e2i,2)(e2T + esC) m-l 1=0 m
= n 2 " ^ ' {«2(T - Tp) + e2(C - C,)} •
(3.153)
76
Commutative Complex Numbers in Four Dimensions
Due to the relations (3.108), the polynomial jPm(^^) <^*an be written as a product of factors of the form m Prn{u) = n
2 " / ' {^l(^ - ^P) + ^^l(^' - ^V) + ^ 2 ( T - T^) + e2(C ~ Cp)} •
(3.154) This relation can be written with the aid of Eq. (3.106) in the form (3.142), where Up = V2{ei^p + eiVp + e2Tp + e2Cp)-
(3.155)
The roots ei(^p+eiVp and e2Tp-f e2Cp defined in Eqs. (3.152) and respectively (3.153) may be ordered arbitrarily. This means that Eq. (3.155) gives sets of m roots ?/i,...,?/„, of the polynomial Pm('^^)^ corresponding to the various ways in which the roots ei^p-\- ei^jp and e2Tp + e2(,p are ordered according to p for each polynomial. Thus, while the hypercomplex components in Eqs. (3.152), Eqs. (3.153) taken separately have unique factorizations, the polynomial Pni{u) can be written in many different ways as a product of linear factors. The result of the circular fourcomplex integration, Eq. (3.140), is however uuicjue. If, for example, P{u) = ?/^ 4- 1. the possible factorizations are P = (7i — ei ~ e2){u -f Ci 4- ^2) and P = (?/ - ex + e2)(?i + ei — 62), which can also be written as u^ -\- 1 = {u — a)(it -f a) or as ?i'^ 4- 1 = (w — P){u 4- P). The result of the circular fourcomplex integration, Eq. (3.140), is however unique. It can be checked that {±e\ ± €2)^ ~ —e] — <^2 = — 13.1.9
R e p r e s e n t a t i o n of circular fourcomplex n u m b e r s by irreducible m a t r i c e s
If T is the unitary matrix,
(3.156)
T = T2
"
'
'-k
it can bo shown tliat the matrix TUT T f / T - ' = I •;,'
;;
1.
' has the form (3.157)
Hyperbolic complex numbers in four dimensions
77
where U is the matrix in Eq. (3.41) used to represent the circular fourcomplex number u. In Eq. (3.157), 14,^2 are the matrices
In Eq. (3.157), the symbols 0 denote the matrix (3.159) The relations between the variables x + t,y + z^x — t, y — z for the multiplication of circular fourcomplex numbers have been written in Eqs. (3.34)(3.37). The matrix TUT~^ provides an irreducible representation [7] of the circular fourcomplex number u in terms of matrices with real coefficients.
3.2
Hyperbolic complex numbers in four dimensions
3.2.1
Operations with hyperbolic fourcomplex numbers
A hyperbolic fourcomplex number is determined by its four components (x, y, z, t). The sum of the hyperbolic fourcomplex numbers (x, y, ^, t) and [x\ y', z\ t') is the hyperbolic fourcomplex number (x -hx'^y~\- y', z + z\t-\t/). The product of the hyperbolic fourcomplex numbers (x^y^z^t) and {x\y'^z'^t') is defined in this section to be the hyperbolic fourcomplex number {xx^ -f yy' + zz^ + tt\ xy' + yx' + zi! + tz', xz* + zx' + yt' + ty^, xt' + tx'-f y2' + ^y'). Hyperbolic fourcomplex numbers and their operations can be represented by writing the hyperbolic fourcomplex number [x,y^z,t) as u = X -{- ay + I5z -i- yt, where a, (3 and 7 are bases for which the multiplication rules are a =1^/3
= 1, 7^ = 1, a/3 = /3a = 7, 07 = 7 a = /3, /37 = 7/3 = a. (3.160)
Two hyperbolic fourcomplex numbers ii = x + ay + Pz + jt, u' ~ x' + ay' + ^z' + 7<' are equal, u = n', if and only if x = x',y = y',z ^ z'A ^ <'. If u — X -{• ay ^^ (3z -\- 7^, u' — x' + ay' + ^z' + jt' are hyperbolic fourcomplex numbers, the sum u + u' and the product uu' defined above can be obtained by applying the usual algebraic rules to the sum (x + ay + Pz + yt) + {x' +
78
Commutative Complex Nmnbers in Four Dimensions
ay' + pz' + 7^') and to the product {x -f ay + flz + jt){x' + ay' -f- pz' -h 7^'). and grouping of the resulting terms. u + u' = x + x' + a(y + y') + ft{z + z') + j{t + t'),
(3.161)
mi = xx' + yy' -f zz' -f tt' -f a(x7j -f yx' + zt! -h tz') ^I5{xz' + zx! + yi! + ^](/') -f 7(:rf' + f:r' -f yz' -f ^y').
(3.162)
If u, u\ u" are hyperboUc fourcomplex numbers, the multipHcation is associative {UIL)U'
= u{v!u'')
(3.163)
and commutative uu ^u'u,
(3.164)
as can be checked through direct calculation. The hyperbolic fourc:omplex zero is 0+a-O-f/?0-f7'0, denoted simply 0, and the hyperbolic fourcomplex unity is l + a O - f / 3 - 0 4 - 7 - 0 , denoted simply 1. The inverse of the hyperbolic fourcomplex number u ~ x-\- ay -i-pz-hyt is a hyperbolic fourcomplex immber u' = x/ -f ay' -f flz' + jt' having the property that W = 1.
(3.165)
Written on components, the condition, Eq. (3.165), is xx' yx' zx' tx'
-\-yy' -{- zz' -^tf/ = 1, -h xy' + tz' + zt' = 0, + ttj' + xz' + yt' = 0, -\- zy' + yz' 4- xt' = 0.
(3.166)
The system (3.166) has the solution
V
' y =
y{-^^^y^~^^"t')-^2xzt .
(3.168)
V
,' . ti--^-v'-z^
+ f) + 2.yz^
^3^^^^
Hyperbolic complex numbers in four dimensions
79
provided that v' ^ 0^ where ly = x^ + y4 ^ ;^4 + t^ ^ 2{x'^y'^ + x^z^ + xH'^ + y^z^ + yh^ + zh^) +8xyzt.
(3.171)
The quantity u can be written as u = ss's"s''\
(3.172)
where s = x + y + ^ + < , / = x-y+z-t,s''
= rr+y-^:-/,^"' = a:-y-^+f.(3.173)
The variables 5,5', 5", 5'" will be called canonical hyperbolic fourcomplex variables. Then a hyperbohc fourcomplex number u — x + ay + /3z + jt has an inverse, unless 5 = 0, or 5' = 0, or 5" = 0, or 5'" = 0.
(3.174)
For arbitrary values of the variables x, y, z^ i, the quantity ly can be positive or negative. If ^' > 0, the quantity /j. = u^^"^ will be called amplitude of the hyperbolic fourcomplex number x + ay + pz + 'yt. The normals of the hyperplanes in Eq. (3.174) are orthogonal to each other. Because of conditions (3.174) these hyperplanes will be also called the nodal hyperplanes. It can be shown that if mi' = 0 then either i/ = 0, or w' = 0, or g, q' belong to pairs of orthogonal hypersurfaces as described further. Thus, divisors of zero exist if one of the hyperbolic fourcomplex numbers w, u' belongs to one of the nodal hyperplanes and the other hyperbolic fourcomplex number belongs to the straight line through the origin which is normal to that hyperplane, X + y -i- z + t = 0, and a;' = y^ = z' = t',
(3.175)
x-y
(3.176)
or + z-t=^0,
diud x' = -y
=z' = -t\
or x + y-z-t^Q,
and x' = y' = - z ' = - t ' ,
(3.177)
X - y - z + t = 0, and x' = -~y' = -z' = t'.
(3.178)
or
80
Commutative Complex Numbers in Four Dimensions
Divisors of zero also exist if the hyperbolic fourcomplex numbers n,u' belong to different members of the pairs of two-dimensional hypersurfaces listed further, X + y = 0^ z + t = Q and x/ - y' = 0, 2' - t' = 0,
(3.179)
or x + z = 0, y + t = 0 8ind x' -z'
= 0, y -t'
:=0,
(3.180)
y + z = 0, x + t = 0 and y' - z' = 0, x/ - ^' = 0.
(3.181)
or
3.2.2
Geometric representation of hyperbolic fourcomplex numbers
The hyperbolic fourcomplex number x -\- ay -i- /3z + jt can be represented by the point A of coordinates {x,y,z,t). If O is the origin of the fourdimensional space x^y^z^t^ the distance from A to the origin O can be taken as (f ^ x ^ + y^ + z^ + f,^.
(3.182)
The distance d will be called modulus of the hyperbolic fourcomplex number x -\- ay + ftz -^ 7/, d = \u\. Ifw = x-^ay-^f3z+^t,U] = ^i-fayi-h/32]+7^1,1x2 = ^2+<^y2+^^2+7^2, and u — u\U2, and if Sj = Xj -f yj + Zj + tj, s'j = Xj - yj -f zj - tj, s] = Xj + yj - Zj Sj -=Xj-yj~
Zj + tj,
tj,
(3.183)
for j = 1,2, it can be shown that /
S = S\S2,
f
I
fl
S = S1S2, S
II
If
III
— .S|.S2, S
III
III
— .Sj ^2 .
/ O 1 O /f A
(3.184)
The relations (3.184) are a consequence of the identities (:ri.T2 + 2/ji!/2 + z\^2 + ^ih) + (^12/2 + ?/i^2 + ^1*2 + ^1^2) +{x\Z2 + ZYX2 + y\t2 + UV'i) + (^'1^2 + hx2 + y\Z2 + z\y2) = {xi +yi+ z\ + tx){x2 + ?/2 + Z2 + ^2), (3.185) (,'rij;2 + VWi + z\Z2 + ^1^2) - (^i2/2 + 2/1^:2 + ^1^2 + ^1^2) +(0:12:2 + 2:1^:2 + yiti + hV'i) - (^1^2 + t\^2 + yi2:2 + ziy2) = (:i^i - 2/1 + ^i - ti){^'2 ~-y2 + Z2~ ^2),
(3.186)
Hyperbolic complex numbers in four dimensions
81
(xi:r2 + yiy2 + ^1^2 +
(3.189)
^j = Sjs'js'jsfj = 1,2.
(3.190)
where
The hyperbohc fourcomplex numbers ___ l + a + P + ^ f _ l - a + ^ ~ 7 „_ l+ a~^~7 ^~ 4 ' ^ "~ 4 ' ^ "^ 4
e- = iz^zA+I
(3.191)
are orthogonal, ee' = 0, ee" = 0, ee'" = 0, e'e" = 0, e'e'" = 0, e"e'" = 0,
(3.192)
and have also the property that e2 = e, e'2 = e', e"^ = e", e'"^ = e'".
(3.193)
The hyperbolic fourcomplex number u = x -{- ay -\- ^z -{- ^t can be written as x + ay + I3z + ^t == {x + y + z + t)e + {x - y + z - t)e^ ^{x + y - z - t)e" + {x-y~z
+ t)e"\
(3.194)
or, by using Eq. (3.173), u = se + s'e' + s"e" + s'"e'",
(3.195)
The ensemble e,e\e'\e"' will be called the canonical hyperbolic fourcomplex base, and Eq. (3.195) gives the canonical form of the hyperbolic fourcomplex number. Thus, if Uj = SjC + s'jc' + s'je'^ 4- 5^-'e'", j = 1,2, and u = U1U2, then the multiplication of the hyperbolic fourcomplex numbers
82
Commutative Complex Numbers in Four Dimensions
is expressed by the relations (3.184). The moduli of the bases e,e',e",e'' are
\e\ = \. \e'\
1 \e"\ = i 2'
\e"'\ = \.
(3.196)
The distance rf, Eq. (3.182), is given by
d^^Us^
+ s'^ + s"^ + s"''),
(3.197)
The relation (3.197) shows that the variables s, s\ s", s"' can be written as 5 = 2d cos-0 cos (/), s' = 2dcos'0sin(/), .s" = 2(isini/^cosx, 5'" = 2(isinV^sinXi
(3.198)
where 0 is the azimuthal angle in the 5,5' plane, 0 < 0 < 27r, x is the azimuthal angle in the 5", 5'" plane, 0 < x < 27r, and ij) is the angle between the line OA and the plane s,s\ 0 < ip < 7r/2. The variables x,y^z,t can be expressed in terms of the distance d and the angles (/>, x, '^ as X = {d / 2) {cos ijj cos (p + cos 1/? sin 0 + sin'0 cos x + sin-0 sin x), y = (d!/2)(cos'0cos) - cos 1/) sin (/> + sin i/; cos x - sin t/? sin x), (3.199) z — {d/2) (cos t/j cos (/) + cos X/J sin (f) — sin ^ cos x — sin ip sin x), t = (d/2) (cos-0 cos 0 — cos ip sin (f) — sin-0 cos x + sin ^ sin x)If u = ttitX2, and the hyperconiplex numbers 1x1,1x2 are described by the variables di, 0i, Xh ^1 and respectively ^2,02i X2,^2, then from Eq. (3.198) it results that tan (j) ~ tan 01 tan 02? tan x = tan xi tan X2i tan^ ij) sin 2x tan'^ t/^i sin 2xi tan^ t/j2 sin 2x2 sin 202 sin 20 sin 201
(3.200)
The relation (3.189) for the product of hyperbolic fourcomplex numbers can be demonstrated also by using a representation of the multiplication of the hyperbolic fourcomplex numbers by matrices, in which the hyperbolic fourcomplex number u = x + ay -^ ^z + 'jt is represented by the matrix / X y t \ y X t z A = z t X y \ t z y X I
(3.201)
Hyperbolic complex numbers in four dimensions
83
The product u = x + ay + 0z + ^t of the hyperbolic fourcomplex numbers uj = xi + ayi + I3zi -f 7^1, ^2 = 2:2 + «y2 + 0^2 + ih-) can be represented by the matrix multipHcation A = ^1.42.
(3.202)
It can be checked that the determinant det(A) of the matrix A is detA = u.
(3.203)
The identity (3.189) is then a consequence of the fact the determinant of the product of matrices is equal to the product of the determinants of the factor matrices.
3,2.3
Exponential form of a hyperbolic fourcomplex number
The exponential function of a hypercomplex variable u and the addition theorem for the exponential function have been written in Eqs. (1.35)(1.36). If u = X + ay + /?^ + 7f, then expn can be calculated as expu = expo: • exp(ay) • ex'p{^z) • exp(7t). According to Eqs. (3.160),
where m is a natural number, so that exp(ay), exp(^^) and exp(7i) can be written as exp(ay) = coshy + asinhy, exp(/32;) = coshy + /^sinh^, exp(7^) = cosh t + 7 sinh t.
(3.205)
From Eqs. (3.205) it can be inferred that (cosh t-^r a sinh tY^ = cosh m.t -\- a sinh mt^ (cosh^ H- /3sinht)'^ = coshm^ + /3sinhm^, (cosh t 4- 7 sinh t)'^ = cosh mt + 7 sinh mf,
(3.206)
The hyperbolic fourcomplex numbers u =^ x + ay -\- I5z -{- ^yt ior which s=i x + y + z + t>Q, s' = x-'y-\-z-t>Q, s^^ = x+ y - z ~t > 0, s'^' = X — y — z -{-t> Q can be written in the form x + ay + pz + ^t = e^^+^^^+^'^+^^i.
(3.207)
The conditions s = x+y + z-\-t > 0, 5' = x-y + z — t > 0, s" = x+y — z — t > 0, s'" =^x — y — z + t>0 correspond in Eq. (3.198) to a range of angles 0 < (^ < 7r/2,0 < X < 7r/2,0 < ip < n/2. The expressions of xi,yi,^i,fi as functions of x, y, z^ t can be obtained by developing e^^^, e^^^and e'^^^ with
84
Commutative Complex Nmnhers in Four Dimensions
the aid of Eqs. (3.205), by multiplying these expressions and separating the hypercomplex components, X = e^^(coshj/i cosh^i coshti + sinhyi sinh^i sinhti),
(3.208)
y = e^^ (sinhyi cosh2:1 cosh^i + coshyi sinhzi sinh^^i),
(3.209)
z = e^^(coshyi sinh^i cosht] 4- sinhyi coshzi sinhti),
(3.210)
t = e^^ (sinhyi sinh^ri cosh^i + coshyi cosh2:1 sinhyi).
(3.211)
It can be shown from Eqs. (3.208)-(3.211) that 1 , / / // ///v 1 , -55" 1 , ss' 1 , ssffi XI = -\n{sss s ), y, = - I n - — , zi = - I n - — , t^ = - In .
s's'"' '^
4"'s"s"''''
4"'s's"
(3.212) The exponential form of the hyperbolic fourcomplex number u can be written as /l ss" 1 ss' 1 ss"'\ u = Mexp (^-a In - ^ + -/:(In ^ ^ ^ + - 7 I n — j ,
(3.213)
/i = (^.s'.sV)'/^
(3.214)
where
The exponential form of the hyperbolic fourcomplex number u can be written with the aid of the relations (3.198) as f^ 1 1 . 1.., sin 20 , 1 tanx\ u — Li exp -a In ; h -p In -—^ , . ^—I- - 7 In . V4 tan(/)tanx 4 tan^'0sm2x 4 tan)/ (3.215) The amplitude /i can be expressed in terms of the distance d with the aid of Eqs. (3.198) as fi - dsin^/'^2t/^sin^/^2(/)sin^/^2x.
(3.216)
The hypercomplex number can be written as u = dsin^/2 2i/^sin^/^ 2(/)sin^/'^ 2xexp ( T a i n — \4 tan 0 tan X 1 sin2(/) 1 tanx\ + ~rpln-—'2 , ' r, + 7 7 1 " ; — 7 . 4 tan^^sm2x 4 tan^/ which is the trigonometric form of the hypercomplex number u.
^Q917\
(3.217)
Hyperbolic complex numbers in four dimensions
3.2 A
85
Elementary functions of a hyperbolic fourcomplex variable
The logarithm ui of the hyperboUc fourcomplex number u, ui = Inw, can be defined for s > 0, s' > 0, s" > 0,5'" > 0 as the solution of the equation w = e^S
(3.218)
for ui as a function of u. From Eq. (3.213) it results that 1 1 ss" 1 ss' 1 ss'" Inn = _lnM + + - a l n ^ ^ + ^ ^ I n ^ ^ + 4 7 l n ^ .
(3.219)
Using the expression in Eq. (3.215), the logarithm can be written as , 1, 1 , 1 1 ^, sin 20 lnu= -lnu+-Qln ; + 7^1n—o , : ^ 4 4 tan
1 , tan v +77ln ^. 4 tan
It can be inferred from Eqs. (3.219) and (3.184) that \n{uiU2) = Initi + lnu2The explicit form of Eq. (3.219) is
(3.221)
In(x + ay + l3z + 'yt) = -{l + a + ^ + j) ln(a; + y + z + t) + - ( 1 - a + ^ - 7) ln(a; - y + z - t) + - ( 1 +a-l3-j) + -{l-a-^
ln(a; +
y-z-t)
+ j)ln{x-y-z
+ t).
(3.222)
The relation (3.222) can be written with the aid of Eq. (3.191) as In n = e In s + e' In s' + e" In 5" + e'" In s'".
(3.223)
The power function u^ can be defined for s > 0, s' > 0, s" > 0, s'" > 0 and real values of n as «" = e"'"«.
(3.224)
It can be inferred from Eqs. (3.224) and (3.221) that (W1M2)" = < u^-
(3.225)
86
Commutative Complex Numbers in Four Dimensions
Using the expression (3.222) for Inu and the relations (3.192) and (3.193) it can be shown that ix + ay + Pz + -yt)'' = -{I + a + 0 + 7)(a; + y + z + t)"" + - ( 1 -a + - ( 1 + a+ -{l-a-(3
+ p- 7)(x -y +
z-tf
0- 7)(x + y - 2 - f)" + j){x-y-z
+ tr.
(3.226)
For integer n, the relation (3.226) is valid for any x, y, z, t. The relation (3.226) forn = - 1 is 1 _lfl+a + P + 'y l-a + ZS-j x + ay + (iz-V^t ^\x + y-\-z + t x -y -\- z — t l-ha-/5-7 l-a~B + ^\ + —7 + . 3.227) x + y~z-t X - y - z -ht J The trigonometric functions of the hypercomplex variable u and the addition theorems for these functions have been written in Eqs. (1.57)(1.60). The cosine and sine functions of the hypercomplex variables ay^/3z and 7^ can be expressed as
K^
cos ay = cosy, sin ay ~ asiny,
(3.228)
cos/3y = cosy, sin/3y = psiny^
(3.229)
cos7y = cosy, sin7y = 7siny. (3.230) The cosine and sine functions of a hyperbolic fourcomplex number x-{-ay-{/3z 4- jt can then be expressed in terms of elementary functions with the aid of the addition theorems Eqs. (1.59), (1.60) and of the expressions in Eqs. (3.228)-(3.230). The hyperbolic functions of the hypercomplex variable u and the addition theorems for these functions have been written in Eqs. (1.62)-(1.65). The hyperbolic cosine and sine functions of the hypercomplex variables ay, fiz and ^t can be expressed as cosh ay = coshy, sinhay = asinhy,
(3.231)
cosh/3y = coshy, sinh/?y = /?sinhy,
(3.232)
cosh7y = coshy, sinh7y = 7sinhy.
(3.233)
The hyi)erbolic cosine and sini^ functions of a hyperbolic fourcomplex number x-\-ay-hl3z-]-'yt can then be expressed in terms of elementary functions with the aid of the addition theorems Eqs. (1.64), (1.65) and of the expressions in Eqs. (3.231)-(3.233).
Hyperbolic complex numbers in four dimensions
3.2.5
87
Power series of hyperbolic fourcomplex variables
A hyperbolic fourcomplex series is an infinite sum of the form ao + ai + a2 + • • • + a/ + • • •,
(3.234)
where the coefficients a/ are hyperbolic fourcomplex numbers. The convergence of the series (3.234) can be defined in terms of the convergence of its 4 real components. The convergence of a hyperbolic fourcomplex series can however be studied using hyperbolic fourcomplex variables. The main criterion for absolute convergence remains the comparison theorem, but this requires a number of inequalities which will be discussed further. The modulus of a hyperbolic fourcomplex number u =^ x + ay + ^z + jt can be defined as |u| = (a:2 + y2 + ^2 +^2)1/2^
(3235)
so that according to Eq. (3.182) d = \u\. Since \x\ < |n|,|j/| < \u\,\z\ < I^M^I ^ l^h a property of absolute convergence established via a comparison theorem based on the modulus of the series (3.234) will ensure the absolute convergence of each real component of that series. The modulus of the sum ui +U2 of the hyperbolic fourcomplex numbers ui,U2 fulfils the inequality \\ui\ ~ \U2\\ < \ui + U2\ < \ui\ + \U21
(3.236)
For the product the relation is \uiU2\<2\ui\\u2l
(3.237)
which replaces the relation of equality extant for regular complex numbers. The equality in Eq. (3.237) takes place for x\ = y{ = z\ — t{ and :r2/:ri = y2/yi = ^2/^2^1 = hlh^ In particular h ? | < 2(0:^ + ^ 2 ^ 2 ^ ^ f),
(3.238)
The inequality in Eq. (3.237) implies that |n^|<2^"^|u|^
(3.239)
From Eqs. (3.237) and (3.239) it results that \au^\ <2^\a\\u\K
(3.240)
A power series of the hyperbolic fourcomplex variable u is a series of the form ao + a\u + a2U^ +
f- aiu^ H
.
(3.241)
88
Commutative Complex Numbers in Four Dimensions
Since
J2aiu^\
(3.242)
a sufficient condition for the absolute convergence of this series is that lim ' 7 7 ' ' < 1. /->oo |a/| Thus the series is absolutely convergent for
(3,243 ^ ^
\u\ < Co,
(3.244)
where
The convergence of the series (3.241) can be also studied with the aid of the formula (3.226) which, for integer values of /, is vahd for any x, y, z, t. If ai = aix + aaiy + /3au + lau, and Ai = aix + aiy + aiz + an, A = aij, - aiy -f- au - an,
(3.246) (3.247)
A'l = aij: -f- aiy - au - au,
(3.248)
^T = «/.x - «/t/ - (iiz + a/f,
(3.249)
it can be shown with the aid of relations (3.192) and (3.193) that aie = Aie, aic = AiC , aie = Aie , aie
= Ai e ,
(3.250)
so that the expression of the series (3.241) becomes oo
£
[Ais'e + A[s"e' + A'ls"'e" + A'l's""e!') ,
(3.251)
1=0
where the quantities s,s',s",s'" have been defined in Eq. (3.173). The sufficient conditions for the absolute convergence of the series in Eq. (3.251) are that hm /->oo
,, I A/I
< 1, hm — ~ — < 1, hm — ~ . — < 1, /->oo
\A\\
/->oo
\A'l\
Hyperbolic complex numbers in four dimensions
89
Thus the series in Eq. (3.251) is absolutely convergent for \x -\-y -i- z + t\ < c^ \x — y -h z --t\ < c', \x -{-y — z — t\ < c", Ix-y-z
+ tlK d\
(3.253)
where
\M
,.
,
,.
\A\\
„
,.
\A'l\
,„
,.
\Ar\ (3.254)
The relations (3.253) show that the region of convergence of the series (3.251) is a four-dimensional parallelepiped. It can be shown that CQ = (l/2)min(c, c',c",c'"), where min designates the smallest of the numbers c^d^d\c!^^, Using Eq. (3.197), it can be seen that the circular region of convergence defined in Eqs. (3.244), (3.245) is included in the parallelogram defined in Eqs. (3.253) and (3.254). 3.2.6
A n a l y t i c functions of h y p e r b o l i c fourcomplex variables
The fourcomplex function j[u) of the fourcomplex variable u has been expressed in Eq. (3.117) in terms of the real functions P{x, y, z, t), Q{x, y, z, t), R{x^y^z^t)^S{x^y^z^t) of real variables x,y,z,t. The relations between the partial derivatives of the functions P^Q^R^S are obtained by setting succesively in Eq. (3.118) Ax -> 0, Ay = Az = Ai = 0; then Ay -> 0,Ax = Az = At = 0; then Az -)- 0, Aa; = Ay == At = 0; and finally At -^ 0, Ax = Ay = Az = 0. The relations are
dP dx
dS '' dt'
(3.255)
_dP _ ds dR dy ~'' dz ~ dt'
(3.256)
dR _dS _ dP dQ dx dy '' dz ~ dt'
(3.257)
dx
_ ag _ dR dy
dS _dR _ dQ dx dy ~' dz
dP
df
(3.258)
The relations (3.255)-(3.258) are analogous to the Riemann relations for the real and imaginary components of a complex function. It can be
90
Commutative Complex Numbers in Four Dimensions
shown from Eqs. (3.255)-(3.258) that the component P is a solution of the equations
dz^
dt^ ~
' dx'
df
~
' c^y2
^^2 - 0.
(3.259)
and the components Q, i?, S are solutions of similar equations. As can be seen from Eqs. (3.259), the components P,Q,R,S of an analytic function of hyperbolic fourcomplex variable are solutions of the wave equation with respect to pairs of the variables x, y, z, t. The component P is also a solution of the mixed-derivative equations dxdy
dzdt'
dxdz
dydt'
dxdt
dydz'
and the components Q, /?, S are solutions of similar equations. 3.2.7
Integrals of functions of hyperbolic fourcomplex variables
The singularities of hyperbolic fourcomplex functions arise from terms of the form l/(w — UQ)^^ with m > 0. Fimctions containing such terms are singular not only dX u = UQ, but also at all points of the two-dimensional hyperplanes passing through WQ and which are parallel to the nodal hyperplanes. The integral of a hyperbolic fourcomplex function between two points A^ B along a path situated in a region free of singularities is independent of path, which means that the integral of an analytic function along a loop situated in a region free from singularities is zero, i f{u)du = 0,
(3.261)
where it is supposed that a surface E spanning the closed loop F is not intersected by any of the two-dimensional hyperplanes associated with the singularities of the function / ( n ) . Using the expression, Eq. (3.117), for f{u) and the fact that du = dx + ady + ^dz + ^dt, the explicit form of the integral in Eq. (3.261) is i f{u)du = I [{Pdx + Qdy + Rdz + Sdt) +a{Qdx + Pdy + Sdz + Rdt) + ^[Rdx + Sdy + Pdz + Qdt) -^^[Sdx -f Rdy + Qdz + Pdt)].
(3.262)
Hyperbolic complex numbers in four dimensions
91
If the functions P^Q^R^S are regular on a surface E spanning the loop r , the integral along the loop F can be transformed with the aid of the theorem of Stokes in an integral over the surface E of terms of the form dP/dy-'dQ/dx, dP/dz-dR/dx, dP/dt-dS/dx, dQ/dz-dR/dy, dQ/dtdS/dy^ dR/dt — dS/dz and of similar terms arising from the a , ^ and 7 components, which are equal to zero by Eqs. (3.255)-(3.258), and this proves Eq. (3.261). The exponential form of the hyperbolic fourcomplex numbers, Eq. (3.215), contains no cyclic variable, and therefore the concept of residue is not applicable to the hyperbolic fourcomplex numbers defined in Eqs. (3.160).
3-2.8
Factorization of hyperbolic fourcomplex polynomials
A polynomial of degree m of the hyperbolic fourcomplex variable u = x + ay -\- /3z -\- jt has the form Pm{u) = u ^ + aiu^~^ + • • • + am-iu + am,
(3.263)
where the constants are in general hyperbolic fourcomplex numbers. If ^m = (^mx + OiCLmy + Po.mz + JO'mt') ^nd with the notations of Eqs. (3.173) and (3.246)-(3.249) applied for / = 0,1, • • •, m , the polynomial Pm(^) can be written as Pm = [s"^ + Ais^-'
+ . . . + Am-is + Am] e
4- [V'- + A'{s"^-' + ... + < _ i 5 " -f Al] e" + [5'"^ + A'/'y"—1 4- • • • + < _ i 5 ' ' ' -f A!;:] ^\
(3.264)
Each of the polynomials of degree m with real coefficients in Eq. (3.264) can be written as a product of linear or quadratic factors with real coefficients, or as a product of linear factors which, if imaginary, appear always in complex conjugate pairs. Using the latter form for the simplicity of notations, the polynomial Pm can be written as m,
m
m
m
Pm = U('-si)e+TJ{s'-s[)e'+Yl{s''-s'l)e''+]lis'''-sr)e''', 1=^1
1=1
1=1
(3.265)
1=1
where the quantities si appear always in complex conjugate pairs, and the same is true for the quantities 5J, for the quantitites s['^ and for the
92
Commutative Complex Numbers in Four Dimensions
quantities s[". Due to the properties in Eqs. (3.192) and (3.193), the polynomial Pmiu) can be written as a product of factors of the form m
HY
J•
/=i
These relations can be written with the aid of Eq. (3.194) as
(3.266)
m
(3.267) where Ui
= 5pe + 4e' + 5;V' + 4V".
(3.268)
The roots Sr, :»p,5p,5p,5p' of the corresponding polynomials in Eq. (3.265) may be ordered arbitrarily. This means that Eq. (3.268) gives sets of m roots u\ 5..., u^ of the polynomial P^{VL)^ corresponding to the various ways in which the roots s^p^s^^.s^'^^s^!^ are ordered according to p in each group. Thus, while the hypercomplex components in Eq. (3.264) taken separately have unique factorizations, the polynomial Pm{^) can be written in many different ways as a product of linear factors. If P{u) = w^ — 1, the factorization in Eq. (3.267) is i/^ -- 1 = (u -u\){u — 1x2)1 where u\ = ±e ± e' db e" ± e'", 1/2 = —u\, so that there are 8 distinct factorizations of w^ ~ 1, {u-e-e'-e" - e'") (w + e + e' + e" + e'") e-e'-e" + e'") (u + e + e' + e" - e'") = {uu" = {u-e-e' + e"- e'") {u + e + e' -e" + e'") = {u-e + e'-e"e'") {u + e-e' + e" + e'") (3.269) = {u-e-e' + e" + e'") {u + e + e'-e"e'") = (u-e + e'-e" + e'") (u + e-e' + e"- e'") u = {u-e + e' + e"- e'") (u + e - e' - e" + e'") V?= iu-e + e' + e" + e'") [u + e-e'-e"e'"). It can be checked that {±e ± e' ± e" ± e"'f = e + e' + e" + e'" = 1
3.2.9
Representation of hyperbolic fourcomplex numbers by irreducible matrices
If T is the unitary matrix, / i
T= \
1
1
?
1^
\
2
2
2
Mf
1)
(3.270)
Planar complex numbers in four dimensions
93
it can be shown that the matrix TUT"^ has the form TUT~^ fx+y+z+t 0 0
\
0 x-y+z-t 0
0 0 x+y-z-t
0
0
0
0 0 0
^
x-y-z+tJ (3.271)
where U is the matrix in Eq. (3.201) used to represent the hyperboUc fourcomplex number u. The relations between the variables x + y + z + t^x — y + z — t^x + y — z — t^x — y — z + t for the multiplication of hyperbolic fourcomplex numbers have been written in Eqs. (3.185)-(3.188). The matrix TUT~^ provides an irreducible representation [7] of the hyperbolic fourcomplex number u in terms of matrices with real coefficients.
3.3 3.3.1
Planar complex numbers in four dimensions O p e r a t i o n s w i t h planar fourcomplex n u m b e r s
A planar fourcomplex number is determined by its four components (x, y, z^ t). The sum of the planar fourcomplex numbers (x, y, z^ t) and {x\ y\ z\ t') is the planar fourcomplex number (x + a;', y + y', z-\-z\t + i'). The product of the planar fourcomplex numbers {x, y, z, t) and {x\ y', z\ t') is defined in this section to be the planar fourcomplex number (xx' - yt' ~ zz' - ty\ xy' + yx' - zt' - tz\ xz^ + yy' + zx' - tf, xt' + yz' + zy' 4- tx'). Planar fourcomplex numbers and their operations can be represented by writing the planar fourcomplex number {x, y^z^i) asu = x-^ay + l3z + 7^, where a, p and 7 are bases for which the multiplication rules are a2 = /?,/32 = - l , 72 = - / ? ,
ap = /3a = 7, aj = ja = -I, 0j = -Y^ = - « .
(3.272)
Two planar fourcomplex numbers u — x+ay+^z+yt^ u' = x'+ay'+0z'+jt' are equal, u = n', if and only if x = x',y = y^z = z\t — t'. If w = x+ay-\-Pz-\-^t^ u' — x'-Voty^-\-pz'+^t' are planar fourcomplex numbers, the sum u + u' and the product uu' defined above can be obtained by applying the usual algebraic rules to the sum {x + ay + ^z+^t) + {x* + ay' -\-pz'+^t') and to the product {x + ay -\- ^z + 7t)(x' + ay' + ^z' + 7*'), and grouping of the resulting terms, u + u' ^x-\'x'
+ a{y + y')+ 0{z + z') + ^{t +1'),
(3.273)
94
Commutative Complex Nmnhers in Four Dimensions W = xx' - yt' - zz' - ty -f a{x\j -f yx - zf! - tz') -hftixz' -f yy' 4- zx' - f£) + ^{xt' + yz' + zy' + tx).
(3.274)
If w,ifc',w" are planar fourcomplex numbers, the multiplication is associative (W)w" = u{u'y!')
(3.275)
and commutative W = u'u,
(3.276)
as can be checked through direct calculation. The planar fourcomplex zero i s 0 + a - 0 + /3*0 + 7 - 0 , denoted simply 0, and the planar fourcomplex unity is 1 + a • 0 + /S • 0 + 7 • 0, denoted simply 1. The inverse of the planar fourcomplex number u =:^ x + ay + /3z + jt is a planar fourcomplex number n' = x^ -\- mj -f Pz' -f jt' having the property that W = 1.
(3.277)
Written on components, the condition, Eq. (3.277), is xx^ — ty' — zz' — yt' yx' + xy' - tz' - zt' = zx' + yy' -f xz' - tt' = tx' -f zy' + yz' + :r^' =
— 1, 0, 0, 0.
(3.278)
The system (3.278) has the solution , x(x^ -f- z^) - ^(y^ - t^) + 2OT^ re' = -^ ~ —,
y =
y[x''-z'')^-t{y^-Vt') 5
+ 2xzt
(3.279)
,
(3.280)
J = - 4 ^ ' + ^') + y - ' ' ) + 2 . ! / t ,
,3.281)
,' = - ' ( ^ ' - ^ ' ) + i ' t o H ' ' ) - ^ ^ ! ' ^ P
,3.282,
provided that p ^ 0, where
+2a:^z^ 4- 2y^f + 4.i;''^y< - Axy^z + 4a;^ri2 - ^^^f.
(3.283)
Planar complex numbers in four dimensions
95
The quantity p will be called amplitude of the planar fourcomplex number X + ay + Pz + 7t. Since
(3.284)
P'^PIPI, where
2
P'
=
/
., _ j _ ^ \ 2
( ^ - ^ ) ^ ( - ^ ) '
a planar fourcomplex number u = x + ay + /3z + ^t has an inverse, unless a; + ^
= 0, ^ + ^ = 0 ,
(3.286)
or x-y^=0,z-^=0.
(3.287)
Because of conditions (3.286)-(3.287) these 2-dimensional hypersurfaces will be called nodal hyperplanes. It can be shown that if uu' = 0 then either ti = 0, or u' = 0, or one of the planar fourcomplex numbers is of the form x + a{x + z)/\/2 + /3z — j{x ~ z)/y/2 and the other of the form x' - a{x' + z')/V2 + pz' + ^{x' - z')ly/2. 3.3.2
G e o m e t r i c r e p r e s e n t a t i o n of planar f o u r c o m p l e x numbers
The planar fourcomplex number x-{-ay^-pz-\-^t can be represented by the point A of coordinates {x^y^z^t). If O is the origin of the four-dimensional space x^ y, z^ t^ the distance from A to the origin O can be taken as d^ ^x'^ ^y'^-^z'^
+ f.
(3.288)
The distance d will be called modulus of the planar fourcomplex number a; + ay + /32; + 7*, d — |u|. The orientation in the four-dimensional space of the line OA can be specified with the aid of three angles (/>, Xi "0 defined with respect to the rotated system of axes ^~'V2
2'^""^2
C = - ^ + ! ^ .
2 ''''^y/2
2
' (3.289)
96
Commutative Complex Numbers in Four Dimensions
The variables ^, v, r, ( will be called canonical planar fourcomplex variables. The use of the rotated axes ^, t;, r, ( for the definition of the angles (/>, x, ^ is convenient for the expression of the planar fourcomplex numbers in exponential and trigonometric forms, as it will be discussed further. The angle (f) is the angle between the projection of A in the plane ^, v and the O^ axis, 0 < (f) < 27r, x ^^ ^he angle between the projection of A in the plane T, C and the Or axis, 0 < x < 27r, and if) is the angle between the line OA and the plane TO(^^ 0 < ^ < 7r/2, as shown in Fig. 3.1. The definition of the variables in this section is different from the definition used for the circular fourcomplex numbers, because the definition of the rotated axes in Eq. (3.289) is diflFerent from the definition of the rotated circular axes, Eq. (3.18). The angles (/> and x will be called azimuthal angles, the angle ip will be called planar angle. The fact that 0 < t/^ < 7r/2 means that ip has the same sign on both faces of the two-dimensional hyperplane vO^. The components of the point A in terms of the distance d and the angles ), x? ^ are thus 4= + ^ ^ v2 2 ^
V2 V2
= dcos(/>sini/;,
(3.290)
^ ~ * =dcosxcosV^,
(3.291)
= d sine/) sin t/j,
(3.292)
2
2
~ ^ + ^=rfsinxcost/;.
(3.293)
It can be checked that p^ = \/2ds'mip,p- = \/2dcos^. The coordinates X, y, z^ t in terms of the variables d, <^, x, V' are X = —p(cos0sini/'+ COSXCOS1/'), v2
(3.294)
y = ~p[sin(<;^ + 7r/4)sin'0 + sin(x - 7r/4)cos'V^], v2
(3.295)
;2; = --p (sin 0 sin z/^ — sin x cos ^ ) , v2 t = - - p [ - cos((?!) + 7r/4) sinip + cos(x - 7r/4) cos t/^]. v2
(3.296)
(3.297)
Planar complex numbers in four dimensions
97
The angles (p, x, tp can be expressed in terms of the coordinates x, y,z,t as
. ^
z + {y + t)/V2
sin
1
,
-z + iy + t)/V2
sinx
P-
tanV' =
x+
{y-t)/V2
(3.298)
, cos^ x-{y-
<)/v/2
, cos X =
(3.299) (3.300)
P+IP--
The nodal hyperplanes are (,0v, for which r = 0, C = 0, and TOC, for which ^ = 0, V = 0. For points in the nodal hyperplane ^Ov the planar angle is ij) = 7r/2, for points in the nodal hyperplane TOC, the planar angle is V' = 0. It can be shown that ifui = xi+ayi+0z\-\r'yti,U2 = X2+ay2+0Z2+jt2 are planar fourcomplex numbers of amplitudes and angles pi, <^i, xi»V"! and respectively P2,2,X2,'^2, then the amplitude p and the angles ,Xi'4^ of the product planar fourcomplex number uiU2 are P = Pip2,
(3.301)
(p = 2, X = Xi + X2, tani/» = tam/*! tani/'2.
(3.302)
The relations (3.301)-(3.302) are consequences of the definitions (3.283)(3.285), (3.298)-(3.300) and of the identities (a;ia;2 - 21^2 - J/i*2 - iiy2) +
(Xiy2 + yiX2 - Zit2 - hZ2)
V2
{Xit2 + tiX2 + Ziy2 + yiZ2) {Xiy2 + ^1X2 - 21^2 - tlZ2) + ixiZ2 + Z1X2 + yit/2 - ^1*2) +
V2
+
(Xit2 + tiX2 + Ziy2 + yiZ2y^ ^/2 Xi +
(
21 +
V2
^ y2-t2^^ y2-t2Y
^ (
t2^^ . y2 + ^2^ «2 +
(a;ia;2 - Z122 - yi<2 - *iy2) -
{xiy2 + yiX2 - zit2 - tiZ2) ^/2
(3.303)
98
Commutative
Complex Numbers in Four
Dimensions
{Xit2 + tiX2 + 212/2 + yiZ2)'\^ v/2 ixiZ2 + ziX2 + yiy2 -
+ +
{Xit2 + tiX2 + Ziy2 +
{x\y2+y\X2
ht2)-
v/2
Zit2-tiZ2)
yiZ2y^
v/2 yi - h v/2
(
m-hV^f
ixiX2 -ZxZ2-y\t2
y2 + t2\
(3.304)
-fl2/2)
{xiy2 + y\X2 - z\t2 - t\Z2) - {xit2 + hx2 + Ziy2 + y\Z2) v/2 yi - t \ 1 + '^^-^^j (-^2 + v/2 v/2 yi + t\\ ( , y2 + t2\
+-
(3.305)
{xiZ2 + ZiX2 + J/1 J/2 - tlt2) {Xiy2 + yiX2 -Zit2tiZ2) + ixit2 + tiX2 + Ziy2 + yiZ2) v/2 J/2 -h^ J/2 + ^2 v/2 )
(3.306) •
(rEia;2 -2^1-22-2/1*2 - ^ 1 ^ 2 ) (a:iy2 + y\X2 - ^1*2 - UZ2) - {x\t2 4- hX2 + ^12/2 + 2/12^2)
v/2
y\-h 3^1
-^1 +
7^— v/2 2/1 + * i
v/2
,, 2:2
2/2-*2\ v/2 ; -22 +
2/2 + ^2
(3.307)
v/2
-(rriX2 + 2]X2 + ?;i2/2 - *l*2) (aJl2/2 + yiX2 - 21*2 - *122) + (3^1*2 + t\X2 + 2l2/2 + 2/l22) +v/2
Planar complex numbers in four dimensions
99
(3.308)
The identities (3.303) and (3.304) can also be written as P+--= PI + P2+
(3.309)
pl-
(3.310)
= P1-P2-
where PH = [Xj +
+
V2 J
«? =f.'ru -
( ^ ' - ^ ) ^
1£
^-IZ.
N/2
r
(3.311)
f o r i = 1,2. The fact that the amplitude of the product is equal to the product of the amplitudes, as written in Eq. (3.301), can be demonstrated also by using a representation of the multiplication of the planar fourcomplex numbers by matrices, in which the planar fourcomplex number w = a; 4- ay + /?^ -f 7* is represented by the matrix
A^
{ ^ -t —z
\-y
X
z y
z
-t
X
y
—z -t
X
y
t \ (3.312) )
The product u = x + ay + ^z + ^t oi the planar fourcomplex numbers ui = x\-\- ay I + Pz\ + 7^1, ^2 = ^2 + 0^1/2 + /3^2 + 7^2 ^ can be represented by the matrix multiplication A = A1A2.
(3.313)
It can be checked that the determinant det(A) of the matrix A is detA
(3.314)
The identity (3.301) is then a consequence of the fact the determinant of the product of matrices is equal to the product of the determinants of the factor matrices.
100 3.3-3
Commutative Complex Numbers in Four Dimensions T h e planar fourdimensional c o s e x p o n e n t i a l functions
The exponential function of a hypercomplex variable u and the addition theorem for the exponential function have been written in Eqs. (1.35)(1.36). li u = X + ay + l3z -j- 7^, then expix can be calculated as expn = expx • exp{ay) • exp{0z) • exp(jt). According to Eq. (3.272), ^8m ^ l^^Sm^l
^ ^^c^8m+2 ^ ^^ c^8m+3 ^ ^^
^8m-f4 ^ ^l^c^8m+5 ^ _^c^^ c^8m+6 ^ _ ^ ^ ^^Sm+T ^ ^ ^ ^ ^4m ^ 1^^4m+l ^ ^^ ^4m+2 ^ ^ 1 ^ ^ 4 m + 3 ^ _ ^ ^
(3 3 I 5 )
7^"^ = 1 ^8m+l _. ^ ^8m-f2 = _ / 3 ,^8m+3 _. ^ ^8m-f4 _ _ 2 ^8m4-5 _. _ ^ ^8m+6 __ ^ ^8m+7 __ _ ^
where n is a natural number, so that exp(a?/), exp(/52:) and exp(7f) can be written as expiPz) = cosz + P sin z,
(3.316)
exp(ay) = /4o(y) + c^/41 (y) + 0f42{y) + ihM.
(3.317)
exp(7<) = /4o(^) + 7/41 (i) - /?/42(i) + «/43(*),
(3.318)
and
where the four-dimensional cosexponential functions /40, fnjf42, fis are defined by the series fioix) = 1 - a:V4! -f 2:^8! - - •, /41 (a:) = a: -- a^^S! + x^/9\ Mx)
(3.319) ,
(3.320)
= xy2\ - xVe! + :r^VlO!
,
(3.321)
/43(a;) = xVs! - a:77! + a;^ V H '
•
(3-322)
The functions /4o,/42 are even, the functions /4i,/43 are odd, hoi-u)
= /4o(w), /42("-w) = /42(w), / 4 l ( - w ) = - / 4 l ( t x ) ,
/43(-ti) = ^fMu).
(3.323)
Addition theorems for the four-dimensional cosexponential functions can be obtained from the relation exp a{x 4- y) = exp ax - exp ay, by substituting the expression of the exponentials as given in Eq. (3.317),
Uo{x + y) = ho{x)Uo{y) - /4i(^)/43(y) - /42(^)/42(y) -f43{x)f4i{yh
(3.324)
Planar complex numbers in four dimensions f4i{x + y) = ho{x)f4\{y) + hi{x)My) -/43(x)/42(y),
101 - /42(a;)/43(y) (3.325)
f42{x + y) = /4o(a;)/42(y) + /4i(a;)/4i(y) + /42(a:)/4o(y) -/43(a;)/43(y),
(3.326)
f43ix + y) = /4o(a;)/43(y) + /4i(a;)/42(y) + /42(a;)/4i(y) +Mx)fAQ{y).
(3.327)
For a; = y the relations (3.324)-(3.327) take the form M2x) = /|o(x) - fl(x) - 2Mx)Mx),
(3.328)
M2x)
(3.329)
= 2hQ{x)Ui{x)
/42(2x) = flix)
- flix)
- 2Mx)f43{x), + 2/4o(x)/42(x),
/43(23;) = 2/4o(a;)/43(a;) + 2/4i(x)/42(a;).
(3.330) (3.331)
For x = -y the relations (3.324)-(3.327) and (3.323) yield /4o(^) - fU^) + V4i{x)f4z{x) fliix)
= 1,
(3.332)
- /43(^) - 2/4o(^)/42(a;) = 0.
(3.333)
Prom Eqs. (3.316)-(3.318) it can be shown that, for m integer, (cos ;2: + ^9 sin z)"^ = cos mz + /3 sin mz^
(3.334)
and [/4o(y) + a/4i(y) + I3f42iy) + 7/43(y)]"^ = f4o{my) + a/41 (my) + /S/42(my) + 7/43('Tiy),
(3.335)
[/40(«) + 7/4l(<) - Pf42{t) + 0 / 4 3 ( i ) r = /4o(mf) + 7/41 (mf) - /9/42(mf) + a/43(mf).
(3.336)
Since (a - 7)2"» = 2"", (a - 7)2'"+i = 2'"(a - 7),
(3.337)
it can be shown from the definition of the exponential function, Eq. (1.35) that exp(a - 7)0; = cosh \/2x + ^ ^ = ^ sinh V2x.
(3.338)
102
Commutative Complex Numbers in Four Dimensions
Substituting in the relation exp(a ~ 7)^: = expaxexp{-yx) the expression of the exponentials from Eqs. (3.317), (3.318) and (3.338) yields /40 + /11 + fh + /43 = cosh V2x,
(3.339)
/40/41 - /40/43 + /41/42 + /42/43 = -y^ sinh \/2x,
(3.340)
where /40, /41, /421 /43 are functions of x. From relations (3.339) and (3.340) it can be inferred that fii-fi3\
, (^
. /41 + /43
exp Vza;,
/4I + /43''
v/2
s/2
j =exp(-v/2a;),
(3.341)
(3.342)
which means that
^«^^r^(^«+ h40
/41 - /43
)
v/2 7
/4i+/4.3V
+(/42
V
/ 4 1 + / 43
v/2
= 1.
(3.343)
An equivalent form of Eq. (3.343) is /40 + /41 + /42 + /43 + 2(/4o/42 + /41/43)
+4(/|o/4i/43 + /40/42/I3 - /40/4V42 - /41/I2/43) = 1.
(3.344)
The form of this relation is similar to the expression in Eq. (3.283). Similarly, since (a + -y)2'" ^ (_i)'"2'", (a + 7)2'"+! = (-l)'"2"*(a + 7),
(3.345)
it can be shown from the definition of the exponential function, Eq. (1.35) that /— Q; + 7 /— (3.346) exp(a + 7)x = cos v2x H -7=- sin v2a;. Substituting in the relation exp(a + 7).x = expa.Texp7a; the expression of the exponentials from Eqs. (3.317), (3.318) and (3.346) yields / 4 0 - / 4 i + / 4 2 - / « = ^-osv/2a:,
(3.347)
/40/41 + /40./43 - /41/42 + /42/43 = - ^ sin \/2.x,
(3.348)
Planar complex numbers in four dimensions
103
where /40, /41, /42, IAS are functions of x. Expressions of the four-dimensional cosexponential functions (3.319)(3.322) can be obtained using the fact that [(1 H- i)/\/2]^ = —1, so that
/4o(^) = 2 {^^^^ ~UT^ "^ ^^^ ~JTJ '
(3.349)
/42(a:) = 2^ (cosh —7^^ - cos — ^ a ; j ,
(3.351)
^«^"^ = v / 2 ( - l + i) (^'"^ ^ "
- ''" ^ ^ )
•
^'-''^^
Using the addition theorems for the functions in the right-hand sides of Eqs. (3.349)-(3.352), the expressions of the four-dimensional cosexponential functions become X
X
fioi^) — COS - ^ cosh —p:, v2 v2 /4i {X) = - L (sin ^ 57
(3.353)
cosh ^
+ sinh ^
cos ^ ) ,
X
f42{x) = sin - ^ sinh ~ ^ , /43(.) = ^
(3.354)
(sin ^
(3.355)
cosh ^
- sinh j= cos ^ ) .
(3.356)
It is remarkable that the series in Eqs. (3.353)-(3.356), in which the terms are either of the form x^^^ or a:'*"^"^\ or x'^^'^'^^ or x^'^'^^ can be expressed in terms of elementary functions whose power series are not subject to such restrictions. The graphs of the four-dimensional cosexponential functions are shown in Fig. 3.3. It can be checked that the cosexponential functions are solutions of the fourth-order differential equation d^C whose solutions are of the form ({u) = ^4/40(1*) + Bfii{u) + C/42(u) + DfA:^{u). It can also be checked that the derivatives of the cosexponential functions are related by dfio
r
df4\
~J— = - - / 4 3 , -J-
au
au
.
df^2
-
dJAz
= /40, --7— = / 4 b -J-
au
au
. = /42.
.
.
(J.dDOj
104
Figure
Commutative Complex Numbers in Four Dimensions
3.3:
The
planar
fourdimensional
cosexponential
functions
Planar complex numbers in four dimensions 3.3.4
105
The exponential and trigonometric forms of planar fourcomplex numbers
Any planar fourcomplex number u = x -\~ ay + 0z ^ jt can be writen in the form
(3.359)
x + ay + 0z + jt = e^i+«^i+^^1+^*1
The expressions of xi.yijZi^ti as functions of x, y, 0, t can be obtained by developing e^^Se^^i and e'^*i with the aid of Eqs. (3.316)-(3.318), by multiplying these expressions and separating the hypercomplex components, and then substituting the expressions of the four-dimensional cosexponential functions, Eqs. (3.353)-(3.356), yi + h , yi - ^1 ;=— cosh y/2 V2 . yi + h . 2/1 — sm zi sm 7=— smh v/2 V2
= e^i (
cos zi cos
yi + t\
^xi
y =€
sm
+ cos I Zi +
V2
.
(3.360)
,yi-ti
sinh
N/2
T^\ . yi+h , — sm y^— cosh 4/ v/2
v^ J '
(3.361)
. yi + *i . , y i - h 7=^— smh •V^ v/2
= e^' ficos 2i Sin + sin zi cos
t — = e^xi
yi +
v/2
u yi - * cosh y/2
cos ( z i + I )
-hsm
0'
(3.362)
yi + h . , 2/1 - *i ^ sinh ^ cosh •
v/2
(3.363)
T h e relations (3.360)-(3.363) can be rewritten as XH
2+
a;
;=- = e^^ cos I zi + ^E L ±r-l i ^ Ipe{ y i - t i ) / ^ v^
v/2
- = e ^ ^ s i n-21( +. . - - i ) . ( - -h)/V2 ^ v/2 v/2 y-i ^ = e"'' cos { zi
^ l ± l l ^ p-(yi-t.)/N/2 v/2
(3.364) (3.365) (3.366)
106
Commutative Complex Numbers in Four Dimensions y+t
yi±K\
z—
^-(yi-ti)iV2
(3.367)
By multiplying the sum of the squares of the first two and of the last two relations (3.364)-(3.367) it results that e^^' = PIPI
(3.368)
e^' = p.
(3.369)
or
By summing the squares of all relations (3.364)-(3.367) it results that (f = p2 cosh [v^(y, - < ] ) ] .
(3.370)
Then the quantities y\.,z\.,ti can be expressed in terms of the angles (/>, x, V' defined in Eqs. (3.290)-(3.293) as ^1 +
x/2
(3.371)
= >,
(3.372)
,(yx-li)/V2
v^[coshv^(yi-«i)]
-{.Vl-ll)/V2 1/2
sini/',
yyj = C O S ^ .
v/2[coshv/2(?/i-
From Eq. (3.373) it results that y \ - t \ - -^Intan^,
(3.374)
SO that
(3.375) Substituting the expressions of the quantities xi,yi,zi,t] yields u — pexp _1 ~2
in Eq. (3.359)
^^i=(«-7)l.>ta„,f+i(/J + 2±2'|^
\{^-"-^)X
(3.376)
Planar complex numbers in four dimensions
107
which will be called the exponential form of the planar fourcomplex number u. It can be checked that 1
tt-7
/I
,a-Y\
2^/2•'U•^-^J'^'^^'^ (3.377)
X 2
2
2x/2
V2
2 2x/2 /
cosx (3.378)
2v/2 /
which shows that e[^+('«+T)/v^I<^/2 and e-t^-(°+T)/^lx/2 are periodic functions of (f) and respectively X) with period 2n. The exponential of the logarithmic term in Eq. (3.376) can be expanded with the aid of the relation (3.338) as exp
L2v/2
(a — 7)lntan'
(sin 2^^)1/2 cos
(*-j)
+ ^=»(*-?)
Since according to Eq. (3.300) tan?/? = p-\-/p-, sin V^ cos V' = -75
(3.379)
then
^ ,
(3.380)
pl-hpV and it can be checked that
pl+pl
(3.381)
= 2d\
where d has been defined in Eq. (3.288). Thus (3.382)
p^ = d^ sin 2^", so that the planar fourcomplex number u can be written as u =d
/
7r\
0 - 7 . / ,
I"
exp
K"^^'* (3.383)
which will be called the trigonometric form of the planar fourcomplex number u.
108
Commutative Complex Numbers in Four Dimensions
liui^U2 are planar foiircomplex numbers of moduli and angles di, 0i, Xi i ^pl and respectively d2^2iX2i'^2i the product of the factors depending on the planar angles can be calculated to be ct — y
cos(t/;i - 7r/4) H
y=- sin(^i - 7r/4) v2
cos(^2 - '7r/4) H
Of — 'Y
1
j=r- sin(t/?2 - ^/4) I v2 J
Of — 7
cos(^i - ^^2)
-;;=- cos(t/^i + ip2)
(3.384)
The right-hand side of Eq. (3.384) can be written as Of —
cos(t/^i - ^2)
^
-1=- cos(^i + 1P2)
= [2(cos^ t/j] cos'^ ip2 -f sin^ tp\ sin^ ^2)]*^^[cos(t/? - 7r/4) + ^^-^sin(t/;-7r/4)l,
(3.385)
here the angle ip^ determined by the condition that tan(t/? — 7r/4) = —cos(t/?i + 1P2)/ cos{ipi — '^2)1
(3.386)
is ^2^2? ?which is consistent with Eq. (3.302). IS given given by by tani/j tan %l)= =tant/?i tan ^)\tantan dulus dd of of the the product product u\U2 The modulus i/iti2 is then d = \/2d\d2 (cos^ ^1 cos^ t/^2 + sin"^ ^1 sin^ t/^2)
3.3.5
(3.387)
E l e m e n t a r y functions of planar fourcomplex variables
The logarithm u\ of the planar fourcomplex number ix, u\ = Intx, can be defined as the solution of the ecjuation oUi
(3.388)
written explicitly previously in Eq. (3.359), for ui as a function of ti. From Eq. (3.376) it results that 1 1/ o + 7\ \nu = Inp + r~^(<^ - 7) I n t a n ^ + 2 V "*" " T T " / ^
4(/»-^)x,
(3.389)
Planar complex numbers in four dimensions
109
which is multivalued because of the presence of the terms proportional to (f) and X' It can be inferred from Eqs. (3,301) and (3.302) that l n ( W ) = lnw + lntx',
(3.390)
up to multiples of 7r[/3 + {a + 7)/\/2] and 7t[p -{a + 7)/\/2]. The power function u^ can be defined for real values of n as ^m^^mlnu
(339!)
The power function is multivalued unless n is an integer. For integer n, it can be inferred from Eq. (3.390) that ( W ) ' ^ = i/^t/'^.
(3.392)
If, for example, m = 2, it can be checked with the aid of Eq. (3.383) that Eq. (3.391) gives indeed {x + ay + ^z + 7*)^ = x'^ - z'^ - 2yt + 2a{xy zt) + Piy'^ -t'^ + '^xz) -f 2-f{xt + yz). The trigonometric functions of the hypercomplex variable u and the addition theorems for these functions have been written in Eqs. (1.57)(1.60). The cosine and sine functions of the hypercomplex variables ay.pz and jt can be expressed as cos ay = Uoiy) - ^/42(y), sin ay = a/4i(y) - 7/43(y),
(3.393)
cos/3z = cosh^, sin ^2: = /3sin\iz,
(3.394)
cos7^ = Mt)+/3f42{t),
(3.395)
sin7^ = 7/4i(f) - aMt),
The cosine and sine functions of a planar fourcomplex number x + ay + I3z + ^t can then be expressed in terms of elementary functions with the aid of the addition theorems Eqs. (1.59), (1.60) and of the expressions in Eqs. (3.393)-(3.395). The hyperbolic fimctions of the hypercomplex variable u and the addition theorems for these functions have been written in Eqs. (1.62)-(1.65). The hyperbolic cosine and sine functions of the hypercomplex variables ay, pz and yt can be expressed as cosh ay = My)
+ /3/42(?/), sinhay = a/4i(y) + 7/43(2/),
(3.396)
cosh/Jz = cos^, sinh/3^ = /9sin^,
(3.397)
cosh7t = Mt)
(3.398)
- I3h2{t). sinh7* = 7/4i(t) + ^AsW-
The hyperbolic cosine and sine functions of a planar fourcomplex number x+ay+l3z+^t can then be expressed in terms of elementary functions with the aid of the addition theorems Eqs. (1.64), (1.65) and of the expressions in Eqs. (3.396)-(3.398).
110
3.3.6
Commutative Complex Numbers in Four Dimensions
Power series of planar fourcomplex variables
A planar fourcomplex series is an infinite sum of the form ao + aj -f- a2 -f • • • -f a/ -f- • • •,
(3.399)
where the coefficients a/ are planar fourcomplex numbers. The convergence of the series (3.399) can be defined in terms of the convergence of its 4 real components. The convergence of a planar fourcomplex series can however be studied using planar fourcomplex variables. The main criterion for absolute convergence remains the comparison theorem, but this requires a number of inequalities which will be discussed further. The modulus of a planar fourcomplex number u = x -{- ay -h /3z + jt can be defined as \u\ = {x? + y^ + z^ + t^)^l^,
(3.400)
so that, according to Eq. (3.288), d — \u\. Since |:c| < |?/|,|?/| < hiM-^:! < 1^1,1^1 ^ I^L a property of absolute convergence established via a comparison theorem based on the modulus of the series (3.399) will ensure the absolute convergence of each real component of that series. The modulus of the sum u\ -\- U2 of the planar fourcomplex numbers wi,W2 fulfils the inequality \\ux\ - 1^211 < K + xi2\ < \ui\ -f \U2\.
(3.401)
For the product the relation is \uiU2\ < \/2\ui\\u2l
(3.402)
as can be shown from Eq. (3.387). The relation (3.402) replaces the relation of equality extant for regular complex numbers. The equality in Eq. (3.402) takes place for cos^('0i - 1/^2) = 1, cos'^{xpi -f 7/^2) = 1, which means that XX + ivi ~ fl)/^/2 = 0, zi + (yi + tl)/^/2 = 0, X2 + {y2 - ^2)/\/2 = 0, ^2 + {y2 + t2)/y/2 = 0, or Xl-{y^-tx)/^/2 = 0, ^1 ~(?/i +^i)/\/2 = 0, X2-(y2^2)/\/2 = 0, Z2 - (t/2 + h)l\/^ = 0. The modulus of the product, which has the property that 0 < |?/]t/2|7 becomes equal to zero for cos^(z/^i - ^^2) = 0, cos'^(V^i + 1P2) — 0, which means that x\ + [y\ ~ ti)/\/2 = 0, ^i + (y\ + ti)/v/2 = 0, X2 - {y2 - /2)/\/2 = 0, ^2 - {y2 + ^2)/^/2 = 0, or .xi - {y^ ti)/v/2 = 0, zx-{yi+ti)/V2 = 0, X2+{y2-t2)/y/2 = 0, ^2+(?;2+^2)/^/2 = 0. as discussed after Eq. (3.287). It can be shown that x' + y' + z' + f < \u'\ < y/2{x^ -f ?/ ^ z^-h f).
(3.403)
Planar complex numbers in four dimensions
111
The left relation in Eq. (3.403) becomes an equality for sin^ 2-0 = 1, when p^ = P - , which means that x{y — t) -f z{y + t) = 0. The right relation in Eq. (3.403) becomes an equality for sin^ 2^ = 0, when x -{- {y — t)/\/2 = 0, ^ + (y + t)/V2 = 0, or ar - (y - t)/V2 = 0, z - {y + t)/V2 = 0. The inequality in Eq. (3.402) implies that |^/| <2(^"i)/2|^|^
(3.404)
Prom Eqs. (3.402) and (3.404) it results that \au^\ < 2^/^\a\\4.
(3.405)
A power series of the planar fourcomplex variable w is a series of the form ao + aiu + a2U^ H
+ a/tx' H
.
(3.406)
Since oo
J^aiu^
(3.408)
\ai\
i-^oo
Thus the series is absolutely convergent for \u\ < c,
(3.409)
c = lim J,^^' .
(3.410)
where
The convergence of the series (3.406) can be also studied with the aid of the transformation x + ay + /3z + jt = \/2(ei^ + eiv + e2T + e2C), where ^,t^,T, ( have been defined in Eq. (3.289), and 1
a —7
^
P
a +7
1 a~7
(3.411)
112
Commutative Complex Nmnbers in Four Dimensions
It can be checked that e i = e i , e? = ~ei, e i e i = e i , 62 = 62, ej =-62, 61^2=0, 6162 = 0, 6162 = 0, 6261=0.
6262 = 62, (3.413)
The moduli of the bases in Eq. (3,412) are
and it can be checked that \x'{'ay + /3z + jt\'^ =. ('^ + v^ + r^-i- (^.
(3.415)
The ensemble 61,61,62,62 will be called the canonical planar fourcomplex base, and Eq. (3.411) gives the canonical form of the planar fourcomplex number. If u = u'u", the components ^,t;,r, ( are related, according to Eqs. (3.305)-(3.308) by ^ ^ v / 2 ( ^ r - v'v\
V = ^{i'v"
+ v'^),
r = x/2(rV" - C'C"),
C = v/2(rr' + C r ) ,
(3.416)
which show that, upon multiplication, the components ^,z; and r, ( obey, up to a normalization constant, the same rules as the real and imaginary components of usual, two-dimensional complex numbers. If the coefficients in Eq. (3.406) arc a/ = Q'lo •+- aan + y0a/2 + 7«/3,
(3.417)
Ml = ^/o +
-^—,
and JE—, All = 0.12 + —-j^—, A12 = aiQ
^/2 = -a/2 +
^^—,
(3.418)
the series (3.406) can be written as 00
J22'^^ [(eiAi + e i ^ a ) ( e i ^ + ^wY + [ciAn + e2Ai2)ie2T + eaC)'] • 1=0
(3.419) Thus, the series in Eqs. (3.406) and (3.419) are absolutely convergent for p+<cuP-
(3.420)
Planar complex numbers in four dimensions
113
where ci = lim
5^
r ^ , C2 == lim
r-rr. (3.421)
It can be shown that c = (l/\/2)min(ci,C2), where min designates the smallest of the numbers ci, C2. Using the expression of \u\ in Eq. (3.415), it can be seen that the spherical region of convergence defined in Eqs. (3.409), (3.410) is included in the cylindrical region of convergence defined in Eqs. (3,420) and (3.421).
3.3.7
Analytic functions of planar fourcomplex variables
The fourcomplex function f{u) of the fourcomplex variable u has been expressed in Eq. (3.117) in terms of the real functions P{x, y, z, t), Q{x, y, z, t), R{x^y^z^t)^S{x^y^z^t) of real variables x^y^z^t. The relations between the partial derivatives of the functions P^Q^R^S are obtained by setting succesively in Eq. (3.118) Ax --> 0, Ay = A2; = At = 0; then Ay -^ 0,Ax = Az = At = 0; then Az -^ 0,Ax = Ay = At = 0; and finally At -> 0, Ax = Ay = A^ = 0. The relations are dP _ dQ _dR _ dS dx dy dz dt '
(3.422)
dR _dS _ dy dz
(3.423)
dx
OR _ dS _ dx dy dS _ dx
dP _ dy
dP _ dz dQ _ dz
dP dt ' dQ dt ' dR dt
(3.424) (3.425)
The relations (3.422)-(3.425) are analogous to the Riemann relations for the real and imaginary components of a complex function. It can be shown from Eqs. (3.422)-(3.425) that the component P is a solution of the equations Q2p
Q2p
Q2p
Q2p
and the components Q, R, S are solutions of similar equations. As can be seen from Eqs. (3.426), the components P^Q^R^S of an analytic function
114
Commutative Complex Numbers in Four Dimensions
of planar fourcomplex variable are harmonic with respect to the pairs of variables x^ y and z, t. The component P is also a solution of the mixedderivative equations
d^_^d^
d ^ _ a ^
d^_d^
d'^p _
&^p
and the components Q, R, S are solutions of similar equations. The component P is also a solution of the mixed-derivative equations ^P___d^ dxdy dzdt'
d'^P _ d^P dxdt dydz'
(3.428)
and the components Q, R, S are solutions of similar equations. 3.3.8
Integrals of functions of planar fourcomplex variables
The singularities of planar fourcomplex functions arise from terms of the form l/(u—uo)^, with m > 0. Functions containing such terms are singular not only at w = WQ? but also at all points of the two-dimensional hyperplanes passing through UQ and which are parallel to the nodal hyperplanes. The integral of a planar fourcomplex function between two points A, B along a path situated in a region free of singularities is independent of path, which means that (f f{u)du = 0,
(3.429)
where it is supposed that a surface E spanning the closed loop F is not intersected by any of the two-dimensional hyperplanes associated with the singularities of the function f{u). Using the expression, Eq. (3.117) for f{u) and the fact that dii — dx + ady + /3dz -f jdt, the explicit form of the integral in Eq. (3.429) is
(3.430)
If the functions P, Q, R, S are regular on a surface E spanning the loop F, the integral along the loop F can be transformed with the aid of the theorem of Stokes in an integral over the surface E of terms of the form dP/dy+dS/dx, dP/dz+dR/dx, dP/dt-hdQ/dx, OR/dy-dS/dz, dS/dtdQ/dy, dR/dt - dQ/dz and of similar terms arising from the a, f3 and 7
Planar complex numbers in four dimensions
115
components, which are equal to zero by Eqs. (3.422)-(3.425), and this proves Eq. (3.429). The integral of the function {u — UQ)^ on a closed loop F is equal to zero for m a positive or negative integer not equal to -1,
i
{u - UQ)'^du = 0, m integer, m 7^ - 1 .
(3.431)
This is due to the fact that J{n — UQ)^du = (w — uo)'^^^/{m + 1), and to the fact that the function {u — I/Q)^"^^ is singlevalued for m an integer. The integral / duj(u — WQ) can be calculated using the exponential form (3.376), 1/ — UQ = pexp l ^ ( a - 7 ) l n . a „ * + 2 ( / 3 - h - ^ j *
4(.-^)x],
(3.43,
SO that du
dp
1 /
X ,,
/
1 /^
c^ + 7 \ . /
Since p and -0 are singlevalued variables, it follows that j^dp/p = 0, ^ d l n t a n i / ? = 0. On the other hand, 0 and x ^^^ cyclic variables, so that they may give a contribution to the integral around the closed loop r . Thus, if C-i- is a circle of radius r parallel to the ^Ov plane, whose projection of the center of this circle on the ^Ov plane coincides with the projection of the point no on this plane, the points of the circle Cf. are described according to Eqs. (3.289)-(3.293) by the equations C = ^0 + ^ sin ip cos 0, u = t;o + r sin ^ sin >, r = TQ + r cos if) cos x, C = Co + ^ cos V? sin X,
(3.434)
for constant values of x and tp, ip ^ 0, IT/2, where HQ = XQ + ayo-\-pzo-}-jto^ and (^0^ ^0^ '^0^ Co are calculated from XQ, yo, ZQ, to according to Eqs. (3.289). Then
If C_ is a circle of radius r parallel to the TO( plane, whose projection of the center of this circle on the TO( plane coincides with the projection of
116
Commutative Complex Nmnbers in Four Dimensions
the point WQ on this plane, the points of the circle C_ are described by the same Eqs. (3.434) but for constant values of (f) and i/j, ip ^ 0,7r/2. Then
The expression of §Y^du/(u — UQ) can be written as a single equation with the aid of the functional int(M, C) defined in Eq. (3.134) as du
-TT ( / ?
(3.437)
-j^\ ini{uQrQJrd^
where wo?t;,wor< and r^^;,r^^ are respectively the projections of the point ixo and of the loop F on the planes (^v and r^. If f{u) is an analytic planar fourcomplex function which can be expanded in a series as written in Eq. (1.89), and the expansion holds on the curve r and on a surface spanning F, then from Eqs. (3.431) and (3.437) it follows that / f{u)du
r/^
^^ + 7 \ . w
( / 3 - ^ )
T. X
int(uo.orrc)] / ( ^ o ) ,
(3.438)
where F^^,Fr(; are the projections of the curve F on the planes (^v and respectively r(^, as shown in Fig. 3.2. As remarked previously, the definition of the variables in this section is different from the former definition for the circular hypercomplex numbers. Substituting in the right-hand side of Eq. (3.438) the expression of f{u) in terms of the real components P, Q, /?, 5, Eq. (3.117), yields
i
f{u)du
r w - Wo =
TT
+7 ( " 7-
-IT
7-
1+
V2
Q-[a
v/2 a +7
v/2
v/2 rv — 7
P
1 + /^ Qv/2
« —7
«
1 v/2
R int (MO^,„ r^„)
R mt(MorC-rTc)' (3.439)
Planar complex numbers in four dimensions
117
where P,Q,R,S are the values of the components of / at n = wolf f{u) can be expanded as written in Eq. (1.89) on T and on a surface spanning F, then from Eqs. (3.431) and (3.437) it also results that /
f(u)du - ( / J - 2 ^ )
IT r /
a + 7 \ . ,,
i n t K , { , r , t ) ] /<""(«o),
T^ , (3.440)
where it has been used the fact that the derivative f^^^uo) of order n of f{u) at u = UQ is related to the expansion coefficient in Eq. (1.89) according to Eq. (1.93). If a function f{u) is expanded in positive and negative powers oiu — Uj^ where Uj are planar fourcomplex constants, j being an index, the integral of / on a closed loop F is determined by the terms in the expansion of / which are of the form aj/{u — ^j), / ( « ) = --- + E - U ^—^Uj + ----
(3-441)
Then the integral of / on a closed loop F is
-vr (^/3 - ^ ^ ^ Y. int(u,rc, rrc)«j.
3.3.9
(3.442)
F a c t o r i z a t i o n of planar f o u r c o m p l e x p o l y n o m i a l s
A polynomial of degree m of the planar fourcomplex variable n = x 4- ay -hPz + yt has the form Pm{u) ^u"^ + aiu"^-^ + • • • + am-iu + am,
(3.443)
where the constants are in general planar fourcomplex numbers. It can be shown that any planar fourcomplex polynomial has a planar fourcomplex root, whence it follows that a polynomial of degree m can be written as a product of m linear factors of the form u — Uj^ where the planar fourcomplex numbers iij are the roots of the polynomials, although the factorization may not be unique, m
Pm{u) = l[{u-Uj).
(3.444)
118
Commutative
Complex Nmnbers
in Four
Dimensions
The fact that any planar fourconiplex polynomial has a root can be shown by considering the transformation of a fourdimensional sphere with the center at the origin by the function u^'^^ The points of the hypersphere of radius d are of the form written in Eq. (3.383), with d constant and 0, X, V^ arbitrary. The point u^ is
ir = d^ (3.445)
exp It can be shown with the aid of Eq. (3.387) that uexp
n
{^^"T^yw-^-wh] = hi
[2
(3.446)
so t h a t cos('0 ~ 7r/4) + exp
Of — 7
._ {ip - IT/4) sin , 1 /„
iw^"-
a + 7
±2
v/2
cos('0 - 7r/4) H
rt — 'y j=- sin(V' - 7r/4)
(3.447)
T h e right-hand side of Eq. (3.447) is (^eos.4--^s.n.j
| =
g Q2k
,,^^2m-2k
'
e sm
A:=0
Ik
e
(3.448)
where e = ip - 7r/4, and since Cf^ ^ C^*^!! i^' <^'^n '^^ concluded that cos 6 H
a —7 . 7=- sni 6
|2
v^
(3.449)
> 1.
Then (3.450)
(T < Iw'"! < 2("'-i>/2d"',
which shows that the image of a four-dimensional sphere via the transformation operated by the function w"' is a finite hypersurface. If u' = i i ^ , and u' = d' cos(V'' - 7r/4) + '-^—^ sin(V'' -
V2
exp
n/4)
cv + 7 2 ^ ^
v/2
X
(3.451)
Planar complex numbers in four dimensions
119
then 0' = m(f), x' = ^ X . tan'0' = tan"^ xp.
(3.452)
Since for any values of the angles (/>', x'^ ^ ' there is a set of solutions 0, x, ^ of Eqs. (3.452), and since the image of the hypersphere is a finite hypersurface, it follows that the image of the four-dimensional sphere via the function u^ is also a closed hypersurface. A continuous hypersurface is called closed when any ray issued from the origin intersects that surface at least once in the finite part of the space. A transformation of the four-dimensional space by the polynomial Pm{u) will be considered further. By this transformation, a hypersphere of radius d having the center at the origin is changed into a certain finite closed surface, as discussed previously. The transformation of the four-dimensional space by the polynomial Pm{'^) associates to the point u = 0 the point /(O) = ami and the image of a hypersphere of very large radius d can be represented with good approximation by the image of that hypersphere by the function u^. The origin of the axes is an inner point of the latter image. If the radius of the hypersphere is now reduced continuously from the initial very large values to zero, the image hypersphere encloses initially the origin, but the image shrinks to CLUI when the radius approaches the value zero. Thus, the origin is initially inside the image hypersurface, and it lies outside the image hypersurface when the radius of the hypersphere tends to zero. Then since the image hypersurface is closed, the image surface must intersect at some stage the origin of the axes, which means that there is a point ui such that f{ui) = 0. The factorization in Eq. (3.444) can then be obtained by iterations. The roots of the polynomial Pm can be obtained by the following method. If the constants in Eq. (3.443) are a/ = ciio-haaii'h/3ai2-{-^ais^ and with the notations of Eq. (3.418), the polynomial Pm{y') can be written as m
Pm = X^2(—^)/2(ei^a + e i i / i ) ( e i ^ + eii;)—' m
+ E 2^"-^)/'(e2A,2 + e2i/2)(e2T + e2C)"-^
(3.453)
where the constants Aik^Au^^k = 1,2 are real numbers. Each of the polynomials of degree m in ei^ -f- eiv,e2T -h e2C in Eq. (3.453) can always be written as a product of linear factors of the form ei(^ — ^p) -^ ei{v — Vp) and respectively e2(r — Tp) + e2(C - Cp)? where the constants (.p^Vp^Tp^^p
120
Commutative Complex Nmnhers in Four Dimensions
are real, 771
771
n 2"^/' {ei(^ - ^p) + e.(t; - »^p)} ,
(3.454)
m
= n 2 " ^ ' (^2(r - Tp) + e2(C - Cp)} .
(3.455)
p=i
Due to the relations (3.413), the polynomial Pm{u) can be written as a product of factors of the form m
Pmiu) = n 2^^' {^i(^ - ^p) + ^i(^ - ''p) + ^2(r - T^) + e2(C ~ Cp)} . p=i
(3.456) This relation can be written with the aid of Eq. (3.411) in the form (3.444), where Up = \/2(ei^p + eiVp + e2Tp -f e2Cp).
(3.457)
The roots e\^p+e\i)p and e2Tp-\-e2C,p defined in Eqs. (3.454) and respectively (3.455) may be ordered arbitrarily. This means that Eq. (3.457) gives sets of m roots u\,..., Um of the polynomial Pmiu), corresponding to the various ways in which the roots e\^p-\-e\Vp and e2Tp-\-e2Cp ar^ ordered according to p for each polynomial. Thus, while the hypercomplex components in Eqs. (3.454), (3.455) taken separately have unique factorizations, the polynomial Pm[u) can be written in many different ways as a product of linear factors. The result of the planar fourcomplex integration, Eq. (3.442), is however unique. If, for example, P{u) = li^ 4-1, the possible factorizations are P — {u ei-62){u+61+62) and P = {11-ei+e2)(u+ei -62)^ which can be written as u'^-^l = (w-/3)(w+/3)orasw^ + l = { i / - (a + 7)/\/2} {(w-h (a + 7 ) / \ / 2 } The result of the planar fourcomplex integration, Eq. (3.442), is however unique. It can be checked that (±ei ± 62)^ = —e\ — 62 = —l>
Polar complex numbers in four dimensions 3.3.10
121
R e p r e s e n t a t i o n of planar f o u r c o m p l e x n u m b e r s b y irreducible m a t r i c e s
If T is the unitary matrix, / _L 0 72 1
1 i
0 J_
-i 1
2 ~ 21
v^ 0n
2 21
\ (3.458)
it can be shown that the matrix TUT~^ has the form 1-1
TUT-'
=
'
,/
h
(3-459)
where U is the matrix in Eq. (3.312) used to represent the planar fourcomplex number u. In Eq. (3.459), Vi, V2 are the matrices Vi = (3.460) In Eq. (3.459), the symbols 0 denote the matrix (3.461) The relations between the variables x -\~ {y — t)/y/2^z 4- (y 4- t)/y/2, x — {y — t)/\/2, —2: + (y 4-1)/\/2 for the multiplication of planar fourcomplex numbers have been written in Eqs. (3.305)-(3.308). The matrix TUT'^ provides an irreducible representation [7] of the planar fourcomplex number u in terms of matrices with real coefficients.
3.4
Polar complex numbers in four dimensions
3.4.1
Operations with polar fourcomplex numbers
A polar fourcomplex number is determined by its four components (x, y, z, t). The sum of the polar fourcomplex numbers {x^y^z^t) and [x^^y\z\t') is the polar fourcomplex number {x + a;',y + y',z + z^^t +1'). The product of the polar fourcomplex numbers (x, y, z^ t) and {x\ y', ^', <') is defined in this section to be the polar fourcomplex number (a;a;'-l-y]t'4-2:^' + fy',xj/' + ya;' + zt^ + tz',xz'^-yy' + zx' + tt*,xt' + yz' + zy' + tx^). Polar fourcomplex numbers
122
Commutative Complex Numbers in Four Dimensions
and their operations can be represented by writing the polar fourcomplex number (x, y, z, t) as u = x -i- ay -{- pz -{- jt, where a, p and 7 are bases for which the muItipHcation rules are a^ = /?, ^^ = 1, 7^ — p,a/3 = pa = 7, aj = ja = ~ 1 , /37 = 7/5 = a. (3.462) Two polar fourcomplex numbers u = x+ay+^z-^'yt.u' = x'+ay'+^z'+jt^ are equal, u = tx', if and only ii x = x\y = y\z ~ z\i = t'. \{ u — x + ay+/3z+jt^u' = x'-\-ay'-\-Pz' + ^t^ are polar fourcomplex numbers, the sum u-\-u' and the product uu^ defined above can be obtained by applying the usual algebraic rules to the sum {x-{-ay-^pz-]-'^t)-\-{x'-\-ay* + pz^-V^t') and to the product {x -{• ay -\- pz + ^t){x' -f- ay' -f ^z' -h 7^'), and grouping of the resulting terms, u-^v! = x + x' ^a{y^
y') + p{z -f z') + ^{t 4-1'),
(3.463)
uu' = a:j:' + yt' + zz' + ty' + Qf(a;?/' + ya:' + zt' -f t^') +^(x^' + yy' + ^a;' + tt') + 7(xt' + y / + zy' + ix').
(3.464)
If tz, t/', w" are polar fourcomplex numbers, the multiplication is associative {uu')u" = u{u'u")
(3.465)
and commutative uu = uu,
(3.466)
as can be checked through direct calculation. The polar fourcomplex zero is0 + a - 0 + /3-0-f7-0, denoted simply 0, and the polar fourcomplex unity i s l 4 - a - 0 + / 3 - 0 - f 7 - 0 , denoted simply 1. The inverse of the polar fourcomplex number u = x + ay + j3z + jt is a polar fourcomplex number u' = x' + ay' + pz' + 'yt' having the property that uu' = 1.
(3.467)
Written on components, the condition, Eq. (3.467), is xx' yx' zx' tx'
-^ty' + xy' + yy' + zy'
+ zz' + yf' + zt' + tz' + -f tt' -I- xz' -^ + yz' -f xt'
= 1, = 0, = 0, = = 0.
.^ ^gg^
Polar complex numbers in four dimensions
123
The system (3.468) has the solution
^' = -z{x''-z-')i-x{y^
+ f)-2yzt^
_L z'^) ^ 2 \ -„ yjy'^ t 2 -_ +2^ ^, ^ -tix"^,.2 + ^^) + 2xyz
^^ ^^^^
(3.472)
provided that v ^ 0, where 4 . 4
4
«4
--2x'^z^ + 2y^t^ - 4a;^y^ - 4yzh + ixy'^z + Axzf,
(3.473)
The quantity u can be written as u=^v^v-fi\,
(3.474)
where Vj^-x^-y-^-z
+ t, v^=x-y-\-z-t,
(3.475)
and f^l = {x-zf-h{y-t)'^.
(3.476)
Then a polar fourcomplex number q = x + ay •^' ^z -h jt has an inverse, unless v^ = 0, or t;_ = 0, or jj.^ = 0.
(3.477)
The condition v^ =0 represents the 3-dimensional hyperplane x+y-\-z-{-t = 0, the condition v^ =0 represents the 3-dimensional hyperplane x — y + z — i = 0, and the condition /i^ = 0 represents the 2-dimensional hyperplane X — z^y = t. For arbitrary values of the variables x^y,z^i, the quantity u can be positive or negative. If i/ > 0, the quantity p = u^l"^ will be called amplitude of the polar fourcomplex number x + ay + ^z + jt. Because of conditions (3.477), these hyperplanes will be called nodal hyperplanes. It can be shown that if uu' = 0 then either ti = 0, or n' = 0, or the polar fourcomplex numbers n, u' belong to different members of the pairs of orthogonal hypersurfaces listed further, x + y^z-ht
= 0 amd x' = y' = z' = t\
(3.478)
124
Commutative x-y
Complex Numbers
in Four
Dimensions
+ z - t = 0 and x' = -y' = z' = -t\
(3.479)
Divisors of zero also exist if the polar fourcomplex numbers u, ?/' belong to different members of the pair of two-dimensional hypersurfaces, X-
3.4.2
z = Q,y-t^()
and x -V z' = 0, y' -\-t' = 0.
(3.480)
Geometric representation of polar fourcomplex numbers
T h e polar fourcomplex number x + ay + /3z + jt can be represented by the point A of coordinates {x, y, z, t). If O is the origin of the four-dimensional space X, y, z, f, the distance from A to the origin O can be taken as S = :i^ + y^j^z^ ^f.
(3.481)
T h e distance d will be called modulus of the polar fourcomplex number X -j- ay -i- /3z -{-jt, d = \u\. If w = x+ay+pz+jt,ui = xi+ayi+pZi+jti,U2 = :r2+a2/2+/3z2+7^2, and u = iiiU2^ and if Sj^ = Xj + yj -f Zj -f- tj, Sj^ = Xj — yj -f- Zj — tj,
(3.482)
for j = 1,2, it can be shown that V^ = 5i-f.92-f, V^ = Si^S2-.
(3.483)
T h e relations (3.483) are a consequence of the identities (a;ia;2 + ^1^2 + ^1^2 + yi^2) + (^1^2 + yiX2 + ^1^2 -f- ^1^2) +(.^1^2 + '2^i-'?^2 + 2/12/2 + ^1^2) + (^1^2 -f ^1^2 + ^12/2 + 2/1^2) = (^1 + 2/1 + ^1 + ^i)(^2 + 2/2 + ^2 + ^2),
(3.484)
(a:i:r2 + ^1^:2 + ifi2/2 + 2/1^2) - (^i2/2 + 2/1^2 + ^1^2 + ^1^2) +{xiZ2 + Z]X2 + 2/12/2 + ^1^2) - (^ik = {xi +zx-
+ U^2 + ^12/2 + 2/1^2)
2/1 - tx){x2 + Z2- 2/2 - ^2).
(3.485)
T h e differences v\ ^ X — z^ v\ = y — t can be written with the aid of the radius //4., Eq. azimuthal angle 0, where 0 < > < 27r, as vi = n^ cos(/>, vi = /i^. s i n 0 .
(3.486) (3.476), and of the
(3.487)
Polar complex numbers in four dimensions
125
The variables v-^^V-^vi^v\ will be called canonical polar fourcomplex variables. The distance d, Eq. (3.481), can be written as d' = \<^r-
+ \t^l-
(3-488)
It can be shown that if ui = xi + ayi + Pzi + jti^U2 = X2 + aj/2 + P^2 + 7*2 are polar fourcomplex numbers of polar radii and angles pi-,<^i and respectively p2-7 ^2, then the polar radius p and the angle (j) of the product polar fourcomplex number uiU2 are /x+ = pi_P2-,
(3.489)
0 = 0i+(/)2.
(3.490)
The relation (3.489) is a consequence of the identity [(a:iX2 + ZiZ2 + yit2 -f tiy2) - {xiZ2 + 2:10:2 + yiy2 +
tlt2)f
+ [{Xiy2 + yiX2 + Zit2 + ^12^2) - {Xit2 + tiX2 + Ziy2 + yiZ2)]^
= [{xi - zif
+ (yi -- ti)^] [{X2 - Z2f + (2/2 - ^2)'] ,
(3.491)
and the relation (3.490) is a consequence of the identities (:ria:2 + Z1Z2 + yit2 + tiy2) - [xiZ2 + zxX2 + y\y2 + ^1*2) = (0:1 - zi){x2 - Z2) - (2/1 - ix){y2 - ^2),
(3.492)
[xiy2 + y\X2 -H zxi2 + ^1^2) - (^1^2 + i\X2 + zxy2 4- 2/1^2) = (yi - t\){x2 - Z2) + (xi - zx){y2 - t2).
(3.493)
A consequence of Eqs. (3.483) and (3.489) is that if n = U1U2, and i/j = SjSjPj- , where j = 1,2, then ly = viiJ2^
(3.494)
The angles 6^^ 6^ between the line OA and the v^ and respectively Vaxes are tan04. = ^ ^ , t a n e _ = ^ ^ ,
(3.495)
where 0 < ^.f < TT, 0 < ^_ < TT. The variable p^ can be expressed with the aid of Eq. (3.488) as
126
Commutative Complex Numbers in Four Dimensions
The coordinates x^ y, z^ t can then be expressed in terms of the distance d, of the polar angles 6^^ 9^ and of the azimuthal angle (j) as //4.(tan^-f + tan^_)
I
/x-j_(-tan^++ tan0_)
,
,
. ,
/., .^^x
//_i_(—tan^-f + tan0_) 1 . , t = ^-^^-^= ± ^ - -/i+ sin(/). 2v^tan(9+tan0_ 2'^ ^
/^ ^^^x 3.500
2v2tan^4-tan^_
1
,
2
If t/ = uiU2, then Eqs. (3.483) and (3.489) imply that tan^_f- = -yr tan^i-^ tan^2+7 tan^_ = —= tan ^i_ tan 02-7 v2 v2
(3.501)
tan ei+ = ^ ^ ^ , tan ^,_ = ^^^^.
(3.502)
where
An alternative choice of the angular variables is //4. = \/2dcos0^ v^ — 2dsin0cosA, V- = 2(isin0sinA,
(3.503)
where 0 <0 < 7r/2,0 < A < 27r. If u = uiU2, then tan A = tan Ai tan A2, dcos 0 = y/2d\d2 cos 6\ cos 62^
(3.504)
where pj_ = \/2dj cos9j, Sj = 2dj sinOj cos Aj, 5^' = 2dj s'mOj sin Aj, (3.505) for j = 1,2. The coordinates x, j/, z, t can then be expressed in terms of the distance d, of the polar angles 9^X and of the azimuthal angle (f) as x= -y=sin6>sin(A4-7r/4)+ -pcos6^cos0,
y = —7= sin 0 cos(A 4- n/4) + ~j=. cos 9 sin 0, v2 v2 2: = —P sin 0 sin(A -f 7r/4) v2
^ cos 0 cos >, v2
(3.506)
(3.507) (3.508)
Polar complex numbers in four dimensions t = -7= sin 0 cos(A -f 7r/4) \/2
^ cos 0 sin (/>. v2
127 (3.509)
The polar fourcomplex numbers 6+ =
l + a + /3 + 7 l ~ a + /3-7 ^ , e_ = —, ,
(3.510)
have the property that e\ = e+, e?_ = e_, e+e_ = 0.
(3.511)
The polar fourcomplex numbers 1 -/? ^
a-J
ei = - y - ^ ^ i = - y - -
(3.512)
have the property that ef = ei, e^ = —ei, eiei = ei.
(3.513)
The polar fourcomplex numbers e-|.,e_ are orthogonal to ei,ei, 64. ei = 0, 64- ei = 0, e_ ei = 0, e_ ei = 0. (3.514) The polar fourcomplex number q = x + ay + I3z + ^t can then be written as X -\- ay + ^z + ^t = v^e^ -f v-e^ 4- f iCi + 'Oici.
(3.515)
The ensemble e^,e^,ei,ei will be called the canonical polar fourcomplex base, and Eq. (3.515) gives the canonical form of the polar fourcomplex number. Thus, the product of the polar fourcomplex numbers n, u' can be expressed as uu' = v^v'_^e^ 4- v-v'_e- -f {v\v[ - v\v\)e\ -\- {v\v'i -h v[v\)e\, (3.516) where 1;^ = x' + y' + ;2' -f t', v'_ = x' - y' + 2:' - t', v[ = x' - y', -Di = ^' - t'. The moduli of the bases used in Eq. (3.515) are K | = i , |e_| = i , h | = i = , |ei| = - ^ .
(3.517)
The fact that the amplitude of the product is equal to the product of the amplitudes, as written in Eq. (3.494), can be demonstrated also by using a representation of the multiplication of the polar fourcomplex numbers by
128
Commutative Complex Numbers in Four Dimensions
matrices, in which the polar fourcomplex number u = x + ay + ^z -h 'yt is represented by the matrix
A
/ X y t X z t \y z
z t \ y z X y t XJ
(3.518)
The product u — x + ay + ^z -\- jt of the polar fourcomplex numbers ui = xi + ayi -f /3zi -f 7^1, W2 = X2 4- ay2 -f /3^2 + 7^2? can be represented by the matrix multiplication A = AiA2.
(3.519)
It can be checked that the determinant det(74) of the matrix A is det^ = iy.
(3.520)
The identity (3.494) is then a consequence of the fact the determinant of the product of matrices is equal to the product of the determinants of the factor matrices. 3.4.3
T h e polar fourdimensional c o s e x p o n e n t i a l functions
The exponential function of a hypercomplex variable u and the addition theorem for the exponential function have been written in Eqs. (1.35)(1.36). If u = X -h ay + /3z + 7^, then expu can be calculated as expu = exp.T • exp(ay) • exp{/3z) • exp(7f). According to Eq. (3.462), /32m ^ l^/32m-^i ^p^
(3 521)
^4m _. J ^4m-f-l — ^ ^4m-|-2 _. Q ^4m+3 _. ^
where m is a natural number, so that exp{ay), exp(pz) and exp(72:) can be written as expipz) = cosh z + P sinh z,
(3.522)
and exp(ay) = g^oiv) + ot94i iy) + P9i2{y) + 7543(?/), exp(7i) = g4o{t) + 754i(0 + ft942{f) + agMf).
(3.523) (3.524)
where the four-dimensional cosexponential functions 940, .941,942.943 are defined by the series g^Q{x) = 1 + x^/A\ + j;V8! + • • •,
(3.525)
Polar complex
numbers
g4i ix)=x-\-
in four dimensions
129
x^/5\ + x^/9\ + • • •,
942{x) = x^l2\ ^ x^l%\ + x^^im
(3.526)
-}-•••,
(3.527)
ff43(^) = ^V3! + ^77! + ar^Vll! + • • •.
(3.528)
T h e functions 540 5 5^42 are even, the functions 5^41,5^43 are odd, 54o(-t/) = gm{u), 542(-w) = ff42(w), gwi-u)
= -941 (w),
i^43(--t^) = -ff43(w).
(3.529)
It can be seen from Eqs. (3.525)-(3.528) that 940 + 941 + 942 + 943 = e^, 940 - 941 + 942 - 943 = e""^,
(3.530)
940 - 942 = cos a:, 941 - 943 = s i n x ,
(3.531)
and
so t h a t (940 + 941 + 942 + 943)(940 - 941 + 942 " 943) [(940 - 942)^ + (941 - 943)^] = 1,
(3.532)
which can be also written as 4 o - ^41 + ^42 - ^43 - 2(9|o942 ^ 54lff43) -4(940941943 + 942541943 "" 94l940942 - 943540942) = l-
(3.533)
T h e combination of terms in Eq. (3.533) is similar to that in Eq. (3.473). Addition theorems for the four-dimensional cosexponential functions can be obtained from the relation expQf(rr -i- y) = expax • e x p a y , by substituting the expression of the exponentials as given in Eq. (3.523), 940 (^ + y) = 94o(a:)94o(y) + 9ii{x)g43{y)
+ 942(^)942(y)
+943(^)941 (y), 94l(^ + y)=
(3.534)
9i0{00)9Al{y) + 94l(^)940(y) + 942(^)943(y)
+943(^)942(y),
(3.535)
942(^ + y) = 940(^)942(y) + 94l(^)94l(y) + 942(^)940(y) +943(^)943(y),
(3.536)
130
Commutative Complex Numbers in Four Dimensions gAz{x -f y) = y4o(^).943(y) + 9\\{x)g^2{y) -I- g'\2{x)gA\ (y) +543(x)^4o(y).
(3.537)
For x = y the relations (3.534)-(3.537) take the form 34o(2x) = gjoix) + gj^ix) -f 2g4i{x)g43{x),
(3.538)
g4i{2x) = 2^4o(.^)ff4i(^) + 2gi2{x)gis{x),
(3.539)
942(20:) = glix)
(3.540)
+ gl:^{x) + 2.940 (.T).g42(rr),
943(2x) = 294o(^)fl'43(^) + 2g4i(x)gi2{x).
(3.541)
For :r = - y the relations (3.534)-(3.537) and (3.529) yield 540(^) + .942(^) - 294l(^)543(^) = 1,
(3.542)
94i{^) + 543(^) ~ 294o(x)542(a:) = 0.
(3.543)
From Eqs. (3.522)-(3.524) it can be shown that, for m integer, (cosh z -\- p sinh z)^^ = cosh mz + /3 sinh m,z^
(3.544)
[gioiy) + aguiy) + I3g42{y) + igMy)]"^ = g4o{rny) + ag^ (my) -f I3g42{my) + jgA^imy),
(3.545)
and
[g4o{t) + 7541 (0 + /3g42{t) + a943(^)]^ = g4o{f^nt) + 7941 (m^) + ^g42{mt) + ^943(^1^).
(3.546)
Since (a + 7)2^ = 2^^^-^(1 + /?), (a + 7 ) ' ' ^ ^ ' = ^^'''{a + 7),
(3.547)
it can be shown from the definition of the exponential function, Eq. (1.35) that exp(a + 7),T = ei + ——^ cosh2x + ^~^
sinh2,T.
(3.548)
Substituting in the relation exp(a + j)x = cxpaa;exp7.T the expression of the exponentials from Eqs. (3.523), (3.524) and (3.548) yields 2
2
2
2
.940 + .941 + .942 + 94.! =
1 + cosh 2.7;
n
'
(^^-549)
Polar complex numbers in four dimensions — 1 -f c o s h 2a;
^40^42 + 941943 =
2
131 /o r e n \
.
940941 + 940943 + 941942 + 942943 = -^ sinh2x,
(3.550) (3.551)
where 940? 941^942? 943 sire functions of a:. Similarly, since ( a - 7 ) 2 ^ = (~l)^22^-Hl-/3), (a -- 7)2^+1 = (-l)^22"^(a - 7),
(3.552)
it can be shown from the definition of the exponential function, Eq. (1.35) that exp(a — 7)3: = —
h ei cos2x + ei sin 2a:.
(3.553)
Substituting in the relation exp(a — 7)0; = expaxexp(—70;) the expression of the exponentials from Eqs. (3.523), (3.524) and (3.553) yields 2 2 , 2 2 1 + c o s 2a: 940 - 941 + 942 - 943 = 2 ' l-COS2x 940942 - 941943 =
,
940941 - 940943 - 941942 + 942943 = ^ sin2a:,
/o ccix ^^'^^^ /o c:rr:^ (3.555) (3.556)
where 540,941,942 7 943 are functions of x. The expressions of the four-dimensional cosexponential functions are 940(a;) = -(cosha; + cos a:),
(3.557)
94i(^) = -(sinha: -f sina;),
(3.558)
942(^) = -(cosha; - cosx),
(3.559)
943 (^) = -(sinha: — sin a:).
(3.560)
The graphs of these four-dimensional cosexponential functions are shown in Fig. 3.4.
132
Commutative Complex Numbers in Four Dimensions
u
^40
>
\^.2
h 0 \-
S.3
^
u
A./
-1 V-
-2 h
/
-3
[
-2
_i
^ 0
1
.1
1
X
Figure
3.4:
540K94l7ff42i943'
The
polar
fourdimensional
cosexponential
functions
Polar complex numbers in four dimensions
133
It can be checked that the cosexponential functions are solutions of the fourth-order differential equation ^ = < ,
(3.561)
whose solutions are of the form C{u) = Ag4{}{u) + Bgii{u) -f Cg42(u) + Dgisiu). It can also be checked that the derivatives of the cosexponential functions are related by ^940 ^941 ^342
aw
3.4.4
aw
aw
.^ ran\ (3.562)
aw
The exponential and trigonometric forms of a polar fourcomplex number
The polar fourcomplex numbers u = x + ay + ^z + ^yt (or which v^ = x + y-\~z-\-t>0^ V- =x — y-\-z — t>0 can be written in the form x + ay + 0z-^jt
= e^i+^^i^^^^^^^^.
(3.563)
The expressions of xi.yi^zi^ti as functions of x,y^z^t can be obtained by developing e^^Se^^i and e^^' with the aid of Eqs. (3.522)-(3.524), by multiplying these expressions and separating the hypercomplex components, and then substituting the expressions of the four-dimensional cosexponential functions, Eqs. (3.557)-(3.560), x + y + z + t = e^i+J/i+^i+^i,
(3.564)
x-z
(3.565)
= e^'-'' cos(yi - h),
x--y^z-t y-t==
= e'^'-y'-^''-^',
(3.566)
e^^-"i sin(yi - ti).
(3.567)
It can be shown from Eqs. (3.564)-(3.567) that ^1 = 2 ^"(/^+^-)' y^ ^ 2^^+^)^
^^ " 2 ^" ~'
^^ ^ 2^'P'-^)^ (3.568)
where nl = {x + zf ~ (y + tf = v^v^,
v^ > 0, v^> 0.
(3.569)
The quantities (f) and a; are determined by COS0 = {x- z)lii^,
sine/) = (y - t)jii^,
(3.570)
134
Commutative Complex Nmnhers in Four Dimensions
and cosho; - {x-i- z)/^i-., sinha; = (y -f 0//^--
(3.571)
The explicit form of a; is 1,
X -\- y -{- z
-^t
If w = ^1^2, and //|_ = [xj + ^j)'^ - (?/j + tjf'.j with the aid of Eqs. (3.483) that
= 1,2, it can be checked (3.573)
/i- =//i_/i2-.
Moreover, if coshcjj = {xj + Zj)/iij-, can be checked that
sinhcjj = {yj -f tj)/^j^,j
= 1,2, it (3.574)
CJU = CUI+UJ2.
The relation (3.574) is a consequence of the identities (j:ia:2 -f <2^1^2 + 2/1^2 + ^l?/2) + {xiZ2 + ZxX2 4- yi?/2 +
tit2)
= {xi + zi){x2 + Z2) + (yi + U){y2 + ^2),
(3.575)
(^l?/2 +y\X2 + Zit2 + t]Z2) + {X\t2 ^ t\X2 •\- Z\]j2 + y\Z2) = (yi + tx){x2 + Z2) 4- (x-i 4- ^l)(2/2 + ^2).
(3.576)
According to Eq. (3.570), (^ is a cyclic variable, 0 < 0 < 27r. As it has been assumed that v^ > 0, t;_ > 0, it follows that x -\- z > Q and X •¥ z > \y -{- t\. The range of the variable u is —oc < u; < oc. The exponential form of the polar fourcomplex number u is then u = pexp - ^ In ^
-f -a{ij
+ 0) 4- -j{uj - 0)
(3.577)
where p=
{fi+fi^) 1/2
If n = ttiit2, and pj = {f^ij^fij^y^'^.j (3.573) it results that
(3.578) — 1,2, then from Eqs. (3.489) and
P = P\P2'
(3.579)
It can be checked with the aid of Eq. (3.513) that 1 4-/i exp (ei(/>) = —~ h e\ cos 0 4- ei sin (j).
(3.580)
Polar complex numbers in four dimensions
135
which shows that e^" ^^^/'^ is a periodic function of 0, with period 2iv. The modulus has the property that \uexp{ei(f))\ = \u\
(3.581)
By introducing in Eq. (3.577) the polar angles 6^,0^ defined in Eqs. (3.495), the exponential form of the fourcomplex number u becomes u = pexp
^ i" ^ ^ ^ ^ ^ ^ " ^
1 V5 - - ( a - /9 + 7) In - — — + ei^ 4 tan y_
(3.582)
where 0 < ^+ < 7r/2,0 < ^_ < 7r/2. The relation between the amphtude p, Eq. (3.578), and the distance d, Eq. (3.481), is, according to Eqs. (3.495) and (3.496), 2^1'd (tan^-). tan^_) 1/4
1 1 Qj^. + tan2 6. 1 + tan2 :—oTr-
-1/2
(3.583)
Using the properties of the vectors e+,e_ written in Eq. (3.511), the first part of the exponential, Eq. (3.582) can be developed as exp
i<°+''+^"°i^-5("-''+^""i:;^ 1/4
( - tan 6+ tan 0_ j
^/2 V2 ei + e + + e_tan 6^ ' tan &_
(3.584)
The fourcomplex number u, Eq. (3.582), can then be written as w = dV2 ( l + + , 2/) ) V tan^&+ ta.n'^O-J exp(ei0),
v/2 ( ^1 + ^+:^—F + e. tan 6- I y tan ^4 (3.585)
which is the trigonometric form of the fourcomplex number u. The polar angles ^-f-,0_, Eq. (3.495), can be expressed in terms of the variables ^, A, Eq. (3.503), as tan^+ =
1 1 , tan^-. = tan ^ sin A' tan 9 cos A
(3.586)
so that 1
1
19+ + tan2 6^_ 1 + tan^ :—^r^
cos2(9'
(3.587)
136
ConmmtMive Complex Numbers in Four Dimensions
The exponential form of the fourcomplex number ix, written in terms of the amplitude p and of the angles ^, A, cp is u = pexp -(of + /3 -h 7) ln(v^tan<9cos A) 4^ •-{a - /? + 7) ln(\/2 tan(9sin A) + ei0
(3.588)
where 0 < A < 7r/2. The trigonometric form of the fourcomplex number u, written in terms of the amplitude p and of the angles 6^ A, (j) is u = d v 2 fei cos^ -f- e_j_v2sin^cos A + e_\/2sin6^sinAj exp(ei0). (3.589) Ifu — u\U2^ it can be shown with the aid of the trigonometric form, Eq. (3.585), that the modulus of the product as a function of the polar angles is d' = 4dl4 (^2 + tan2^i+tan2^2+ "^ tan^ 0i_ t a n ^ ^ s - j V ^ t a n 2 0,+ " ^ t a n ^ e i ^ j
V'^
i^i^0^'^
i^^?^)
' (3.590)
The modulus d of the product u\U2 can be expressed alternatively in terms of the angles ^, A, (j) with the aid of the trigonometric form, Eq. (3.589), as (f = 4:d\dl i - cos^ Ox cos^ 62 + sin^ 6>i sin^ 62 cos^ Aj cos^ A2 -f sin^ 9\ sin^ ^2 sin^ Ai sin^ A2J . (3.591)
3.4.5
Elementary functions of polar fourcomplex variables
The logarithm u\ of the polar fourcomplex number n, u\ ~ Inu, can be defined as the solution of the equation ^ = e^^
(3.592)
written explicitly previously in Eq. (3.563), for u\ as a function of u. From Eq. (3.577) it results that \nu = \np+ l-PIn ^— + l-a{uj + 0) 4- IJ{U; - 0). 2 //-f 2 2
(3.593)
Polar complex numbers in four dimensions
137
If the fourcomplex number u is written in terms of the ampHtude p and of the angles O^^O^^cp^ the logarithm is
(3.594) where 0 < ^-}- < IT/2, 0 < 6J^ < 7r/2. If the fourcomplex number u is written in terms of the amplitude p and of the angles 0, A, <^, the logarithm is Inw = ln/9+ ~(Q:-f;0 + 7)ln(\/2tanecosA) - - ( a - /? + 7) ln(V^tan(9sin A) + eicf),
(3.595)
where 0 < ^ < 7r/2,0 < A < 7r/2. The logarithm is multivalued because of the term proportional to (f). It can be inferred from Eq. (3.594) that \n{uiU2) = Inni -h lnu2,
(3.596)
up to multiples of 7r(a — 7). If the expressions of p, //-[-,//_ and u in terms of x,y,z,t are introduced in Eq. (3.593), the logarithm of the polar fourcomplex number becomes mu = +
ln(x + y -i- z + t) 1 — a + /3 — 7 j ^ ln(a; '-y + z-t)
+ ei ln/i+ + eicf),
The power function u^ can be defined for v^ > 0,vvalues of m as
(3.597) > 0 and real
The power function is multivalued unless m is an integer. For integer m, it can be inferred from Eqs. (3.593) and (3.598) that {uiU2r = uTu^.
(3.599)
Using the expression (3.597) for \nu and the relations (3.511)-(3.514) it can be shown that u ^ = e^v"^ + e-.v^ + p^ (ei msm(^ + ei sinm(/)). For integer m, the relation (3.600) is valid for any x,y,z,t. this relation can be written as u"^ = e^v"^ + e^v'^ + [ei{x ~ z)-h ei(y-- t)^ ,
(3.600) For natural m (3.601)
138
Commutative Complex Numbers in Four Dimensions
as can be shown with the aid of the relation ei cos m0 + ei sin m^ = (ei cos 0 + ei sin 0 ) ^ ,
(3.602)
vaHd for natural m. For m = - 1 the relation (3.600) becomes 1 _ 1 /^l-f a - f / 3 4 - 7 l~a + /?-7\ a; + ay + /?2 + 7^ 4- (V- ^ + ?/ + '2 + ^ x — y + z — tj 4 Va: l(l-0){x-z)-{a-^)(y-t) ^2 {X - zf + {y- ty •
(^-^^^^
If m = 2, it can be checked that the right-hand side of Eq. (3.601) is equal to (x-^ay+^z+jtf = x'^-\'Z^+2yt+2a{xy+zt)+P(y'^-^f+2xz)+2y{xt+yz). The trigonometric functions of the hypercomplex variable u and the addition theorems for these functions have been written in Eqs. (1.57)(1.60). The cosine and sine functions of the hypercomplex variables ay,/3z and 7^ can be expressed as cos ay = gio ~ /3.g42, sin ay = ag4i - 7^43,
(3.604)
cos/3y = cosy, sin/3y = Psiny,
(3.605)
cos7y = ^40 - )0542, sin7y = 7541 - a943.
(3.606)
The cosine and sine functions of a polar fourcomplex number x+aj/4-/3^+7^ can then be expressed in terms of elementary functions with the aid of the addition theorems Eqs. (1.59), (1.60) and of the expressions in Eqs. (3.604)-(3.606). The hyperbolic functions of the hypercomplex variable u and the addition theorems for these functions have been written in Eqs. (1.62)-(1.65). The hyperbolic cosine and sine functions of the hypercomplex variables ay, /3z and jt can be expressed as cosh ay ^ 340 + %12, sinhay = agn + 7343,
(3.607)
cosh Py = cosh y, sinh Py = p sinh y,
(3.608)
cosh7y = ^40 + /3g42, sinhyy = 7^741 + ag43.
(3.609)
The hyperbolic cosine and sine functions of a polar fourcomplex number x + ay-\-f3z+'yt can then be expressed in terms of elementary functions with the aid of the addition theorems Eqs. (1.64), (1.65) and of the expressions in Eqs. (3.607)-(3.609).
Polar complex numbers in four dimensions 3.4,6
139
P o w e r series of polar f o u r c o m p l e x variables
A polar fourcomplex series is an infinite sum of the form ao + ai -f a2 H
1- a/ H
,
(3.610)
where the coefficients a/ are polar fourcomplex numbers. The convergence of the series (3.610) can be defined in terms of the convergence of its 4 real components. The convergence of a polar fourcomplex series can however be studied using polar fourcomplex variables. The main criterion for absolute convergence remains the comparison theorem, but this requires a number of inequalities which will be discussed further. The modulus of a polar fourcomplex number u = x -h ay + I3z + jt can be defined as \u\^{x^-hy^
+ z^ + ty/\
(3.611)
so that according to Eq. (3.481) d = \u\. Since \x\ < \u\,\y\ < \u\,\z\ < |w|,|t| < \u\, a property of absolute convergence established via a comparison theorem based on the modulus of the series (3.610) will ensure the absolute convergence of each real component of that series. The modulus of the sum Ui + U2 of the polar fourcomplex numbers u\^U2 fulfils the inequality \\ui\ - K l l < 1^1 + U2\ < \ui\ + \u2l
(3.612)
For the product the relation is \uiU2\ <2|ui||u2i,
(3.613)
which replaces the relation of equality extant for regular complex numbers. The equality in Eq. (3.613) takes place for xi = yi = zi =^ ^1,^2 — y2 = Z2 = t2, or xi — -j/i ~ zi — -t\,X2 — -y2 = ^2 = -^'2- In particular In^l < 2(x^ + y^ + ^2 + t^).
(3.614)
The inequality in Eq. (3.613) implies that \vt\<2^~^\u\K
(3.615)
From Eqs. (3.613) and (3.615) it results that \avt\ < 2 ' | a | h ^ ^
(3.616)
A power series of the polar fourcomplex variable n is a series of the form ao + axu + a2U^ + • • • + a/w' + • • •.
(3.617)
140
Commutative Complex Numbers in Four Dimensions
Since I 00
00
I
1
(3.618)
1=0
a sufficient condition for the absolute convergence of this series is that lim ^
1
l-^^
\ai\
^ < 1.
(3.619)
Thus the series is absohitely convergent for |n| < Co,
(3.620)
co= l i m - M ^ .
(3.621)
wliere
The convergence of the series (3.617) can be also studied with the aid of the formula (3.601) which, for integer values of/, is valid for any x^y.z^t. If ai = aix + aaiy 4- /3aiz + ^au, and Ai^ = aix + aiy -f (iiz -f ait, A,^^ai,-aiy-\-ai,-au,
(3.622)
it can be shown with the aid of relations (3.511)-(3.514) and (3.601) that the expression of the series (3.617) is oo
Y, [Af+<e+ + Ai^v-t^
-f (eiA/1 4- hAn)
{cxvx -f eitJi)'] , (3.623)
where the quantities ?;_|-,?;_ have been defined in Eq. (3.475), and the quantities v\,v\ have been defined in Eq. (3.486). The sufficient conditions for the absolute convergence of the series in Eq. (3.623) are that lim l-^^tl'"'!'"' < 1- l i - i % 4 ^ /->oc
\Aij^\
/->oc
|A/_|
< 1, lira ^ /->oo
< 1, (3.624) Ai
where the real and positive (juantity Ai^ > 0 is given by Af = Al+Af,.
(3.625)
Thus the series in Eq. (3.623) is absolutely convergent for \x + y + z + t\< c+, \x -y + z -t\<
c_, //.f < ci,
(3.626)
Polar complex numbers in four dimensions
141
where c^ — lim 7-^—^^, c_ = lim —
:, c\ = lim
.
(3.627)
The relations (3.626) show that the region of convergence of the series (3.623) is a four-dimensional cylinder. It can be shown that CQ = (1/2) min(c4-,c_,ci), where min designates the smallest of the numbers in the argument of this function. Using the expression of \u\ in Eq. (3.488), it can be seen that the spherical region of convergence defined in Eqs. (3.620), (3.621) is included in the cylindrical region of convergence defined in Eqs. (3.627).
3*4.7
Analytic functions of polar fourcomplex variables
The fourcomplex function / ( u ) of the fourcomplex variable u ha^ been expressed in Eq. (3.117) in terms of the real functions P{x^ y, z, t), Q{x, y^ z, t)^ R{x^y,Zjt)^S{x,y,z,t) of real variables x^y^z.t. The relations between the partial derivatives of the functions P, Q, R, S are obtained by setting succesively in Eq. (3.118) Ax -^ 0,Ay = Az = At = 0; then Ay ~> 0,Ax = A^ = A/ = 0; then Az -> 0, AJ: = Ay = At = 0; and finally At -> 0, Ax = Ay = A2: = 0. The relations are dP _dQ
_dR_
dS
_dR _ dS dx dy '' dz
dP dt'
(3.629)
dP dQ dz ~ dt'
(3.630)
OR dx
dy
dS _dP _ dQ dR dx dy dz ~' df
(3.631)
The relations (3.628)-(3.631) are analogous to the Riemann relations for the real and imaginary components of a complex function. It can be shown from Eqs. (3.628)-(3.631) that the component P is a solution of the equations a2p
d'^p
^
d'^p
d'^p
^
.^^^^^
and the components Q, i?, 5 are solutions of similar equations. As can be seen from Eqs. (3.632)-(3.632), the components P,Q,R,S of an analytic
142
CoinniutMive Complex Numbers in Four Dimensions
function of polar foiircomplex variable are harmonic with respect to the pairs of variables x,y and z,t. The component P is also a solution of the mixed-derivative equations
and the components Q, /?, S are solutions of similar equations. The component P is also a solution of the mixed-derivative equations d^P _ d^ dxdy dzdt'
d^ _ d^^P dxdt dydz'
and the components Q, i?, S are solutions of similar equations.
3-4.8
Integrals of functions of polar fourcomplex variables
The singularities of polar fourcomplex functions arise from terms of the form l/(?i--txo)^i with m. > 0. Functions containing such terms are singular not only at u — UQ^ but also at all points of the two-dimensional hyperplanes passing through UQ and which are parallel to the nodal hyperplanes. The integral of a polar fourcomplex function between two points A^ B along a path situated in a region free of singularities is independent of path, which means that the integral of an analytic function along a loop situated in a region free from singularities is zero, (f f{u)du = Q,
(3.635)
where it is supposed that a surface E spanning the closed loop F is not intersected by any of the hyperplanes associated with the singularities of the function / ( n ) . Using the expression, Eq. (3.117) for / ( n ) and the fact that du = dx + ady -h ftdz -{- jdt, the explicit form of the integral in Eq. (3.635) is i f{u)du = i [{Pdx + Sdy + Rdz + Qdt) +a{Qdx + Pdy + Sdz + Rdi) + p{Rdx + Qdy + Pdz + Sdt) +^{Sdx + Rdy + Qdz + Pdt)].
(3.636)
If the fvmctions P,Q,R,S are regular on a surface S spanning the loop F, the integral along the loop F can be transformed with the aid of the theorem of Stokes in an integral over the surface S of terms of the form
dP/dy-dS/dx, dQ/dy,
dP/dz-dR/dx,
dP/Of-dQ/dx,
OR/dy-dS/dz,
dS/dt-
dR/dt - OQ/dz and of similar terms arising from the aj-i and 7
Polar complex numbers in four dimensions
143
components, which are equal to zero by Eqs. (3.628)-(3.631), and this proves Eq. (3.635). The integral of the function {u — UQ)^ on a closed loop F is equal to zero for m a positive or negative integer not equal to -1, / .r
{u - uo)'^du = 0, m integer, m^
-I.
(3.637)
This is due to the fact that J{u - uo)'^du = [u- UQ)'^^^/{m. 4-1), and to the fact that the function {u - UQ)^"^^ is singlevalued for m an integer. The integral §p du/{u~U{)) can be calculated using the exponential form (3.582), 1 \/2 u — ifco = p exp - ( a + /3 + 7)ln4 tan 0+ (3.638) •
^
^
"
-
^
^
^
^
^
^
^
^
"
"
'
^
^
^
so that du dp 1, ^ , ,, \/2 - ^ + -;-(a + ^ + 7)dln u — UQ P 4 tan ^+ +eid0.
1. ^ , ,, \/2 - ( a - / 3 + 7)dln4 tan ^_ (3.639)
Since p, tan ^4- and tan d- are singlevalued variables, it follows that §^ dp/p = 0, ^ r f l n V^/tan^-f = 0, and ^ d In \ / 2 / t a n ^4. = 0. On the other hand, (^ is a cyclic variables, so that it may give a contribution to the integral around the closed loop F. The result of the integrations will be given in the rotated system of coordinates
^ "^ 71^^ "" ^^' ^ "" 71^^ - ^), T = -(J: -f 2/ + ^ + t), v = -{x-y
+ z-t),
(3.640)
Thus, if C|| is a circle of radius r parallel to the (^Ov plane, and the projection of the center of this circle on the ^Ov plane coincides with the projection of the point UQ on this plane, the points of the circle C\\ are described according to Eqs. (3.486)-(3.487) by the equations C = Co + 7^cos), V = uo + r s i n 0 , r = TQ, C = Co,
(3.641)
where WQ = XQ + ayo + /3^o + 7^o? and ^o,'^o,'7'o,Co are calculated from ^o,?/o,'2ro,^o according to Eqs. (3.640).
144
Commutative Complex Numbers in Four Dimensions Then
i
= 27rei.
(3.642)
C|, U - Wo
The expression of ^. du/{u-7io) can be written with the aid of the functional int(M,C) defined in Eq. (3.134) as
i
du
^ ^ . / _ . = 27rei mt{uo^y, T^t, ,
(3.643)
where UQ^y and T^y are respectively the projections of the point UQ and of the loop r on the plane ^v. If/(?i) is an analytic polar fourcomplex function which can be expanded in a series as written in Eq. (1.89), and the expansion holds on the curve F and on a surface spanning T, then from Eqs. (3.637) and (3.643) it follows that
i
^ ^^^^
= 2iTeimt{uo^y,r^y)f{uo).
(3.644)
where T^y is the projection of the curve F on the plane ^i;, as shown in Fig. 3.5. Substituting in the right-hand side of Eq. (3.644) the expression of f{u) in terms of the real components P, Q, /?, 5, Eq. (3.117), yields
i
^ ^ ^ = TT [(/? - 1)(Q - 5) + ( a - - 7 ) ( i " ~/?)] m^ r ^ ^u-uo (3.645)
where P, Q, R^ S are the values of the components of / at w = WQ- If the integral is written as
i
(3.646)
/ + /« + //J + 7^ = 0.
(3.647)
= ^ ^ ^ = I + aIa+ /3l0 + 7 / v iv u- UQ it results from Eq. (3.645) that
If f(u) can be expanded as written in Eq. (1.89) on V and on a surface spanning T, then from Eqs. (3.637) and (3.643) it also results that / , ^^""^t+i = -^'-^ int.(«o«.,r^.) /('^'Huo), (3.648) / r {u - uo)^"^^ rn! where the fact has been used that the derivative J^^'^^UQ) of order m of /(w) at u = Ho is related to the expansion coefficient in Eq. (1.89) according to Eq. (1.93).
Polar complex numbers in four dimensions
145
Figure 3.5: Integration path T and the pole UQ^ and their projections F^^ and UQ^y on the plane ^f. If a function f{u) is expanded in positive and negative powers of u — Uj^ where Uj are polar fourcomplex constants, j being an index, the integral of / on a closed loop F is determined by the terms in the expansion of / which are of the form aj/{u — Uj)^ (3.649)
fiu) = -.. + Y.U — Uj Then the integral of / on a closed loop F is j> f{u)du =
3.4.9
2n€iY^int{uj^y,r^y)aj.
(3.650)
Factorization of polar fourcomplex polynomials
A polynomial of degree m of the polar fourcomplex variable u = x -{- ay -\-(3z + jt has the form Prniu) - ix^ + aiu'''-^ +'"+
am-iu + a^,
where the constants are in general polar fourcomplex numbers.
(3.651)
146
Commutative Complex Numbers in Four Dimensions
If am = o.mx + otamy + ftdmz + icimi-, and witli the notations of Eqs. (3.475) and (3.622) applied for 0,1, • • • ,m , the polynomial Pm{y) can be written as Pm = [ < + A i < ~ ^ + • • • + Am~\v^ + Am] e+ r
+
m
{eivi + eit)!)"" -f Y, (eiAn + e i i a ) (eitn + eiv^)"^"^ ^=1
(3.652) where the constants Ai^,Ai^,Aii^Ai\ are real numbers. The polynomial of degree m in {eiv\ 4- ei^i) can always be written as a product of linear factors of the form [ex {v\ —vii)-\-e\ {vi —1)1/)], where the constants i;y, vn are real. The two polynomials of degree m with real coefficients in Eq. (3.652) which are multiplied by e^ and e_ can be written as a product of linear or quadratic factors with real coefficients, or as a product of linear factors which, if imaginary, appear always in complex conjugate pairs. Using the latter form for the simplicity of notations, the polynomial Pm can be written as m
m
rn
+ n [^1 (^1 ~ ^^1^) + ^1 (^1 ~ ^1^)]'
(3.653)
where the quantities si^ appear always in complex conjugate pairs, and the same is true for the quantities cS/_. Due to the properties in Eqs. (3.511)(3.514), the polynomial Pmi'^i') can be written as a product of factors of the form
1=1
+ (ei(t;i - vxi) + ei(Oi - vu))].
(3.654)
These relations can be written with the aid of Eq. (3.515) as m
Pm{u) = Y[{u - Up),
(3.655)
where Up = Sp+e+ + Sp_e_ +eiwip + eivip,p= l,...,m.
(3.656)
Polar complex numbers in four dimensions
147
The roots Sp^,Sp^,vipei + v\pe\ of the corresponding polynomials in Eq. (3.653) may be ordered arbitrarily. This means that Eq. (3.656) gives sets of TTi roots 1/1,..., Um of the polynomial Pm{u)^ corresponding to the various ways in which the roots Sp4-,5p_,^ipei -f v\pe\ are ordered according to p in each group. Thus, while the hypercomplex components in Eq. (3.652) taken separately have unique factorizations, the polynomial Pm{'^) can be written in many different ways as a product of linear factors. The result of the polar fourcomplex integration, Eq. (3.650), is however unique. If P{u) — xi^ — \^ the factorization in Eq. (3.655) is u^ — 1 = (u — u\){u — ui)^ where u\ = ±e^- ± e_ ± ei,W2 = —^h so that there are 4 distinct factorizations of tx^ — 1, u - l = {u-(3){u [u —
l-hQ
+ ^),
^)(„+i±«rto),
(3.657)
It can be checked that {ie-f ± e_ ± ei} = e-f + e_ + ei = 1. 3.4.10
R e p r e s e n t a t i o n of polar f o u r c o m p l e x n u m b e r s by irreducible matrices
If T is the unitary matrix, / 1 i i
' 1 2
\ 2
? 21 v/2
1 0 V 0 " 73 •TJ V2 it can be shown that the matrix TUT "^ has the form X + y -^ z -i-t 0 0 \ TUT-^ = I 0 x-y-^z t 0 0 0 V, J
(3.658)
(3.659)
where U is the matrix in Eq. (3.513) used to represent the polar fourcomplex number u. In Eq. (3.659), Vi is the matrix Vi
( X— z y — t \ ^ -y + t x-z J'
(3.660)
The relations between the variables x + y + z + t^x — y + z — t^x — z^y — t for the multiplication of polar fourcomplex numbers have been written in Eqs. (3.484),(3.485), (3.492), (3.493). The matrix TUT~^ provides an irreducible representation [7] of the polar fourcomplex number u in terms of matrices with real coefficients.
This Page Intentionally Left Blank
Chapter 4
Complex N u m b e r s in 5 Dimensions A system of complex mimbers in 5 dimensions is described in this chapter, for which the multiphcation is associative and commutative, which have exponential and trigonometric forms, and for which the concepts of analytic 5-complex function, contour integration and residue can be defined. The 5-complex numbers introduced in this chapter have the form u = xo + hixi + /^2^2 + hsx:i + /i4X4, the variables xo^xi,X2^xs^X4 being real numbers. If the 5-complex number u is represented by the point A of coordinates xo^x\,X2^x:^,X4, the position of the point A can be described by the modulus d = {XQ 4- x'i -\- X2 + x'^ -\- x'ly^'^, by 2 azimuthal angles 01^02, by 1 planar angle t/?i, and by 1 polar angle 0^. The exponential function of a 5-complex number can be expanded in terms of the polar 5-dimensional cosexponential functions g^kiv), k = 0,1,2,3,4. The expressions of these functions are obtained from the properties of the exponential function of a 5-complex variable. Addition theorems and other relations are obtained for the polar 5-dimensional cosexponential functions. Exponential and trigonometric forms are given for the 5complex numbers. Expressions are obtained for the elementary functions of 5-complex variable. The functions f{u) of 5-complex variable which are defined by power series have derivatives independent of the direction of approach to the point under consideration. If the 5-complex function f{u) of the 5-complex variable u is written in terms of the real functions Pk{xo^xi.X2,x:i,X4),k = 0,1,2,3,4, then relations of equahty exist between partial derivatives of the functions P^. The integral /^ f{u)du of a 5-complex function between two points A, B is independent of the path cormecting A, 5 , in regions where / is regular. The fact that the exponen149
150
Complex Numbers in 5 Dimensions
tial form of the 5-complex numbers depends on the cyclic variables (f)\, ^2 leads to the concept of pole and residue for integrals on closed paths, and if f{u) is an analytic 5-complex function, then ^ f{u)du/{u-HQ) is expressed in this chapter in terms of the 5-complex residue f{uo). The polynomials of 5-complex variables can be written as products of linear or cjuadratic factors. The 5-complex numbers described in this chapter are a particular case for n = 5 of the polar complex numbers in n dimensions discussed in Sec. 6.1.
4.1
Operations with polar complex numbers in 5 dimensions
A polar hypercomplex number u in 5 dimensions is represented as U = XQ + hiXi -f h2X2 -f /^3^3 -f h4X4. The multiplication rules for the bases h],h2,hs^h^
(4.1) are
h\ = /i2, /i2 ~ ^Mi hi = /ii, fil = /i3, hih'2 = /13, hih's = /M, hih4 — 1, /i2^3 = 1, ^2^4 = h\,h:ih.\ = /i2.
(4.2)
The significance of the composition laws in Eq. (4.2) can be understood by representing the bases hj^h^ by points on a circle at the angles Oj = 27rj/5,ak = 27rA;/5, as shown in Fig. 4.1, and the product hjhk V>y the point of the circle at the angle 27r(j + k)/b. If 27r < 2n{j -f k)/b < 4n, the point represents the basis hi of angle cv/ = 2n{j -f k)/5 — 27r. The sum of the 5-complex numbers u and u' is u+ii' =
xo+XQ+h](xi+x\)+h2(x2+X2)+h:]{x:i+x':^)+h4{x4+x\).(4.3)
The product of the numbers ?/, w' is then W
= XQXQ + X\x[^ + X2X'^ + X:^X2 + X4X\
+hi{x{ix\ + X\XQ + ,T2X4 + a;3:r3 + .7:4X2) +h2{xQx'2 + x\x\ 4- .T2.T0 + 0:30:4 + X4.T3) -f/i3(x'oo;3 -f x\x'2 4- 0:20:1 -f rr3o:Q + .0:40:4) ^-h4{x^x[4 + :rio;3 4- 0:20:2 + :r3o:i 4- 0:4X0).
(4.4)
Geometric representation of polar complex numbers in 5 dimensions
151
Figure 4.1: Representation of the polar hypercomplex bases 1, /ii, ^2, /^a, ^^4 by points on a circle at the angles ak = 27rfc/5. The product hjhic will be represented by the point of the circle at the angle 2n(j -f fc)/5, i^k = 0,1,..., 4, where /IQ = 1. If 27r < 27r(j + k)/5 < 47r, the point represents the basis hi of angle a/ = 2n{j + k)/b — 2n. The relation between the variables v^^vi^vi^V2',V2 stnd XQ^xi,0:2,3:3,X4 can be written with the aid of the parameters p = (\/5 — l)/4,g' = ^(5 + N / 5 ) / 8
as
/ ^+ \
/ I
Vi
=
1
V2
\
^2
1
1 0
Vi
)
\0
p q
2p^-l 2pq
1
1
1
2p2_l 2pq
2p2-l
P -q 2p2_i
P -9
-2pq P Q
W a;o \ Xl X2
xz
-2pg J \X4 J (4.5)
The other variables are v^ — 1^2?^3 = ~'^2^'^\ = "^17^4 = —^i- The variables i;^,t;i,7)i,i;2,'02 will be called canonical 5-complex variables.
4.2
Geometric representation of polar complex numbers in 5 dimensions
The 5-complex number XQ •\-h\X\-\- /i2^2 + ^3^3 + ^4^4 can be represented by the point A of coordinates (xo,a:^i, 3:2,^3, ^4). If O is the origin of the
152
Complex Numbers in 5 Dimensions
5-dimensional space, the distance from the origin O to the point A of coordinates {x(),xi,X2,x:i,X4) has the expression (f = xl + x\ + xj +xl + xj.
(4.6)
The quantity d will be called modulus of the 5-complex number u. The modulus of a 5-complex number u will be designated by d = \n\. The modulus has the property that IwV'l < y/b\u'\\2/'l
(4.7)
The exponential and trigonometric forms of the 5-complex number u can be obtained conveniently in a rotated system of axes defined by the transformation
J_
m
1
/I
V '72 y
\
p
0
q
1 0
2p2 - 1 2pq
A:
v/2 ,2 2p^
1 2p 2pg -2p9 P P -q q
1 v/2
Xi
P
X2
2p2_i -2P9
X3
/
\X4
/
(4.8)
The lines of the matrices in Eq. (4.8) gives the components of the 5 basis vectors of the new system of axes. These vectors have unit length and are orthogonal to each other. The relations between the two sets of variables are v+ = y/hi^,Vk = y:^^k.h
= y~Vk,k
= 1,2.
(4.9)
The radius p^ and the azimuthal angle cpk in the plane of the axes Vk, Vk are Pk = ^k + ^h ^^^^k = Vk/pk, sin(/)fc = Vk/pk^
(4.10)
0 < 0A: < 27r, A: = 1,2, so that there are 2 azimuthal angles. The planar angle ^i is (4.11) where 0 < ^i < 7r/2. There is a polar angle 0^, tan0+ =
y/2pi v^
(4.12)
Geometric representation of polar complex nmnbers in 5 dimensions
153
where 0 < ^.f < TT. It can be checked that (4.13) The amplitude of a 5-complex number u is p = (V^PIPI)
,1/5
' •
(4.14)
If zx = u'u", the parameters of the hypercomplex numbers are related by
v+ = v'^v'l,
(4.15)
Pk = p'kPk^
(4.16)
tan 6+ = —j= tan Q\ tan ^", v2
(4.17)
tan V"! = tan V"! tan i^^ >
(4.18)
k = 4>k + 'L
(4.19)
^k = v'kv'k - v'kv',^, Vk = v'kv'k + v'kv'^,
(4.20)
P = P'p",
(4.21)
where A; = 1,2. The 5-complex number u = XQ -^ hixi + h2X2 + h^x^ -f /14X4 can be represented by the matrix / Xo
Xi
X2
Xs
X4
X4
XQ
XI
X2
X^
Xs
X4
Xo
Xi
X2
X2
X3
X4
XQ
XI
Xi
X2
X3
X^
XQ
[/ = \
\
(4.22) )
The product u = w'u" is represented by the matrix multipUcation \J
154
Complex Numbers in 5 Dimensions
4.3
T h e polar 5-dimensional cosexponential functions
The polar cosexponential functions in 5 dimensions are oo
95k{y) = Yly''^"'/ik
+ 5py.,
(4.23)
for k = 0, ...,4. The polar cosexponential functions g^k do not have a definite parity. It can be checked that 4
Yl9f^k(y) = ^''
(4-24)
The exponential of the quantity hky^ k = 1,..., 4 can be written as e^'^ e^'^ e^^y e^^^
= = = =
g^oiv) + hig^iiy) + ^2^52(1/) + h^g^^iy) + M54(y), gm{y) + higssiy) + ^2ff5i(l/) + h^g^^iy) + /i4352(y), g^Q{y) + hig^2{y) + M54(j/) + hsg^iiy) + h^gb^{y), g5o{y) + hxg^iy) + M53(y) + hsgr,2(y) + h^gbiiy)-
/^ 25)
The polar cosexponential functions in 5 dimensions can be obtained by calculating e^^^^^^"^^ and e*'^^"^'*^^ and then by nultiplying the resulting expression. The series expansions for e^'*^"^'^'*)^ and e^^^'^'^^y are 00
,(/.l+/^4)t/ ^
^
3(/ii-/i4)2/ ^
^
00
^
_ ( ^ , ^ ^ /l4)^^y'",
(4.26)
.
_ ( / , ^ _ /i4)^^^.
(4.27)
^ m! The powers of h\ + /14 have the form [h, + IHT = ^m(/H + /i4) + Bm[h2 + h) + Cm-
(4.28)
The recurrence relations for Am^Bm-^Cni are ^ m + l = Bm + C'm, 5y7i_|-i = Am. + Sm, C'm+1 = 2^4^71,
(4.29)
a n d ^ i = l,J5i = 0 , C i = 0 , ^ 2 = 0 , ^ 2 = l,C2 = 2,>l3 = 3, Bg = 1,0.3 =
0. The expressions of the coefficients are A,„ = ^
+ ^ a — ^
+ ( - 1 ) - 3 ^ ( 1 + ar-\m
> 3, (4.30)
The polar 5-diniensional cosexponential a- 1 5
5 2^ 5
,771 — 3
functions
155
_oa + 2 (l-^a^-^m>3, (-1) m-3
(4.31)
4 ~ 6a m-4 . ( ]^\m-4^Q + ^ ^ / i , ^\m-4 ( l + a ^ - ^ m > 4,(4.32) 5 5
where a is a solution of the equation a^ + a — 1 = 0. Substituting the expressions of Am,Bm,Cm from Eqs. (4.30)-(4.32) in Eq. (4.26) and grouping the terms yields 5
5 +{hi +h4) +{h2 + /is)
5 5
r
2y
a - h i -(l+a)?/ a 5 5 a + 1 e«v 4. ^e-(»+o)j/ 5 5
(4.33)
The odd powers of /ii — /i4 have the form {hx - /i4)^'"+' = Dmihi - /14) + Em{h2 -
(4.34)
h).
The recurrence relations for Dm, Em are Dm+l
(4.35)
= ~ 3 D m — Em, £'m+l = " ^ ' m ~ 2£?^,
and Di = —3,Ei = —1,D2 = 10, E2 = 5. The expressions of the coefficients are Dm = {b+ l)6'"-i + (-l)"'-2(6 + 4)(5 + br-\m E,
6 + 2"
> 1,
+^-^^i^^^-\m>l,
(4.36) (4.37)
where 6 is a solution of the equation 6^ + 56 + 5 = 0. The even powers of hi — h^ have the form (hi - /14)''" = Fm{hx + hi) + Gm{h2 + /la) + Hm.
(4.38)
The recurrence relations for Ffji^Gyn^Hyyi are
(4.39) and Fi = 0,Gi = 1,/fi = - 2 , F 2 = l,G2 = -4,/f2 = 6. The expressions of the coefficients are J- m —
1
5(6 + 2)
b"' + (-l)'"-i-A±i-(5 + 6)"^,m>l, 5(6 + 2)
(4.40)
156
Complex Numbers in 5 Dimensions Gm —
5(6 + 2)
^^
1 (5 + 6 ) ' " , m > 1, 5(6 + 2)
'
^-=-|^'''""-^(-^)i(^+^)'"'"^^^'
(4.41)
(4.42)
where 6 is a solution of the equation 6-^ + 56 + 5 = 0. Substituting the expressions of Dm, Em, Fm, Gm, Hm from Eqs. (4.36)(4.37) and (4.40)-(4.42)in Eq. (4.27) and grouping the terms yields g(/M-/«4)!/ = 1 + ? cos(^/^y) + ? cos(\/5T6y) 5 5 ' 5 p— cos(V-6y) H — cos(v5 + by) 0 o o 1 ^ + 2 , /-T X '' + 3 , /s—r X +{h2 + h^) - H r— cos( V-6y) — cos( V5 + by) +{hi +/i4)
0
0
o
+ ( / l l - / l 4 ) —p— sm{\f^)y) 0
26 + 5 +(/i2 - /J3)
+ -j== sm(v/5 + 6y) y—bo
sin{\f~by) + -7=== sin(\/5 + fey) V-56
(4.43)
On the other hand, e^^*^ can be written with the aid of the 5-dimensional polar cosexponential functions as .2/112/
_
^5o(2y) + /ii^5i{2y) + /i2^52(2y) + /i3553(2y) + /i4ff54(2y).(4.44)
The multiphcation of the expressions of e^^^^^^^^ and e^'^*"^^)^ in Eqs. (4.33) and (4.43) and the separation of the real components yields the expressions of the 5-dimensional cosexponential functions, for a = (\/5 l)/2,6 = - ( 5 + v/5)/2, as 9BO
(2y) = le^y + le^y c o s ( / ^ y ) + ^e-^^^^)^ cos{VbTby),
95i(2y) = ^ e ' ^ + pe«^
+ ^e
-{l+a)y
(4.45)
-l V^ sm( ^ f vf-r —+ v/5 cos(f v/—r - o y )s H, ^ +y==- o y )\
l + v/5
cos(\/5 + 6y) + 4/—7 sin(\/5 + 6y) (4.46)
The polar 5-dimensional cosexponential
- 1 + N/5
-(l+a)3/
•1 +
, /^—r . cos(v5 + by)
1 + ^5
953(2y) = le^' + ^e"^ , i
157
1 +v/5 , / - r X / 5 . , A-T ^ — c o s ( V - b y ) + W —sm(V-fcy)
552(2?/) = i e 2 , ^ l^ay 4.1e-{i+a)j/ ^5
functions
N/5
^5
5 + \/5 . / - — r • p=r sin( v 5 + by) 2V-b (4.47)
cos(v—fej/)
cos(\/5 + by) +
5 + \/5
sin(v-6t/) sin(\/5 + 6y) (4.48)
554(2y) = ^ e 2 y ^ l g a j ,
4.ip-(l+a)y
^5
-1 + \/5
1 + v^
^-v . 5 + v/5 cos(v-cw/) — 2V-b ^—^
cos{\/5 + by) - i—--sin{V5
/-jsm(v-by) +by) (4.49)
The polar 5-dimensional cosexponential functions can be written as 1 4 . /2nl\ s m ( ^-^) —j 95k{y) = c H ^ x p ycos ( ^ — j J c o s [ yysm\
A; = 0,...,4.
2nkl — (4.50)
The graphs of the polar 5-dimensional cosexponential functions are shown in Fig. 4.2. It can be checked that
E dliy) - U^"" + |e(^-i)!'/2 + ^e-(^+i)2'/2.
(4.51)
A;=0
The addition theorems for the polar 5-dimensional cosexponential functions are
158
Complex Numbers in 5 Dimensions
>.
Figure 4.2: Polar cosexponential functions gmiOm^95219531954-
Exponential and trigonometric forms of polar d-complex numbers
9bo{y -i- z) = gbo{y)9bo{z) 4- gbi{y)9b4{z) + •^953{y)9b2{z) +95i{y)9bl{z), 9b\{y -f ^) = 9m{y)9b\{z) + 9b\[y)9bo{z) + +953{y)9b3{z) +9bA{y)9b2{z), 9b2{y + z) = 9bo{y)9b2{z) + 95i{y)95i{z) + +9b3{y)9bA{z) + 9b4{y)9b3{z), 553(2/ + z)= 9bo{y)9b3{z) + 9bi{y)9b2{z) + +9b3{y)9bo{z) + gbi{y)gbA{z), 9bA{y ^z) = 9bo{y)9b4{z) + 9bi{y)gb3{z) + +9b3{y)9bi{z) + 9bi{y)9bo{z)>
159
952{y)953{z) 952{y)9b4{z) 9b2iy)9bo{z) 9b2[y)9bi[z) gb2{y)9b2{z) (4.52)
It can b e shown t h a t {9bo{y) + higsiiy) + h2gb2{y) + h39b3{y) + /i4554(y)}' = 9bo{ly) + hi9bi{ly) + h2gb2(ly) + h^9b3{ly) + h^gb^ily), {95o(y) + hig^^iy) + /i295i(y) + hsgbiiy) + h49b2{y)y = 9bo{ly) + hig^sily) + /i2ff5i(iy) + hgbiily) + h^gb2{ly), {ff5o(y) + hig^2[y) + M54(y) + Hgb\{y) + M53(y)}' = 9b^{ly) + hi9^2{ly) + h2gbA{ly) + hsg^iily) + h^g^^{ly), {9b^{y) + higbAiy) + h2gb3{y) + hzgb2{y) + h^gb\{y)y = 9b^{ly) + higu{ly) + h2gb3{ly) + h39b2{ly) + h^gbiily)(4.53) T h e derivatives of t h e polar cosexponential functions are related by 9bA,
du
4.4
dg '51 ^553 ^554 dgb2 = 550, du = 551, du = 552, du du
553-
(4.54)
Exponential and trigonometric forms of polar 5-complex numbers
The exponential and trigonometric forms of 5-complex numbers can be expressed with the aid of the hypercomplex bases
ei V 62 /
1 / 1 2 f 2 1 P 0 Q 1 2p2-l 2pq
\o
1 2
1 2
1 2
2p2-l 2pq
2p2-l -2pq
P -9
P
P -9 2p2_l
Q
-2pq
/
1 \ hi h2
\h4 J (4.55)
160
Complex Numbers in 5 Dimensions The multiplication relations for these bases are e^ = 64., e^€k = 0, e^€k = 0, 4 = ^A:, 4 = -^k^ ^kh = h^ ekCi = 0, eA;e/ = 0, ekCi = 0, A;,/= 1,2, A; 7^/. (4.56)
The bases have the property that e + + e i +62 = 1.
(4.57)
The moduli of the new bases are 1 |e^l = ^ ,
. le.l
vl' l^^^l = vl'
(4.58)
for fc = 1,2. It can be checked that xo + h\X\ + h2X2 + /i3^3 + h^XA = e+?;-f + eii;i + e]Vi + 62^2 + ^2^2(4.59) The ensemble e-|.,ei,61,62,62 will be called the canonical 5-complex base, and Eq. (4.59) gives the canonical form of the 5-complex number. The exponential form of the 5-complex number n is
ii
u — pexp < -{h\ + /i2 + /i3 + h^) In
+
10
N/2
{hi + h^) - ^^——{h2 + h) lntani/)i 10 (4.60)
for 0 < 6»+ < 7r/2. The trigonometric form of the 5-complex mimber u is
"=^^'J) ( ^
-1/2
+1+
62 e+ v/2 + 61 + tan^-i tan t/^i I
exp(ei(/)i +6202).
(4.61)
The modulus d and the amplitude p are related by 1/2 d = p—F^ (tan9^ t'du ipi j
I
tan^ OJ
+ 1 + tan'^ ^l)l J
(4.62)
Elementary functions of a polar S-complex variable
4.5
161
Elementary functions of a polar S-complex variable
The logarithm and power function exist for v^ > 0, which means that 0 < 0-f- < 7r/2, and are given by 1 \/2 Intx = Inp + -{hi + h2 + hs + h^) In 7——5 tan 0^
+
v/5 + 1 , , 10
, ,
\ / 5 - 1,, 10
In tan ^1 (4.63)
n ^ = e^v^ + p^{ei cosmcpi + e\ sinm)i) 4-p2^(e2C0sm>2 + e2sinm02).
(4.64)
The exponential of the 5-complex variable u is e^ = e+e^"^ + e^^ (ei cos?)i + ei sin{;i) + e^'^ (e2Cos^2 + e2sin{;2). (4.65) The trigonometric functions of the 5-complex variable u are 2
COS!/ = 6-1- cosi;+ 4- ^
(cjt cost;^cosh-O^ — e^sini;^sinhi;jk) ?
(4.66)
2
sinti = 64- sin^+ + ^Z (^^ ^^'^'^^ ^^^^^^ "^ ^^ cos i;jt sinh Vk).
(4.67)
The hyperbolic functions of the 5-complex variable u are cosh w = e^- cosh?;^. + ^
(e^ cosht^/^ cosi;^ -f Cjt sinh?;/^ sini;^), (4.68)
2
sinh u = e^ sinh t;^ + 2 J (^ifc sinh ?;jt cos v^ + e^ cosh i^/t sin Vk). (4.69) k=:l
4.6
Power series of 5-complex numbers
A power series of the 5-complex variable u is a series of the form ao + ci\u + a2U^ H
h a/w' H
.
(4.70)
162
Complex Numbers in 5 Dimensions
Since \au^\ < 5'/2|a||u|',
(4.71)
the series is absolutely convergent for \u\ < c,
(4.72)
where \ai\
c = lim
(4.73)
If «/ = Ep=o hpaip, where /IQ = 1, and 4
(4.74) p=0
(4.75) p=o 4
V ^
/
/27rfcp\
(4.76)
for A; = 1,2, the series (4.70) can be written as 2
E /=0
e^Ai^v^^ + Y^{ekAik + ekAik){ekVk +
hh)
(4.77)
k:=l
The series in Eq. (4.70) is absolutely convergent for (4.78)
\v^\ < c+, pk
1/2'
\A+l,k + Al+l,k}
(4.79)
Analytic functions of a polar 5-complex variable
4.7
163
Analytic functions of a polar S-complex variable
I f / ( ^ ) = Ilk^ohkPk{xo,xuX2,x^,X4),
then
dPo dxo
dPi dxi
dP2 dx2
dPs dxz
dPi 8x4
(4.80)
dPs dx2
dPi
dxo
dP2 dx\
8P0 8x4
(4.81)
dPi 3x2
dPo
dxo
dP3 dx\
dPi dxi
(4.82)
dPz^ dxo
dP^ dxi
dPo 8x2
dPx dx3
dP2
dPj _ dPo dxo dx\
dPi 8x2
dP2 dxz
dP3
dX3
dX3
(4.83)
dX4
(4.84)
dX4
and d'Pk dxodxi
d'Pk dx[i/2]dxi_[if2]
d^Pk dxidxi^i d'Pk 8x1^18x4
d^Pk 9^/+H-[(3-0/2]^^4-[(3-0/2] ' (4.85)
d^Pk 8xi^2dxs
for fc,/ = 0, ...,4. In Eq. (4.85), [a] denotes the integer part of a, defined as [a] < a < [a] + 1. In this chapter, brackets larger than the regular brackets [ ] do not have the meaning of integer part.
4.8
Integrals of polar S-complex functions
If f{u) is an analytic 5-complex function, then /
f(u)du
r»
/•/
\ r-
.
/
T-.
\
-
'
/
T^
\i
(4.86) where int(M,C) =
1 if M is an interior point of C, 0 if M is exterior to C,
(4.87)
and uo^^Tfk^ ^^kVk ^^^ respectively the projections of the pole no and of the loop r on the plane defined by the axes ^i^ and r/^, fc = 1,2.
164
Complex Numbers in 5 Dimensions
4.9
Factorization of polar 5-complex polynomials
A polynomial of degree rn of the 5-complex variable u has the form Pm{u) = ii"^ + aiii'"'-^ 4- • • • + arn-iu + a,^,
(4.88)
where ai, for / = l,...,m, are 5-complex constants. If a/ = Ylp=ohpaip, and with the notations of Eqs. (4.74)-(4.76) applied for / = 1,- • • ,m, the polynomial Pm{u) can be written as
l=\
{ekVk + eki^k)"^ + Yli^^kAik 4- ekAik){ekVk + 4%)"^ ' | (4.89) The polynomial Pm{u) can be written, as m
Pm{u)^ll{u~up),
(4.90)
p=\
where Up = e^Vp^ + (eivip -h eivip) -f {e2V2p-^ hhp) ,P = l,--.,rn.
(4.91)
The quantities Vp^, ekVkp + efci}fcp, p = 1,..., m, A; = 1,2, are the roots of the corresponding polynomial in Eq. (4.89). The roots Vp^ appear in complexconjugate pairs, and iJkp^ Vkp are real numbers. Since all these roots may be ordered arbitrarily, the polynomial Pm{u) can be written in many different ways as a product of linear factors. If P{u) = u^ — l^ the degree is 7n = 2, the coefficients of the polynomial are ai = 0,a2 = - 1 , the coefficients defined in Eqs. (4.74)-(4.76) are ^2+ = - l ^ ^21 = - 1 , ^2] = 0, A22 = - 1 , ^22 = 0- The expression of P(w), Eq. (4.89), is v\ - e+ + (ei^i + eit;i)^ - ei + (e2t;2 + 62^2)'^ - e2. The factorization of P{u), Eq. (4,90), is F(w) = {u - u\){u - ti2)j where the roots are ni = ±e^ ± e i ± e2,1/2 = -^^i- If e^., ei, 62 are expressed with the aid of Eq. (4.55) in terms of h\,h2,h?,Ji\^ the factorizations of P{u) are
Representation of polar 5-coinplex numbers by irreducible matrices
165
obtained as u 2 - l = {u + l)(u-l), M2 - 1 = [u + i + 4 ^ 1 (/li + h4) - ^ ( / t 2 + hs)] u2 _ 1 = [„ + i - ^ { h i + h4) + 4 ^ ( / i 2 + ha)]
h - 5 + 4^(/»l + h4) - 4^(/t2 + /i3)] ,
(4.92) It can be checked that {±e+ ±ei±
4.10
€2)^ = e+ + ei + 62 = 1.
Representation of polar S-complex numbers by irreducible matrices
If the unitary matrix which can be obtained from the expression, Eq. (4.8), of the variables ^+,^\,r)i,^^,% in terms of xo,xi, x^,X3,X4 is called T, the irreducible representation [7] of the hypercomplex number u is TUT-^
=
0
Vi
0
,
where U is the matrix in Eq. (4.22), and V^ are the matrices
(4.93)
This Page Intentionally Left Blank
Chapter 5
Complex N u m b e r s in 6 Dimensions Two distinct systems of commutative complex numbers in 6 dimensions having the form u = XQ -I- /ii^i 4- /i2^2 + ^3^3 + /14X4 4- h^x^ are described in this chapter, for which the multiphcation is associative and commutative, where the variables xo,xi,X2,X3,a:4,X5 are real numbers. The first type of 6-complex numbers described in this article is characterized by the presence of two polar axes, so that these numbers will be called polar 6-complex numbers. The other type of 6-complex numbers described in this paper will be called planar n-complex numbers. The polar 6-complex numbers introduced in this chapter can be specified by the modulus d, the amplitude p, and the polar angles 0-f,0_, the planar angle ^ 1 , and the azimuthal angles (/)i,(/>2. The planar 6-complex numbers introduced in this paper can be specified by the modulus d, the amplitude p, the planar angles V^i,V'2^ ^nd the azimuthal angles <^i,>2^>3Exponential and trigonometric forms are given for the 6-complex numbers. The 6-complex functions defined by series of powers are analytic, and the partial derivatives of the components of the 6-complex functions are closely related. The integrals of polar 6-complex functions are independent of path in regions where the functions are regular. The fact that the exponential form of ther 6-complex numbers depends on cyclic variables leads to the concept of pole and residue for integrals on closed paths. The polynomials of polar 6-complex variables can be written as products of linear or quadratic factors, the polynomials of planar 6-complex variables can always be written as products of linear factors, although the factorization is not unique. The polar 6-complex numbers described in this paper are a particular
167
168
Complex Numbers in 6 Dimensions
case forn = 6 of the polar hypercomplex numbers in n dimensions discussed in Sec. 6.1, and the planar 6-coniplex numbers described in this section are a particular case for n = 6 of the planar hypercomplex numbers in n dimensions discussed in Sec. 6.2.
5.1
Polar complex numbers in 6 dimensions
5.1.1
Operations with polar complex numbers in 6 dimensions
The polar hypercomplex number ?/ in 6 dimensions is represented as u = XQ -\- hixi -f h2X2 -f- hsx:^ + h4X4 + h^x^. The multiplication rules for the bases h\,h2,h:^,h4,h^
(5.1) are
h\ = /i2, h^ = /i4, h^ = 1, /i4 = /i2, hi = /i4, hih2 = /13, hih^ — h^, h\h4 — /i5, /11/15 = 1, /i2/^3 = ^5? h2hu\ ~ 1, /i2^5 — h\^ h^h/i = h\, h:ihrj — /i2, h\h^ = h-^,
(5.2)
The significance of the composition laws in Eq. (5.2) can be understood by representing the bases /IJ,/IA; by points on a circle at the angles aj = 7ri/3,aA: = 7r/i:/3, as shown in Fig. 5.1, and the product h^h^ by the point of the circle at the angle 7r(j + A:)/3. If 27r < 7r(j -f /i;)/3 < 47r, the point represents the basis hi of angle a/ = 7r(j -h fc)/3 — 27r. The sum of the 6-complex numbers u and w! is 11 + 7/ = J;Q + j;[j + /?^j (:7;^ "f ^ 1 ) + '^1 (^2 + •'^2) + ^3(.'Z^3 + ^'3)
-^h4{x4 -f x'^) + H{xf, 4- 4 ) -
(5-3)
The product of the numbers u, n' is W = x^x!^ 4- a:i.X5 + a:2.T4 4- 0:30:3 -f- 0:40:2 4- x^x\ 4-/11 (0:00:1 4- O:IO:Q -f 0:2X5 + 0:30:4 -h 0:40:3 4- .0:50:2)
4-/i2(^o4 + ^ i 4 + ^20:0 4- .^3X5 4- 0:4X4 4- 0:50:3) 4-/14(0:04 + ^"i4 + ^20:2 + -^s^ + ^40:0 4- 0:50:5)
,
.
169
Polar complex numbers in 6 dimensions
Figure 5.1: Representation of the polar hypercomplex bases l,/ii,/i2^^3^^4,^5 by points on a circle at the angles ak = 27rfc/6. The product hjhk will be represented by the point of the circle at the angle 2n{j + fc)/6, i, fc = 0,1,..., 5, where ho = 1. If 27r < 27v{j -f fc)/6 < 47r, the point represents the basis hi of angle ai = 2TT{J + k)/Q — 27r. The relation between the variables v^^v-, i^i, Ci, V2, V2 sind XQ, xi, 0:2, X3, X4^xs are VVi
\ V2 J
( 1 1 1 -1 1 21 0 1 2
Vo
2
1 1 1 2
2 2
1 -1 -1 0 1 0
1 1 1 2
_ ^ !
3 2
1
\
~1 1 2 v/3
_\_ v/3 ; 2 /
/ ^0
\
2:2
(5.5)
^3
V ^5 /
The other variables are v\ = t^2?^4 = —^25 ^^5 = ^1,1^5 = —tli. The variables i;_^.,'u_,t;i,{)i,t;2^'S2 will be called canonical polar 6-complex variables.
5.1.2
Geometric representation of polar complex numbers in 6 dimensions
The 6-complex number u = J^Q + h\X\ + h^x^ + ^13X3 + h\X\ + h^x^ is represented by the point A of coordinates (XQ, 0:1,0:2,0:3,0:4, X5). The distance from the origin O of the 6-dimensional space to the point A has the
170
Complex Numbers in 6 Dimensions
expression (f = xl + X? -f x^ -f x^ + x^ -f xl.
(5.6)
The distance d is called modulus of the 6-complex number u, and is designated by d = \u\. The modulus has the property that \u'u"\ < V6\u'\\u"\
(5.7)
The exponential and trigonometric forms of the 6-complex number u can be obtained conveniently in a rotated system of axes defined by a transformation which has the form (^^\
'
v^
v/6
3
6
0
i
e6
^~i
6
V 6J
1
1
v/6
>/6
Te
~76
6
3
y/3 3
\
0
6
f
h
2
3
0
1 v/6 I v^
I
i/3
a^2
(5.8)
:i "A 2
2
The lines of the matrices in Eq. (5.8) gives the components of the 6 basis vectors of the new system of axes. These vectors have unit length and are orthogonal to each other. The relations between the two sets of variables are v^ - \/6^4.,i;_ = \f^i^,Vk
= \f^ik^^k = \/^'nk,k = 1,2.
(5.9)
The radius p^ and the azimuthal angle (/)jt in the plane of the axes Vk, Vk are Pk^'^l-^
^h cos (pk = Vk/pk, sin(f)k = ^k/pki 0 <(f)k <27r, A; = 1,2, (5.10)
so that there are 2 azimuthal angles. The planar angle ipi is tanV^i = /9i//92,0 < ^1 < 7r/2.
(5.11)
There is a polar angle 6^, tan ^4
v/2pi ~,0<(94. <7r,
(5.12)
^^+
and there is also a polar angle 9-,, tan^_ = ^ ^ , 0 < 6 > _ <7r.
(5.13)
Polar complex numbers in 6 dimensions
171
The amplitude of a 6-complex number u is p = {^V+V-pfp^)
(5.14)
.
It can be checked that
d^ = l^l + ^vl + ^{pl + PI).
(5.15)
li u = u'u", the parameters of the hypercomplex numbers are related by (5.16) 1 tan ^+tan ^", v/2
taxi 9
(5.17)
u_ — v'_v"_.
(5.18)
tan 9- = —= tan 9' tan 9",
(5.19)
tanV*! = tan^'i tan«/;'/,
(5.20)
Pk =
(5.21)
p'kP'L
(t>k = 4>'k + 4>'k,
(5.22)
Vk = v'kV',^ - v'kV'k^ Vk = v'kV'l + VfcUfc,
(5.23)
P = P'p".
(5.24)
where k = 1,2. The 6-complex number ti = XQ + hix\ + /i2^2 + ^3^3 + ^4^4 + ^53^5 can be represented by the matrix / 0^0
U
\
Xs
X4
X5 \
X\
X2
X^
XQ
Xi
X2
X3
X4
X4
X^
Xo
Xi
X2
X^
Xs
X4
.X5
Xo
Xi
X2
X2
Xz
X4
X'5 Xo
X\
Xi
X2
X3
X4
Xo /
X^
(5.25)
The product u — u'u" is represented by the matrix multiphcation U = UV".
172
Complex Numbers in 6 Dimensions
5.1.3
The polar 6-dimensional cosexponential functions
The polar cosexponential functions in 6 dimensions are
(5.26) p=0
for fc = 0,..., 5. The polar cosexponential functions g^/^ of even index k are even functions, ge^2p(^y) = 96,2p{y)^ and the polar cosexponential functions of odd index k are odd functions, gcy,2p+\{-y) = -~.96,2p+i(y)5 P = 0,1,2. It can be checked that
Yldekiv) =e^,
(5.27)
k=0
E(-l)'56)i.(?y) = e-
(5.28)
A:=0
The exponential function of the quantity hf^y is e^^y e^'^ e^^y g/i42/ e^'''^
= geo(y) + hgeiiy) + hig^iiy) -i- hsge^iy) + hgeiiy) + hr,g^r^(y), = 9m{y) + 9m{y) + h2{gm{y) + .964 (y)} + /M{S'62(|/) + Qmiy)}, = 960 (y) + 962 (y) + 964 (y) + /i3{96l(y) + 963 (l/) + 965(y)}, :^ ,900(i/) + 963(y) + h2{gm{y) + 965(y)} -f /M{96I (y) + 964(y)}, = 96o(y) + h\g(sfy{y) + h2g6i{y) + h^9m{y) + fugiyiiy) 4- Hg^iiy).
(5.29) The relations for /i2 and /^ can be written equivalently as e^^^ = ^730 + /i293i + h^gyi^e^'^y = r/30 + /^2932 + /i493ii and the relation for /13 can l)e written as el^^^ = (/20 + '^3921 ^ which is the same as e}^^^ = cosh y -f ^3 sinh y. The expressions of the polar 6-dinicnsional cosexponential functions are 9m{y) = \ (*osh y + I cosh | cos ^ 1 / , 301 (y) = \ sinh y + ^ sinh | cos ^ y + ^ cosh | sin ^?y, .962 (y) 963(y) = 964 (y) = 965(y) =
v/3.
I cosh y - I cosh I cos "x?^ + 3 '"^"^'^ I ^^^ Y'V^ | s i n h y - 3 Sinh ^ cos -r-y^ \ cosh y - ^ cosh | cos ^ y - ^ sinh | sin ^ y , ^ shihy + ^ sinh f cos ^ y - ^ cosh | sin ^ y .
(5.30)
The cosexponential ftnictions (5.30) can be written as 9cyk{y) = ^ 5 Z « x p ycosf
T)
. /27r/\ COS
2Trkl
(5.31)
Polar complex numbers in 6 dimensions
173
for k = 0, ...,5. The graphs of the polar 6'dimensional cosexponential functions are shown in Fig, 5.2. It can be checked that Yl dlkiy) = ^ cosh2y + - coshy. ^ ^
(5.32)
A;=0 k=o
The ^he£addition theorems for the polar 6-dimensional cosexponential functions are .960(y + Z)=: 960iy)960{z) + 9ei{y)965{z)
+ 962{y)964{z) + 963iy)963{z)
+9e4{y)962{z) +965iy)9ei{z), 96i{y + z) = 96o{y)96i{z) + 96iiy)96Q{z) + 962{y)9es{z) + 963{y)964{z) -^964{y)963{z) +9e5{y)9e2{z), 962{y + z) = 96o{y)962{z) + 96i{y)9ei{z) -f-9e2iy)96o{z) + 96s{y)965{z) +9e4{y)964{z) +965{y)9e3{z), 963{y -f- ^) = 96o{y)963{z) -^ 96i{y)962{z) + 9e2(y)9ei{z) -^ 9e3{y)96o{z) +964{y)965{z) +965iy)964{z), 964{y + z) := 96o{y)964{z) + g6i{y)g63{z) + 9G2{y)962{z) + 963{y)9eiiz) +964{y)960{z) + 965{y)965{z), 965 (J/ + Z) = 960{y)965{z) + 96l{y)9Q4{z) + 962{y)963{z) + 563 (2/).962 (^) +964{y)9ei{z) -h965{y)96o{z)' (5.33) It can be shown that {9Qo{y) + hi9Q\{y) + h2962{y) + h:sg^^{y) + h^gQ^iy) + h^geb{y)y
= gMh) + hig6i{ly) + M62('y) + h^gmily) + h^g^ily) +h5g65{ly), {960(y) -f 963(y) + /i2{96i(y) + 964(y)} + ^^4{962(9) -f 965(y)}}' = geoily) + 963(^y) + h2{g^i{ly) -f 964('y)} -f tnigeiily) +965('y)}, {960 (y) + 962 (y) + 964 (y) + /i3{96i (y) + 963 (y) + 965 (y)}}' = 960 (^y) + 962 (^y) + 964 (^y) + /i3{96i(/y) + 963 (^y) +965 (^y)}, {960 (y) + 963 (y) + /i2{962(y) + 965 (y)} + /M{96i(y) + 964 (y)}}' = gmily) + gmii'y) + h2{gm{hj) + gebily)} + /M{96i(^y) +964 (^y)}, {96o(y) + /H965(y) + M 6 4 ( y ) + ^3963(9) + ^4962(y) + /V396i(y)}' = 960 (^y) + /ii965(^y) + /i2964(^y) + ^3963 (^y) + /M962(^y)
(5.34)
174
Complex Numbers in 6 Dimensions
Figure 5.2: Polar cosexponential functions gc^o,56i,5'62i^63,ff64,965-
Polar complex
numbers
in 6
175
dimensions
T h e derivatives of t h e polar cosexponential functions are related by dgeo dg%2 dg^'61 — 965, = 360, du du du dg['65
du 5.1.4
561,
dg^z du
362,
dgi'64
du
= 963,
(5.35)
==ge4'
Exponential and trigonometric forms of polar 6-complex numbers
T h e exponential and trigonometric forms of polar 6-complex numbers can be expressed with the aid of the hypercomplex bases
e_ ei ei 62
(^
V 62 /
^0
s 3
1
1
1
1
\
f
f
s
i'
^
S ?
'i
1^
0
h
1 3
6
S
\
3
0 6_
1 3
6
0
N/3
6
1
f
6
V3 6
f
/
1 \ hi h2 hi
(5.36)
6
_v5 J 6
/
The multiplication relations for these bases are e\ = e^, e^ = e_, e+e_ = 0, e+e^ = 0, e^e^ = 0, e-e^ = 0, e-ek = 0, e^ = e^, el = - e ^ , 6)^6^ = 4 , ej^e/ = 0, e^e^ = 0, Ckii = 0 , A;,/ = l,2,fc 7^ /
(5.37)
The bases have the property that (5.38)
e-f. + e_ + e i + e2 = 1. The moduli of the new bases are 1 ^ ,
1 M = - ^ , | e . | = - ^ , f c = l,2.
(5.39)
It can be shown that Xo + hiXi+
h2X2 + ^3X3 + /I4X4 4- h5X5
= e^v+ + e_w_ + eivi + ejui + e2U2 + hv2-
(5.40)
The ensemble e+, e_, ei, ei, 62,62 will be called the canonical polar 6-complex base, and Eq. (5.40) gives the canonical form of the polar 6-complex number.
176
Complex Numbers in 6 Dimensions The exponential form of the G-coniplex number n is f1 \/2 u = pexp I 6-(hi + /i2 + h:i + /14 -Ih^) hi ' tan 6^ 1 \/2 —6^(/ii — /^2 + '^3 — hi +' h^) in tan^_ + ~(/ii +/i2 ~ 2/13 -h h^ -f /t5) In tan-01 +e](/)i +6202 [ ,
(5.41)
for 0 < ^-^ < 7r/2,0 < 6/„ < 7r/2. The trigonometric form of the 6-complex number u is / I I I \-i/2 \ tan"^ OJ^
i'cin^ 9-
tan"^ ^i /
/e4.\/2 e_\/2 ^2 \ , r ^ + T—r + ^1 + r - V ^^P ^^^^ + ^^2>2). ytan0_f. tan^_ tani/jiy The modulus d and the amplitude p are related by
(5.42)
d = /9—-pr (tan6^ tan^_ tan ?/?! j /
1 1 1 \^/^^ TIT + 77^ + 1 + 9-r • Vtan^e^ tan^6^_ tan^i/^i/ 5.1.5
(5-43) ^ '
E l e m e n t a r y functions of a polar 6 - c o m p l e x variable
The logarithm and power functions of the 6-complex rmmber u exist for Vj^ > 0, ^;_ > 0, which means that 0 < 0+ < 7r/2,0 < 0_ < 7r/2, and are given by 1 \/2 Inw = Inp + -(/ii + /i2 -i- ^3 + /i4 + /i5) In tan 9 1 v/2 - -6( / i i - /i2 -I- /i3 - /i4 + ^5) In t a n ^ -
+
+ -{hi +h2 - 2 / i 3 + /i4 + /i5)lntanV^i + e ] 0 i + hh^ o
(5.44)
ix"^ = e^v'^ + e-v^ + p^\e\ cosmcpi + e\ shim(t)i) +P2^{e2 cos m(f)2 + e2 sin m(j)2).
(5.45)
The exponential of the 6-complex variable u is e^^ — e+e^'"^ + e_e^^~ -f- r^^^^ (ei cos?)] 4- e\ ^\nv\) -he^"^ (^2 cos V2 + ^2 sinf;2).
(5.46)
Polar complex numbers in 6 dimensions
177
The trigonometric functions of the 6-complex variable u are cos u = e^ cos V-^ -f e_ cos v2
+ ^
(efc cos Vk cosh Vk - 4 sini;/^ sinhi;^;),
(5.47)
sin u = e^ sin i^-f + e_ sin v^ 2
+ ^
{ck sin ^fc cosh f;^ -f ^fc cos ?;fc sinh v^) •
(5.48)
The hyperbolic functions of the 6-complex variable u are cosh u = e^ cosh v^ -\- e_ cosh v2
4- ^
(efc coshtiA; cosi}^; + 4 sinht;/. sini^A;),
(5.49)
sinhu = 64. sinhi;-!- + e_ sinhi?_ 2
+ ^
(efc sinht;^ cost}^: + ^ik coshi;^: sin?}/j;).
(5.50)
k=l
5.1.6
Power series of polar 6-complex numbers
A power series of the 6-complex variable u is a series of the form tto + CLiu -f a2u'^ H
h aiu^ H
.
(5.51)
Since
\au^\
(5.52)
the series is absolutely convergent for \u\ < c,
(5.53)
c = l i m - J ^ .
(5.54)
where
If tt/ = X]p=o f^p(^lp^ where ho =" I, and 5
Ai^ = Y^aip, p=0
(5.55)
178
Complex Numbers in 6 Dimensions
(5.56)
Mk = 2^0^/^ c o s — - ,
(5.57)
(5.58) for fe = 1,2, the series (5.51) can be written as 2
^4-^+^4 + e_^;_?;'_ + J^ickAik + ekAik){ekVk + f=0
hhY .(5.59)
k=i
The series in Eq. (5.51) is absolutely convergent for \v^\
= 1,2,
(5.60)
where c+ = lim •—
—, c_ = hm 1/2
Ck =
5.1.7
, A; = 1 , 2 .
lim
1/2'
(5.61)
Analytic functions of a polar 6-complex variable
I f / ( « ) = El=o^/feA;(a;o,2;i,X2,a;3,a;4,X5), then 5Fo
C>Pi
dP2
dPs
dxo
dx\
dX2
dX3
dP4 dx4
dP5
dPi
dP2
dPz
dxo
dx\
OP^
OPo
dX2
dPA dx:i
0X4
Ox<s'
dP2 dxo
dPz
dPA
dP5
OPo
dxi
0X2
0X3
0X4
OPi 0x5'
dPs dxo
dPi
dP'o
OPo
OPx
OP2
dx\
8X2
0X3
0X4
5.T5'
dP4 _ ^ OXQ
_dPo OX I 0x2
_m
9X5'
_ OP2 _ 0P3 0x3 0x4 Ox^'
(5.62) (5.63) (5.64) (5.65) (5.66)
Polar complex numbers in 6 dimensions dPr, dxo
dPo dx\
dPi dx2
dP2 dxz
dPs 8x4
179 dP^ dx^'
(5.67)
and d^Pk dxodxi
d^Pk dxidxi^i d^Pk dxi+idx^
d^Pk ^^[//2]^^/-[i/2] d^Pk dxi^2dx4
d^Pk dxi^i^(^4^i)/2]dx^-[(4-i)/2]' (5.68)
for fc, / = 0,..., 5. In Eq. (5.68), [a] denotes the integer part of a, defined as [a] < a < [a] + 1. In this work, brackets larger than the regular brackets [ ] do not have the meaning of integer part. 5.1.8
Integrals of polar 6 - c o m p l e x functions
If f{u) is an analytic 6-complex function, then f
= 27r/(wo) [ei int(no^i^i,r^i^j) + 62 irit{uo^.^r)2^r^^r}2)], (5.69)
where • WTi^ x-rx I X *x M IS an mtenor mt(M,C) = < T A^ • ^ • ^ \ n0 if V ' ^ \ ri ^^ M IS exterior to and uo^^r}k ^^^ ^U^Vk ^^^ respectively the the loop r on the plane defined by the axes 5.1.9
pomt of C, /-. ^ C, projections of the pole (^^ ^.nd %, fc = 1,2.
.^ „ ^ . (5.70) ^ ^ uo and of
F a c t o r i z a t i o n of polar 6 - c o m p l e x p o l y n o m i a l s
A polynomial of degree m of the polar 6-complex variable u has the form Pm{u) = ^ ^ + aiu"^-^ + • • • + am~iu + am,
(5.71)
where a;, for / = l,...,m, are 6-complex constants. If ai = Ylp=o^p^ip^ and with the notations of Eqs. (5.55)-(5.58) applied for / = 1, • • • ,m, the polynomial Pmi'^) can be written as Pm = e+ f < + E A + < - ' ) + e - (v^ + E 2 r + E ^^kVk + hvkT k=l
L
m + ^{ekAik
Ai_v^'^)
+ ejfc^Jfc)(ejkVfc + hhT'^
\
;=1
(5.72)
180
Complex Numbers in 6 Dimensions
where the constants Ai+,Ai^,Aik,Aik are real numbers. The polynomial Pmiu) can bo written, as Pm{u)
(5.73)
Up),
where Up = e^Vp-^ + e_i;p_ -|- {eivip -f eivip) + {e2V2p + e2V2p) ,p = 1, ...,m. (5.74)
The quantities Vp-^^ Vp-^ ^k'^kp + ^k^kp^ P — l,...,m,/j = 1,2, are the roots of the corresponding polynomial in Eq. (5.72). The roots Vp^, Vp- appear in complex-conjugate pairs, and v^p^Vf^p are real numbers. Since all these roots may be ordered arbitrarily, the polynomial Pmi'^^) can be written in many different ways as a product of linear factors. If P{u) = w"^ — 1, the degree is rn = 2, the coefficients of the polynomial are ai = 0, a2 = —1, the coefficients defined in Eqs. (5.55)-(5.58) are A2-\. = - 1 , ^ 2 - = -l,A2\ = -1,^21 = 0,^22 = -1,^22 = 0. The expression of P{u), Eq. (5.72), \sv\-e^-^vl.-e--^{eiV\+e]V\)'^-ei + {e2V2-\-e2V2)^-'e2^ The factorization of P(?/), Eq. (5.73), is P{u) = {u — ui){u~ti2), where the roots are ui = ±e^ ± e_ ± ei ± e2, U2 = —iii- If e_|_,e__,ei, 62 are expressed with the aid of Eq. (5.36) in terms of h], /i2, /i3, /^4, /15, the factorizations of P(u) are obtained as u^ -1 = (n + l ) ( 7 / - 1 ) , 1/2 - 1 = [n 4- ^(1 -f- /ii + /i2 - 2/z3 + hi + hr,) [u - i ( l 4- hi -F h2 - 2/i3 -f hi + /i5)] , u^ - 1 = [u + ^(1 - /ii + h2 -f- 2/i3 + hi - /15)] [ix - i ( l - /ii + /i2 + 2/13 + /^l - /15)] , - 1 = Jn + ^(2 -f- hi - /i2 + /i3 -
/M
-f ^5)]
[ti - 1(2 + hi - h2 + /i3 - hi + A5)] , u 2 - -1 =
-1 7 / 2 - -1 M2--1 TX2-
=
\ + i ( - l -f 2/^2 + 2/M)] [n - ^ ( - 1 + 2/12 + 2/^4)] , u + \{2hi - /i3 + 2/15)] [u - |(2/ii - /i3 4- 2/15)] ,
= (u
+ h:i){u - /i3). = \u + | ( - 2 + hi+ h-i + /i3 + hi + h)\
\u - ^ ( - 2 + hi + /«2 +
/M
+
/M
+ /is)] • (5.75)
It can be checked that (±e+ ± e_ ± e\ ± ej) = e+ + e_ + ei + 62 = 1-
Planar complex numbers in 6 dimensions 5.1.10
181
R e p r e s e n t a t i o n of polar 6 - c o m p l e x n u m b e r s b y irreducible m a t r i c e s
If the unitary matrix which appears in the expression, Eq. (5.8), of the variables ^ + , ^ - , ^ i , ^ i , ^k^Vk in terms of xo,x\,X2.,X3,X4,xs is called T, the irreducible representation of the hypercomplex number u is / u+ 0 TUT-^ = 0 V 0
0 w0 0
0 0 Fi 0
0 \ 0 0 V2 J
(5.76)
where U is the matrix in Eq. (5.25), and Vk are the matrices
5.2 5,2.1
Planar complex numbers in 6 dimensions O p e r a t i o n s w i t h planar c o m p l e x n u m b e r s in 6 dimensions
The planar hypercomplex number n in 6 dimensions is represented as u = XQ -\- hixi + h2X2 -f h^x^ -h hiXi 4- h^x^. T h e m u l t i p l i c a t i o n r u l e s for t h e b a s e s hi^h2^hs^h4^h^
(5.78) are
h\ = /l2, /l2 = /i4, h^ = 1, / i | = ~/i2? /^5 = ~-^4? hlh2 hih:i = /14, hih^
= /15, hih^
= /i3,
= — 1, /i2^3 = ^5? /^2/^4 = ~1?
^ 2 ^ 5 = ~-^i7 /i3'^4 == —hi^ hsh^
= —^2^ /i4/i5 = -~/i3.
(5.79)
The significance of the composition laws in Eq. (5.79) can be understood by representing the bases l,/ii,/i2i^3,/i4?/i5 by points on a circle at the angles a^ = 7rA;/6, as shown in Fig. 5.3. The product hjhk will be represented by the point of the circle at the angle n{j + A:)/12, j ,fc= 0,1,..., 5. If TT < n{j + k)/l2 < 27r, the point is opposite to the basis hi of angle ^i = 7r(j + k)/6
- TT.
T h e s u m of t h e 6 - c o m p l e x n u m b e r s u a n d w' is u + u' = X{)+ X'Q + hi{xi
+ x\) + hi{x2
+h4{x4 + .T4) + h^{xs + 4 ) -
+ X2) + hz{x3
+ x'^)
(5.80)
182
Complex Numbers
in 6
Dimensions
Figure 5.3: Representation of the planar hypercomplex bases ^'>hi^h2^hs^h4^h^ by points on a circle at the angles a^ = 7tk/6. T h e product hjhk will be represented by the point of the circle at the angle 7r{j + k)/l2j i,k = 0,1,...,5. If TT < IT{J + k)/12 < 27r, the point is opposite to the basis hi of angle ai = n{j + k)/6 TT. T h e product of the numbers w, ?i' is UU
=
xix^^ — 0:23:4 — 0:3X3 X4X2
XQXQ
•i-hi{xox[
3:53:^
+ XIXQ — X2x/^ — X3.X4 — 0:4X3 — 0:50:2)
4-/12(2:0^2 + ^ 1 ^ 1 + ^2^0 " ^33^5 -- 3:40:4 - x^x/:^)
(5.81)
4-/13(3:03:3 4- 3:13:2 4- 3:23:1 4- 3:33:0 - 3:43:5 - 3:53:4) +h4{xoX^^ 4- 3:13:3 4- X23:2 -f 3:30:1 4- 0:4X0 - 3:5X5) +h^{xox/^
+ X1X4 4- X23:3 4- X3X2 4- X4X1 4- XSXQ).
T h e relation between the variables t;i,t;i, 1^2,02,1^3,1)3 and xo,3:i,X2,3:3, 3^4? 3:5
are / Vi\ V] V2
h V3
\ '"^3 )
1fl
0 1 0 1
u
v/3 2 1 2
1 2
0 1
-1 0
2 1 2
2
1 2 ^/3 2
0 1 0 -1 0 1
1 2
2
2 1 2
1 0
0 1
1 2 v/3 2
2
V3
'
/ ^'0 \ Xi X2
(5.82)
3^3 X4
^- J\ ^5
/
183
Planar complex numbers in 6 dimensions
The other variables are v^ = V3,V4 = —v^.v^ = V2,v^ = —V2,ve = vi,ve = —vi. T h e variables vi^v\^V2^V2^vs^vs complex variables.
5.2.2
will be called canonical planar 6-
Geometric representation of planar complex numbers in 6 dimensions
The 6-complex number u = XQ + hixi + /i2^2 + ^^3^3 + h^x/^ + /15X5 is represented by the point A of coordinates {xo,xi,X2,xs,X4,X5), The distance from the origin O of the 6-dimensional space to the point A has the expression ct
(5.83)
1 x^ -J- x^ "7" ^ 5 ^
— ^Q 4" ^1^ "T" X2
is called modulus of the 6-complex number u, and is designated by d = \u\ The modulus has the property that
luV'l < V3|u'||«"|.
(5.84)
The exponential and trigonometric forms of the 6-complex number u can be obtained conveniently in a rotated system of axes defined by a transformation which has the form
I
1 N/3
0 1 x/3
I2
0 1
V^~3 J
V3
1 2 1 2\/3
0 1
A21
1 2v/3 2 1
0 1 x/3
0
1__ 2\/3 2 1 V^
1 2\/3
/
^0
0
0 J_
^2
1 v/3
1 2\/3
0
1 2i/3
f
X4
""2
^
V3
0
^3
\
(5.85)
V ^5 y 2 _i_ / 2v/3 / The lines of the matrices in Eq. (5.85) give the components of the 6 vectors of the new basis system of axes. These vectors have unit length and are orthogonal to each other. The relations between the two sets of variables are ^fc
V 0
1 2\/3
1
""2
= \/3(,k.Vk = \/3%,
(5.86)
for fc = 1,2,3. The radius pk and the azimuthal angle (l)k in the plane of the axes v^^ v^ are 2 _ ^,2
Pit = ^fc + ^h cos>fc = Vk/pk, sm(f)k = Vk/pk,
(5.87)
184
Complex Numbers in 6 Dimensions
where 0 < (f)^ <27r, k = 1,2,3, so that there are 3 azimuthal angles. The planar angles ipk-i ^r^ (5.88)
tanv^i = pi/p2, tanV^2 = Pi/ps,
where 0 < i/^i < 7r/2, 0
1/3
(5.89)
It can be checked that (5.90)
rf - 3 ( P l + P 2 + Pl)•
If ?/ = u'u"^ the parameters of the hypercomplex numbers are related by Pk = p'kP'L
(5.91)
t a n V'fc = tan i/)^ tan i/'lt»
(5.92)
(f>k = 'k + 4>k^
(5.93)
'"k = v'kv'l - v'kvl, Vk = v'kvl + v'kv'k,
(5.94)
P = P'P".
(5.95)
where fc = 1,2,3. The 6-complex planar number u = XQ-\-h\x\ +/t2a^2 + ^3a;.3 + ^4X4 + /15X5 can be represented by the matrix /
[/ =
\
^3
X\
.^•5 \
.xo
^2 .Ti
X2
X?,
.T4
-X\
•~.X5
x^
X\
X2
^3
-^3
—Xx -x^
Xo
XX
X2
-X2
-^3
X\
^0
XX
— ^5
-XX
-'^^2
-.T4
-^•5
Xo
-.'^3
~.T4
-X5
(5.96)
^0 /
The product u = u'u" is represented by the matrix multiplication U = U'U".
Planar complex numbers in 6 dimensions 5.2.3
185
T h e planar 6-dimensional c o s e x p o n e n t i a l functions
The planar •cosexponential functions in 6 dimensions are oo
/c+6p
for k = 0, ...,5. The planar cosexponential functions of even index k are even functions, /6,2/(~y) = /6,2/(y)? ^nd the planar cosexponential functions of odd index are odd functions, /6,2/+i("-J/) = -~/6,2/+i(y)? ' = 0,1,2. The exponential function of the quantity hkV is e^'y = feoiy) + hifeiiy)
+ /i2/62(y) + hsfesiy) + h4f64{y) + /is/esCl/),
e^'^ = geoiy) - 563(y) + h2{gQi{y) - ge^iy)} + /i4{562(y) - ffesly)}, e^^^ = /6o(y) - /62(y) + /64(y) + hsifeiiy) - /eaCy) 4- /esCy)}, e^^^ = 96o(y) + 963(y) -/i2{fl'62(y) +565(y)} + /i4{g6i(y) + 564(y)}, e'^^^ = /6o(y) + hife5{y) - h2fM{y) + hshsiy) ~ h^My) + hhiiy)^
(5.98) The relations for /12 and h^ can be written equivalently as e^^^ = /30 + h2hi + h4f^2^^^'^^ = 330 - ^2/32 + h^g^i, and the relation for /13 can be written as e^^^ = /20 + /i3/2i, which is the same as e^^^ = cosy + hssiny. The planar 6-dimensional cosexponential functions fek{y) are related to the polar 6-dimensional cosexponential function g^kiy) by the relations /6fc(y)=e-^'''/V(e^'^/'y),
(5.99)
for k = 0, ...,5. The planar 6-dimensional cosexponential functions fek{y) are related to the polar 6-dimensional cosexponential function gekiy) also by the relations f6k{y)=e~''''/^gek{iy).
(5.100)
for A; = 0,..., 5. The expressions of the planar 6-dimensional cosexponential functions are feoiy) /6i(y) /62(y) fesiy) /64(y) /65(y)
= = = = = =
^cosy4- | cosh ^ y cos | , jsiny-l- ^ s i n h ^ y c o s | + i cosh ^ y sin | , -jcosy + |cosh^ycos| + ^ s i n h ^ y s i n | , - 5 s i n y + § cosh ^ y sin | , 5 cosy - 5cosh ^ y c o s I + ^ s i n h ^ y s i n | , 5 siny - ^ sinh ^ y cos | + | cosh ^ y sin | .
.^ ^^^.
186
Complex Numbers in 6 Dimensions
The planar 6-dimcnsional cosexponential functions can be written as , , . 1 ^ [ /7r(2/-l)\l r . /7r(2/-l)\ Jekiy) = g2l^explj/cosl 1 cos i/sm( ~ j
1=1 7r(2/ - l)k
(5.102)
for k = 0, ...,5. The graphs of the planar 6-dimensional cosexponential functions are shown in Fig. 5.4. It can be checked that
J2fik{y) = \ + lcoshV3y.
(5.103)
The addition theorems for the planar 6-dimensional cosexponential functions are 560(y + z) = g6o{y)g6o{z) - 96i{y)g6^{z) -562(y)p64(^) -9e3{y)963{^) - 9e4{y)962{z) - 965{y)96iiz), geiiy + z) =9eo(y)9e}{z) + 96i(y)96o{z) - 962{y)96b{z) -96:i{y)964{z) - 964(y)963{z) -965{y)962{z)^ 962{y + z) = geo{y)g62{z) + 9m (y)96i(z) + 9^2{y)9m{z) -9e3{y)96^{z) - 964{y)9e4{z) - 96b{y)96:i{z), 963{y-^z) = 96o{y)963(z) -^ 9ei{y)962{z) -i-962{y)96i{z) -^9e3{y)960{z)
- 964(y)965{z)
-
g65(y)964{z),
ff64(y + 2:) = 96o{y)9e4{z) + 9ei{y)963{z) 4-562(y)562(^) -^963{y)96\{z) + 964{y)960{z) - 565(y)65(^), 565 (y + Z) = 560(y)565(^) + 56l(y)564(^) + 9^2{y)963{z) +563(y)562(^) + 564(y)56l(^) + 565(y)56o(^). (5.104) It can be shown that
Planar complex numbers in 6 dimensions
Figure 5.4: Planar cosexponential functions /eo, /ei, /62, /ea, fei, /es-
187
188
Complex Numbers in 6 Dimensions
{My) + hiUiiy) + h'iMy) + h^My) + /M/64(y) + hMv)]^ = Uo(ly) + hifeiily) + h2h2{hi) + hMly) + hif^ily) +h-,U->{ly), {.96o(«/) - 563(2/) + h2{g(ii{y) - 964{y)} + h4{g62{y) - 565(2/)}}' = 96o{ly) -963(ly) + h2{g6i{ly) -.964(/»/)} + h4{ge2(ly)
-gasily)}, {My) - My) + My) + h3{My) - My) + /65(y)}}' = hoiiy) - Miy) + Miy) + h{Miy) - h^iiy) + Miy)}, {geoiy) + 563(2/) - /J2 {562(2/) + 565 (y)} + /i4 {561(2/) + 564(5)}}' = 560(/y) + 563(^5) - h2{gfi2{ly) + ge^ily)} + /i4{56i(ly) +564 (/y)},
{/6o(5) + h]My) - h2My) + hMy) - hiMy) + hMy)Y = Mh/) + hiMiy) - h2M^y) + hsMh) - h^Mh) +h5fe\ily)(5.105) The derivatives of the planar cosexponential functions are related by dfi60
- / .65)
du
dhb du
5.2.4
61 /GO?
du
df62 du
63
hii
du
r
4f64 du
r
/62? -~r~ ^ /63?
(5.106)
/64-
Exponential and trigonometric forms of planar 6-complex numbers
The exponential and trigonometric forms of planar 6-complex numbers can be expressed with the aid of the hypercomplex bases / i 3
6
ei
0
i
62
1 3
0 I
62
0
63
i
\ 63 /
6
3
6 I 6
1 6 /3 6, 1 3
0
1 6 s/Ti 6
0
0
1 3
0 1 3 1 3
1 6
^
?
6 1 6
3
d
0
1 3
1 6 v/3 6
^
(
1 \ h2
(5.107)
v/3 6
1 )
T h e multiplication relations for the bases ek,ek are el = gfc, el = -efc, e^Ck = ejt, 6^:6/ = 0, ejke; = 0, ejte/ = 0,
A;,/= 1,2,3, fc^L
(5.108)
Planar complex numbers in 6 dimensions
189
The moduli of the bases e^^ e^ are |efe| = y^,|efc| = ^ ,
(5.109)
for A: = 1,2,3. It can be shown that 3
xo 4- hixi -f h2X2 -f hsxs 4- h^x^ + h^x^ = ^{e^Vk
-f- e^Vk)'
(5.110)
k=i
The ensemble ei, ei, e2, e2,63,63 will be called the canonical planar 6-complex base, and Eq. (5.110) gives the canonical form of the planar 6-complex number. The exponential form of the 6-complex number u is u = pexp < -(/i2 — /i4)lntani/?i + ~{V3hi — h2 + h^ — \/3/i5)lntan'02 +ei(/)i + e2(t>2 + e3)3} •
(5.111)
The trigonometric form of the 6-complex number u is / \
I ^2
(
1 \-~^^'^ tan^ 7p2 J
tan^ ipi ^3
\
ei + p -h -; p exp (ei>i + e2<)62 + e3>3) • V tan-01 tan ^ 2 / The modulus d and the amplitude p are related by 2I/3 / Y 1 \^/^ d = p—=. (tant/^i tan^^2)^^^ f 1 + 9-7- + ^-— ) . V^ V tan^i/?! tan2t/;2/ 5*2.5
(5.112)
(5.113) ^
E l e m e n t a r y functions of a planar 6-coniplex variable
The logarithm and power functions of the 6-complex number u exist for all xo,...,^5 and are Inw = l n p + -(/i2 — /i4)lntan'0i o -f--(\/3/ii — /^2 + /i4 — \/3/i5)lntan^2 6 +ei0i + e2>2 + e3>3,
(5.114)
3
u^ = y^pfe^(efeCOsm(?!>fc 4-efcsinm(/)fe).
(5.115)
190
Complex Numbers in 6 Dimensions
The exponential of the 6-complex variable u is 3
gu ^ Y^e''^ {ekcosvk + eksinvk).
(5.116)
The trigonometric functions of the 6-complex variable u are 3
cos u = Y^{ek cos Vk (^^sh v^ ~ c^ sin ^^^ sinh v^),
(5.117)
3
sinii = ^
(e^; sin ?;^ cosh f;^ -f ^A; ^os?;/.sinh?;A:) •
(5.118)
The hyperbolic functions of the 6-complex variable u are 3
coshu = ^
(e^ cosh f/i; cos t;yt + e^t sinh ?;jt sin li^t),
(5.119)
(cjt sinht'^COST)jt + e/tcosh?;jtsin0^).
(5.120)
3
sinhi/ = ^ k=:\
5,2.6
P o w e r series of planar 6 - c o m p l e x n u m b e r s
A power series of the 6-complex variable ix is a series of the form ao -h aiu + a2U^ H
+ aiu^ -h • • •.
(5.121)
Since taw^|<3'/2|a||w|',
(5.122)
the series is absolutely convergent for |n|
(5.123)
c= l i m - J ^ .
(5.124)
where
l-^oo V3|a/4.i| lim = Yll:=Qhpaip, and J^ 7r{2k-l)p Aik = 2^o«pCOS -^ p=0
,
/r. ior,\ (5.12o)
Planar complex numbers in 6 dimensions
191
Ai, = f:ai,sm^^^2!lZ^,
(5.126)
P=:0
where k = 1,2,3, the series (5.121) can be written as (5.127)
Y^iekAik + ekAik){ekVk + hn)^ /=0
.k=l
The series is absolutely convergent for p^
(5.128)
where
ck = ,lim
5.2.7
^^
''J
^^,.
(5.129)
A n a l y t i c functions of a planar 6 - c o m p l e x variable
I f / ( " ) = T.l=ohkPkixo,...,X5), dPo dxo
dPx
dP2
dxi
8X2
dP3 dxs
dP4 _dp^____^m dx4 dxfy' dxz
(5.131)
dPo dxi
(5.132)
dxo
dx\
dX2
dP2
dPs dx\
dP^
dxQ
dPs dxo
dP4 dx\
dPr,
dP b dx 3
dX2
dPo dX2
_dPo_ dxi vxi
dP4 dx4
dP^ dx^
dX2
dPi _ dPs _ dxo dxi cx'o dxo
then
UX2 5X2
dP^ dx^'
dPi dx^'
dPi
dP2
dXA
8X5 '
dPi _ dxz 8P2 _ 8x^ cx"3
dP2 _ dx4 8P^ _ 8x4 ux'4
(5.133)
dPs dx^'
(5.134)
dP4 dx^ ax"5
(5.135)
and d'Pk dxodx I
d'Pk dxidxi^i d'Pk dxi+xdx^
d'pk dX[i/2]dXi_^[/2]
d'Pk dxi+2dX4
d^p. dxi+imi_iy2]dX5_[(^i_i)/2]'
(5.136)
192
Complex Numbers in 6 Dimensions
5.2.8
Integrals of planar 6-complex functions
If f{u) is an analytic 6-compIex function, then
i
r
f{u)du
= 2nf{uo) {ei int{uo^^rj,,T^,rii) + ^2 int{uQ^^ri2^r^^r)2) u~uo +63 int(wo^3^3, r^3^3)} , (5.137)
where UQ^^^jk ^^d F^^^^ are respectively the projections of the point UQ and of the loop r on the plane defined by the axes (,k ^nd r/^, A; = 1,2,3. 5.2.9
Factorization of planar 6-complex p o l y n o m i a l s
A polynomial of degree rn of the planar 6-complex variable u has the form P^{u) = u'^ + aiu"^-^ + • • • + am-iu + am,
(5.138)
where a/, for / = l,...,m, are 6-complex constants. If ai = Yll^ohpaip, and with the notations of Eqs. (5.125)-(5.126) applied for / = 1, • • • ,m, the polynomial Pm{u) can be written as
(5.139) where the constants Ai^^Aik are real numbers. The polynomial Pm(^^) can be written as a product of factors rn
Pm{u) = l[{u^Up),
(5.140)
p=l
where 3
% = Z l (^^^^ + ^f'^^^p'^'
(5.141)
for p = l,...,m. The quantities e^Vkp + ^^k^kpi P = l,...,m,fe = 1,2,3, are the roots of the corresponding polynomial in Eq. (5.139) and are real numbers. Since these roots may be ordered arbitrarily, the polynomial Py7i(ix) can be written in many different ways as a product of linear factors. If P{u) =11^ + 1, the degree is rn — 2, the coefficients of the polynomial are ax = 0,a2 = 1^ the coefficients defined in Eqs. (5.125)-(5.126) are ^21 = 1,^21 = 0,^22 = 1,^22 = 0,yl23 = 1,^23 = 0. The expression, Eq. (5.139), is P(u)=(eii;i-feii)i)'^ + ei + (e2i^2 + ^2^^^2)^ + e2 + (e3^;3 4-637)3)^ + 63.
Planar complex numbers in 6 dimensions
193
The factorization of P(ti), Eq. (5.140), is P{u) = {u - ui){u — U2), where the roots are ui = ±ei ± 62 ± 63,U2 = —ui. If ei, 62,63 are expressed with the aid of Eq. (5.107) in terms of h\,h2,hz,h4,h^, the factorizations of P{u) are obtained as u + i(2/ii + /i3 + 2/15)] [u - i(2/ii + /13 + 2/15)] , u2 + l = hi +l{hi
+ N/3/I2 - /13 + \/3/i4 + ^5)]
[« - i(/li + v/3/l2 -h3 + N/3/14 + ^5)] , u^ + 1 = (u + hz)(u - /is),
(5.142)
n^ + l = « + i ( - / l i + V^/l2 + /l3 + V^/14 - /15)] [u - | ( - / i i + \/3/i2 + /i3 + \/3/i4 - /is)] • It can be checked that (±ei db 62 + ±63)^ = —ei — 62 — 63 = —1. 5.2.10
R e p r e s e n t a t i o n of planar 6 - c o m p l e x n u m b e r s by irreducible m a t r i c e s
If the unitary matrix written in Eq. (5.85) is called T, the matric TUT~^ provides an irreducible representation [7] of the planar hypercomplex number u, / Vi TUT-^
=
0
0
0 1 / 2 0
V 0
0
(5.143)
V3
where U is the matrix in Eq. (5.96) used to represent the 6-complex number u, and the matrices Vk are
ior k = 1,2,3.
Vk
Vk
-Vk
Vk
(5.144)
This Page Intentionally Left Blank
Chapter 6
Commutative Complex N u m b e r s in n Dimensions Two systems of complex numbers in n dimensions are described in this chapter, for which the multiplication is associative and commutative, which can be written in exponential and trigonometric forms, and for which the concepts of analytic n-complex function, contour integration and residue can be defined. The n-complex numbers introduced in this chapter have the form u = XQ + hixi + /i2^2 H + /^n-i^n-i^ the variables XQ, ...,Xn-i being real numbers. The multiplication rules for the complex units hi, ...,/in-i discussed in Sec. 6.1 are hjhk = hjj^k if 0 < 7 4- A: < n — 1, and hjhk = hj^k-n if ^ ^ i + fc:^2n — 2. The product of two n-complex numbers is equal to zero if both numbers are equal to zero, or if the numbers belong to certain n-dimensional hyperplanes described further in this chapter. If the n-complex number ?/ = a^o 4- h\X\ -\- h2X2 + • • • + /zn-i^n-i is represented by the point A of coordinates xo^^i? "",Xn-i^ the position of the point A can be described, in an even number of dimensions, by the modulus d = {xl + x\-\ hx'^_i)^^'^, by n / 2 - 1 azimuthal angles >jt, by n/2 — 2 planar angles tph-ii ^i^d by 2 polar angles 6J^,6-, In an odd number of dimensions, the position of the point A is described by d, by (n — l)/2 azimuthal angles ^^k-, by (n —3)/2 planar angles tpk-i-, ^'Wd by 1 polar angle 6^. An amplitude p can be defined for even n as p'^ = V-^v-p\ - • • p^/2_i? ^i^d for odd n as p" = ^+Pi •••P^n-i)/2' where v^ == XQ + • • • + Xn-\,V-. = XQ - Xi H + Xn-2 — ^n-ij ^^^ Pk ^tre radii in orthogonal two-dimensional planes defined further in Sec. 6.1.. The amplitude p, the variables v^.v^, the radii pk-, the variables (l/\/2) tan^^., (l/\/2) tan^_, tan^j^-i are multiplicative, and the azimuthal angles (j)k are additive upon the multiplication of n-complex 195
196
Commutative Complex Numbers in n Dimensions
numbers. Because of the role of the axis i;_|_ and, in an even number of dimensions, of the axis t;_, in the description of the position of the point A with the aid of the polar angle 6^ and, in an even number of dimensions, of the polar angle 0_, the hypercomplex numbers studied in Sec. 6.1 will be called polar n-complex number, to distinguish them from the planar n-complex numbers, which exist in an even number of dimensions. The exponential function of a polar n-complex number can be expanded in terms of the polar n-dimensional cosexponential functions gnkiv) ~ T.T^^y^^^''l{k+pn)l k - 0,l,...,n - 1. It is shown that Qnkiv) = n 5]rJo^^^P{2/^^s(2^'/^0}cos{ysin(27r//n) - 27rkl/n}, k = 0,1,...,n - 1. Addition theorems and other relations are obtained for the polar n-dimensional cosexponential functions. The exponential form of a polar n-complex number, which in an even number of dimensions n can be defined for .TQ H h.x„_ i > 0, XQ —rrj H fXn-2 - Xn-\ > 0, is tx - p e x p J E p r J hp [(1/n) hi x/2/tan^+ + ( ( - l ^ / n ) l n \ / 2 / t a n 0 _ - [2ln)YTk~2^ cos(27rfcp/n) Intani/?^-]] j exp (22^ffj~^ 4(/)fc), where ek = (2/n) Xlp=i hp sm(27Tpk/n). In an odd number of dimensions n, the exponential form exists for XQ + • • • + Xn-i > 0, and is given by u = pexp {YlpZl hp [(1/n) In v/2/ tan 19+ - {2/n) E I I V ^ ^ ^ ^^S {27rkp/n) Intant/^Aj-i]} ^W[YX=I ^k(t>k)' A trigonometric form also exists for an n-complex number u, when u is written as the product of the modulus d, of a factor depending on the polar and planar angles 0^., ^_, t/^/c_i and of an exponential factor depending on the azimutlial angles (p^. Expressions are given for the elementary functions of polar n-complex variable. The functions f{u) of n-complex variable which are defined by power series have derivatives independent of the direction of approach to the point under consideration. If the n-complex function f{u) of the n-complex variable u is written in terms of the real functions Pi^{xo,,,,^Xn-\)^ k = 0,..., n - 1, then relations of equality exist between partial derivatives of the functions Pfc. The integral f^ f{u)du of an n-complex function between two points i4, B is independent of the path connecting A, B, in regions where / is regular. If f{u) is an analytic n-complex function, then §^ f{u)du/{u - uo) Vk^^^km)^ where the functional int takes the values 0 or 1 depending on the relation between t^o^AT?^ ^^^ ^^kVk^ which are respectively the projections of the point UQ and of the loop F on the plane defined by the orthogonal axes (,k and 77^, as expained further in this work. A polar n-complex polynomial can be written as a product of linear or quadratic factors, although the factorization may not be unique.
Introduction to Chapter 6
197
Particular cases for n = 2,3,4,5,6 of the polar n-complex numbers described in Sec. 6.1 have been studied in previous chapters. The multiplication rules for the complex units hi,..., hn-i discussed in Sec. 6.2 are hjh^ = hj^k if 0 < j 4- fc < n — 1, and hjh^ = —hj^k-n if ^ ^ i + ^' ^ £ 2n — 2, where /IQ = 1. The product of two n-complex numbers is equal to zero if both numbers are equal to zero, or if the numbers belong to certain n-dimensional hyperplanes described further in Sec. 6.2. If the n-complex number u = x^ + h\X\ + h2X2 + • • • + /in-i^n-i is represented by the point A of coordinates x c ^ i , ...,0:^-1, the position of the point A can be described, in an even number of dimensions, by the modulus d = [xl + xf-] ha:^_i)^/^, by n/2 azimuthal angles (f)k and by n/2 — 1 planar angles 'ipk-i- An amplitude p can be defined as p^ = p2 . . . p2 where p^ are radii in orthogonal two-dimensional planes defined further in Sec. 6.2. The amplitude p, the radii pk and the variables tan^fc_i are multiplicative, and the azimuthal angles (j)k are additive upon the multiplication of ncomplex numbers. Because the description of the position of the point A requires, in addition to the azimuthal angles (p^^ only the planar angles ipk~ii the hypercomplex numbers studied in Sec. 6.2 will be called planar n-complex number, to distinguish them from the polar n-complex numbers, which in an even number of dimensions required two polar angles, and in an odd number of dimensions required one polar angle. The exponential function of an n-complex number can be expanded in terms of the planar n-dimensional cosexponential functions fnkiv) = E^=.o{-l)V^^7(fc +P^)!,fc = 0,1,...,n ~ 1. It is shown that Ukiv) = (l/n)ELiexp{ycos(7r(2/ - l)/7i)}cos{ysin(7r(2/ - l)/n) - nk{2l - l ) / n } , k = 0, l,...,n — 1. Addition theorems and other relations are obtained for the planar n-dimensional cosexponential functions. The exponential form of a planar n-complex number, which can be defined for all values of XQ, ..., Xn-i, is given by u = pexp {Ep=i hp [—(2/n) Efc=2 ^^s (^(2^ - 1)P/^) IntdLUipk-i] } exp (Efcfa h^Pk), where 4 = (2/n) Ep=i ^pSin(7r(2A; — l)p/n). A trigonometric form also exists for an ncomplex number, being given by n = d {n/2)^^^ (1 + 1/ tan^ -01 + 1/ tan^ -02 + • • • 4- 1/ tan^ ipri/2-i) {ei + Yll~2 ^kl tanipk-i) exp ( E ^ ^ ^ I h(l>k)' Expressions are given for the elementary functions of planar n-complex variable. The functions f{u) of planar n-complex variable which are defined by power series have derivatives independent of the direction of approach to the point under consideration. If the n-complex function f{u) of the ncomplex variable u is written in terms of the real functions Pfc(a;o,..., ^n-i)^ fc = 0,..., n — 1, then relations of equality exist between partial derivatives
198
Commutative Complex Numbers in n Dimensions
of the functions P^. The integral J^ f{ii)du of an n-complex function between two points A, B is independent of the path connecting A, B, in regions where / is regular. If f{u) is an analytic n-complex function, then ^ f{u)du/{u - uo) = 27Tf{uo) YllL] h int(?ioam-'r^Ar7J^ where the functional int takes the values 0 or 1 depending on the relation between u^^^^r^k and r^^.;^^., which are respectively the projections of the point 7io and of the loop r on the plane defined by the orthogonal axes <^^ and %? as expained further in Sec. 6.2. A planar n-complex polynomial can always be written as a product of linear factors, although the factorization may not be unique. For n = 2, the n-complex numbers discussed in this paper become the usual 2-dimensional complex numbers x + iy. Particular cases for n = 4 and n = 6 of the planar n-complex numbers described in this paper have been studied in previous chapters.
6.1
Polar complex numbers in n dimensions
6.1.1
Operations with polar n-complex numbers
A complex number in n dimensions is determined by its n components {xo^Xi^...^Xn~\)' The polar n-complex numbers and their operations discussed in this section can be represented by writing the n-complex number (xo,xi,...,0:^-1) as IX =: xo -f h\X\ -f- /12X2 + ••• + hn-^Xn-i, where /^i? ^2? • • • 5 hn-\ are bavSes for which the multiplication rules are hjhk = hi, I = j -]-k~ n[{j -h k)/n],
(6.1)
for j , fc, / = 0,1,..., n - 1. In Eq. (6.1), [{j 4- k)/n] denotes the integer part of (j -f k)/n, the integer part being defined as [a] < a < [a] + 1, so that () < j -^ h ~ n[{j -f k)/7i] < n - 1. In this chapter, brackets larger than the regular brackets [ ] do not have the meaning of integer part. The significance of the composition laws in Eq. (6.1) can be understood by representing the bases hj.hk by points on a circle at the angles Oj — 2'KJ/n,ak — 2'Kkln, as shown in Fig. 6.1, and the product hjhk by the point of the circle at the angle 27r(j 4- k)ln. If 27r < 27r(j + k)/n < An, the point represents the basis hi of angle ai = 2IT{J + k - n)/n. Two n-complex numbers u = xo+h\Xi+h2X2-\ h/in-i-'^n-i^ '^^ = ^0 + hix'i+h2X2-^ h/in-i3:'n_i are equal if and only iixi = a;-,i = 0, l , . . . , n - l . The sum of the n-complex numbers ?/ and u' is u + u' = xo+ XQ + hi{xi -f j;|) + • • • + hn-]{xn-i
+ .'2^n-i)-
(6-2)
Polar complex numbers in n dimensions
199
Figure 6.1: Representation of the hypercomplex bases l,/ii,...,/in-i by points on a circle at the angles ak = 27rfc/n. The product hjhk will be represented by the point of the circle at the angle 2w{j + k)/n, i,k = 0,1, ...,n — 1. If 2iT < 2n{j + k)/n < 47r, the point represents the basis hi of angle ai = 2TT{J + k)/n — 27r. The product of the numbers u, u' is W = XQXQ + XiXn^i 4- 3:2a:'n-2 + ^3^n-3 + * * * + Xn--ix[ +hi{xoXi + XIXQ + a;2a;'^_i + ^3^n-2 •* ^ ^n-~1^2) 4-/12(^0^^2 + ^ i M + ^2^0 •*- ^ 3 ^ n - l + 1- ^n-1^3)
(6.3)
-+-/in»l(^0<-i + ^l^n-~2 + ^ 2 < _ 3 + ^ 3 < - 4 + • • • 4* ^ n - l ^ ) '
The product uu' can be written as n—1
n-1
W =J2f^f^Yl Xix[_i^^[^^_i^_^^iy^y /c=0
liu^u\u^'
(6.4)
1=0
are n-complex numbers, the multiplication is associative
(W)ti'' = u{v!v!')
(6.5)
and commutative W = v!u,
(6.6)
200
Commutative Complex Numbers in n Dimensions
because the product of the bases, defined in Eq. (6.1), is associative and commutative. The fact that the multipHcation is commutative can be seen also directly from Eq. (6.3). The n-complex zero is 0 + /ii -OH h/^n-i '0, denoted simply 0, and the n-complex unity is 1 + /ii • 0 -f • • • + hn-i • 0, denoted simply 1. The inverse of the n-complex number u — XQ •}- hixi + /i2^2 -)-••• + hfi-iXn-i is the n-complex number u' = OTQ -f hix[ + /i2^2 ^ •" /^n-i^n-i having the property that uu' = 1.
(6.7)
Written on components, the condition, Eq. (6.7), is XQXQ + Xix'^^^ + ^2^n-2 + ^3^n-3 + * * * + ^n-l^i = 1, 0:0:^1 + X\X'Q + X2X'^_Y + Xsx'^_2 ^ ^ ^n-1^2 = 0, XQX2 + XiX\ + X2XQ -f Xsx'^_i -j
f- 3:^-1:2^3 = 0 ,
(6.8)
a:o^n-l + ^l^n-2 + ^2^n-3 + ^3^n-4 + ' ' * + ^n-1^0 = 0.
The system (6.8) has a solution provided that the determinant of the system, u = det(i4).
(6.9)
is not equal to zero, ^ 7 ^ 0 , where (
A= \
^0
^n-1
X\
rro
Xn-2 ^n—l
X2
XX
xo
T.^
1
' ••
xr \
•
X2
'
X3
'
X(\
(6.10) i
If i^ > 0, the quantity (6.11) will be called amplitude of the n-complex number w = a;o + h\X\ + h2X2 + Vhn-\Xn-\' The quantity v can be written as a product of Hnear factors n-l
i^ = ]][ (^0 + ^kX\ + ^\X2 + • • • + ^fc ^Xn-\) ,
(6.12)
^•=0
where 6^ = e^^^^l^, i being the imaginary unit. The factors appearing in Eq. (6.12) are of the form .^0 + ekXi + 6^X2 + • • • + e^-^Xn-i =Vk + ih,
(6.13)
Polar complex numbers in n dimensions
201
where n-l
Vk
= J2xp COS 27rkp n
(6.14)
p-0
sin
Vk
p=0
2'nkp
(6.15)
r)
for fc = 1,2,..., n — 1 and, if n is even, k ^ n/2. Forfc= 0 the factor in Eq. (6.13) is VJ^ =^ XQ + X \ - \
1- Xn-X,
(6.16)
and if n is even, for k = n/2 the factor in Eq. (6.13) is V- -X{^- Xi-{
h Xn-2 - Xn~\'
(6.17)
It can be seen that Vk = Vn-k-^^k — ^^n-k^ k = l,...,[(n ~ l)/2]. The variables v^^V-^Vk-^Vk^k = 1,..., [(n —1)/2] will be called canonical polar ncomplex variables. Therefore, the factors appear in Eq. (6.12) in complexconjugate pairs of the form Vk + ivk and v^^k + ^^n-A: = ^it — i'^ki where k = 1,..., [(n — l)/2], so that the product v is a real quantity. If n is an even number, the quantity u is n/2-l
u:=v^v,
n
(^fc + ^fc)'
(6-18)
and if n is an odd number, u is (n-l)/2
^=^+
n
(^fc+*fc)-
(6.19)
ifc=0
Thus, in an even number of dimensions n, an n-complex number has an inverse unless it lies on one of the nodal hypersurfaces XQ + xi + • • • + Xn-\ = 0, or xo - .Ti + h Xn-2 ~ ^n-1 = 0, or v\ = 0,{ii = 0, ..., or '^n/2-1 = 0?^7i/2-i — 0- I^ ^^ ^dd number of dimensions n, an n-complex number has an inverse unless it lies on one of the nodal hypersurfaces ^0 + ^1 + • • • + Xn~\ = 0, or ^1 = 0, til = 0, ..., or ^(n-l)/2 = 0, i*(n-l)/2 = 0-
202
Commutative Complex Numbers in n Dimensions
6.1.2
G e o m e t r i c representation of polar n - c o m p l e x n u m b e r s
The n-complex number XQ + hiXi -f/i2:r2H h/^n-i^n-i ^an be represented by the point A of coordinates (xo,X],...^Xn~i)' If O is the origin of the n-dimensional space, the distance from the origin O to the point A of coordinates {xo^xi^..,^Xn-i) has the expression d^ = a:Q + ^1 H
(6.20)
h a:^_i.
The quantity d will be called modulus of the n-complex number u = XQ + hixi +h2X2 + h/in-i^n-i- The modulus of an n-complex number u will be designated hy d = \u\. The exponential and trigonometric forms of the n-complex number u can be obtained conveniently in a rotated system of axes defined by a transformation which, for even n, has the form 1 y/n
1 y/n
\
:
fn sm
0 /
2TTk
\
y/n
y/n
\/n
Vk
1
y/n
f2 ^^„
27t(n-2)k
\/|sin
27r(n~2)A; n
N/ICC
27r(n--l)A; n 27r(n~l)fc n
v (6.21)
\ Xn-l J where A; = 1,2,. , n/2 - 1. For odd n the rotation of the axes is described by the relations /
1 y/n
cos
0
-y/n^1
1
27r(n -1) n 27r(n-l) n
27r
vising
\
XQ
Xi X2
Vnn
Vk
0
\ ': J \
V ^ sin
27rfc
n 2^-^2jr(n-l)k n n
\ Xn-l J
Polar complex numbers in n dimensions
203 (6.22)
where A; = 0,1,..., (n — l)/2. The lines of the matrices in Eqs. (6.21) or (6.22) give the components of the n basis vectors of the new system of axes. These vectors have unit length and are orthogonal to each other. By comparing Eqs. (6.14)-(6.17) and (6.21)-(6.22) it can be seen that
i.e. the two sets of variables differ only by scale factors. The sum of the squares of the variables Vk, v^ is, for even n, W2-1 ^ _ 2 Yl (^fc + ^fc) = —IT'^^l + '"+ ^ n - l ) - 2(:roa:2 + • • • + k=l ^ +Xia:3 + • • • + Xn-ZXn-\),
Xn-AXn-2 (6.24)
and for odd n the sum is (n-l)/2
Y^
^ ^ j
(^l+^l) = —;j—(^o+- • • + ^ n - i ) - ( ^ o ^ i + - • •+:rn-2a;n-i).(6.25)
The relation (6.24) has been obtained with the aid of the identity, valid for even n,
E
27rpfc
f ~ 1 , for even p,
.^ ^^.
cos
—I ^ (6.26) n [ 0 , for odd p. ^ ' The relation (6.25) has been obtained with the aid of the identity, valid for odd values of n,
E cosHfi = - l .
(6.27,
From Eq. (6.24) it results that, for even n,
^'-l^^V-^tfo-
(6.28)
and from Eq. (6.25) it results that, for odd n,
S = Ul^t
Y: PI
(6-29)
204
Commutative Complex Numbers in n Dimensions
The relations (6.28) and (6.29) show that the square of the distance d, Eq. (6.20), is the sum of the squares of the projections v^/\/n.pi^s/2jn and, for even n, of the square of v^/y/n. This is consistent with the fact that the transformation in Eqs. (6.21) or (6.22) is unitary. The position of the point A of coordinates (rrg, '^'i, ••., ^n-i) can be also described with the aid of the distance d, Eq. (6.20), and of n - 1 angles defined further. Thus, in the plane of the axes %)k, ?\, the radius pk and the azimuthal angle c/)/^ can be introduced by the relations /^fc = ^fc + ^^l^ COS(/)A: = Vk/pk,
Sin)A: = Vk/pk,()
< (j)k < 27r,
(6.30)
so that there are [{n — l)/2] azimuthal angles. If the projection of the point A on the plane of the axes v^^Vk is Ak, and the projection of the point A on the 4-dimensional space defined by the axes v\,vi,Vk,Vk is Aik, the angle ij^k-x between the line OAik and the 2-dimensional plane defined by the axes Vk-^Vk is tanV^it-i = pi/pk',
(6.31)
where 0 < ipk < 7r/2,A; = 2,...,[(n - l)/2], so that there are [(n - 3)/2] planar angles. Moreover, there is a polar angle 6^, which can be defined as the angle between the line OA]^ and the axis ?;+, where A]^ is the projection of the point A on the 3-diniensional space generated by the axes
tan0+- ^ ^ ,
(6.32)
where 0 < ^-f < TT , and in an even number of dimensions n there is also a polar angle ^_, which can be defined as the angle between the line 0A\- and the axis i;_, where A\^ is the projection of the point A on the 3-dimensional space generated by the axes ?;i,t)i,?;_, tan0_ = ^ ,
(6.33)
V-
where 0 < ^_ < TT . In Eqs. (6.32) and (6.33), the factor \/2 appears from the ratio of the normalization factors in Eq. (6.23). Thus, the position of the point A is described, in an even number of dimensions, by the distance d, by n/2 - 1 azimuthal angles, by n/2 - 2 planar angles, and by 2 polar angles. In an odd number of dimensions, the position of the point A is described by (n - l)/2 azinuithal angles, by (n - 3)/2 planar angles, and by 1 polar angle. These angles are shown in Fig. 6.2.
205
Polar complex numbers in n dimensions
P,
^,
Figure 6,2: Radial distance pk and azimuthal angle (/)fc in the plane of the axes v^^Vk^ ^tnd planar angle 4^k~\ between the line OA\k and the 2dimensional plane defined by the axes Vk^Vk- ^k is the projection of the point A on the plane of the axes Vk^Vk^ and ^ i ^ is the projection of the point A on the 4-diniensional space defined by the axes v\,vi,Vk,Vk- The polar angle 6^ is the angle between the line OAi-^ and the axis v^^ where Ai^ is the projection of the point A on the 3-dimensional space generated by the axes V{,vi^v^, In an even number of dimensions n there is also a polar angle 0_, which is the angle between the line OAi- and the axis ?;_, where Ai_ is the projection of the point A on the 3-dimensional space generated by the axes vi^vi.V-..
206
Commutative Complex Numbers in v, Dimensions
The variables pk can be expressed in terms of d and the planar angles i/'fc as Pk = —^ , tanV'fc-i for k = 2,..., [{n — l)/2], where, for even n, P\
(6.34)
nd^ 1 . 1 1 nd^f / 11 o2 ( .„^2 0^ + tan2 ^_ ^ ^ "^ tan2 ,/,i ^ tan2 xj^^'^" "^ 2 Vtan2^+ Vtan2 -1
+ —Y1
1 .
(6.35)
tan^V'n/2-2/ and for odd n 2 _ nd'^ / 1 1 1 ^1 ~ ^ V t a n 2 6»+ "^ "^ tan^ ^i "^ tan^ ^2 "^ " '
+—W^
)
•
(6.36)
tan^ ^(„-3)/2 / If u' = X'Q + hix\ + h^x'^ -\ 1- /i„_ia;Jj_,, u" = XQ + h\x'{ + h2X2 H (hn-ix'^-i are n-complex numbers of parameters v'^,v'_,p',^,0'^,6'_,'tp'i^,'k^ '•hen the parameters W+, w_, Pk^ 9^, 9-, ipk^ ^k of the product n-complex number u — u'u" are given by v^ = . / + < ,
(6.37)
Pk = p'kP'k,
(6-38)
forA; = l , . . . , [ ( n - l ) / 2 ] ,
tan 6'+ = 4= *^" ^+ **" ^+' v2 tan V"*; = tan V'^ tan ^j^',
(^-^^^ (6.40)
forA; = l , . . . , [ ( n - 3 ) / 2 ] , <^fe = '/'fe + C
(6.41)
for fc = 1,..., [(n - l)/2], and, if n is even, (6.42) tan ^ - = 4 = tan 6''_ tan O'L. \/2
(6.43)
207
Polar complex numbers in n dimensions
The Eqs. (6.37) and (6.42) can be checked directly, and Eqs. (6.38)-(6.41) and (6.43) are a consequence of the relations Vk = v'kv'k
v'kv'k, Vk
(6.44)
4^k + Wfc"lt>
and of the corresponding relations of definition. Then the product v in Eqs. (6.18) and (6.19) has the property that u = i^'u"
(6.45)
and, if v' > 0,i^" > 0, the amplitude p defined in Eq. (6.11) has the property that (6.46)
P'P"-
The fact that the ampUtude of the product is equal to the product of the amplitudes, as written in Eq. (6.46), can be demonstrated also by using a representation of the n-complex numbers by matrices, in which the n-complex number u = XQ + hixi + h2X2 + • • • + hn~iXn-i is represented by the matrix (
U =
X{)
Xi
X2
Xn-l
Xn-l
XQ
Xi
Xn~2
Xn-2
Xn-\
XQ
^n-3
Xi
X2
^3
^0
\
\
(6.47) )
The product u — u^u" is represented by the matrix multiplication U = U^U". The relation (6.45) is then a consequence of the fact the determinant of the product of matrices is equal to the product of the determinants of the factor matrices. The use of the representation of the n-complex numbers with matrices provides an alternative demonstration of the fact that the product of n-complex numbers is associative, as stated in Eq. (6.5). According to Eqs. (6.37, (6.38), (6.42), (6.28) and (6.29), the modulus of the product uv! is, for even n. n/2-l
K P = l(^^.;)2 + i(^_,'_)2 + i Y. iPkP'k?, n
n
n
1
9 {n-l)/2
(6.48)
f-^.
and for odd n (6.49) jfc=i
208
Commutative Complex Numbers in n Dimensions
Thus, if the product of two n-complex numbers is zero, uu' — 0, then v^v\ = 0, pkp'k = 0,fc= 1,..., [(n - l)/2] and, if n is even, ?;_?;'_ = 0. This means that either n = 0, or ?i' = 0, or ?/, ?/' belong to orthogonal hypersurfaces in such a way that the afore-mentioned products of components should be equal to zero. 6.1.3
The polar n-dimensional cosexponential functions
The exponential function of a hypercomplex variable u and the addition theorem for the exponential function have been written in Eqs. (1.35)(1.36). If w = a;o + h\X\ + h2X2 -t- • • • -f /i^i-i^^n-i, then expu can be calculated as expt^ = exp^o • exp(/iixi) • • •exp(/i^_ia::yi_i). It can be seen with the aid of the representation in Fig. 6.1 that hl^v ^hl,p
integer,
(6.50)
for A; = 1, ...,n — 1. Then e^^^ can be written as n-l
g/iA.2/ ^ ^ hp-n[kpln]9np{y)^
(6.51)
p=Q
where the expression of the functions Qnk^ which will be called polar cosexponential functions in n dimensions, is oo
Sn;t(y) = E y ' " ' ' " / ( ^ + p n ) ! ,
(6.52)
forfc= 0,1, ...,n - 1. If n is even, the polar cosexponential functions of even index k are even functions, gn,2p{-y) = gn,2p{y)^ P = 0,1,...,n/2 - 1, and the polar cosexponential functions of odd index are odd functions, gn^2p-^i{—y) ~ —gn,2p+i{y)^ p = 0,1, ...,n/2 - 1. For odd values of n, the polar cosexponential functions do not have a definite parity. It can be checked that
J29nk{y)=ey
(6.53)
and, for even n, E(-'^)''Snk(y) = e-y.
(6.54)
The expression of the polar n-dimensional cosexponential functions is , ^
l^ip^
r
/27r/\1
r
. /2nl\
2nkl]
(6.55)
Polar complex numbers in n dimensions
209
for fc = 0, l,...,n - 1. In order to check that the function in Eq. (6.55) has the series expansion written in Eq. (6.52), the right-hand side of Eq. (6.55) will be written as
1 ^"^
f
r/
I.cos
9nk{y) = - J ^ R e ^ e x p
27r/ . . 27r/\ ,2Trkl h I Sin — y — % }, n n ) n
(6.56)
for A: = 0,1,..., n — 1, where Re{a + ib) = a, with a and b real numbers. The part of the exponential depending on y can be expanded in a series, oo n - 1
(6.57)
9nk{y) = - £ X) ^*^ 1 ~f ^^P r~^P - ^) U" f' for A; = 0,1,..., n — 1. The expression of gnkiv) becomes
(6.58) for A: = 0,1,..., n — 1 and, since 1 n—1
27rl,
,,
— y ^ cos ^^^(jt? — /;;) =
I 1, if p —fcis a multiple of n, 0, otherwise,
(6.59)
this yields indeed the expansion in Eq. (6.52). It can be shown from Eq. (6.55) that n—1
1 n—\
12y cos
n
(?)
(6.60)
It can be seen that the right-hand side of Eq. (6.60) does not contain oscillatory terms. If n is a multiple of 4, it can be shown by replacing y by iy in Eq, (6.60) that n/4-1
E(-i)'5^.(y) A:=0
1 H- COS 2y +
^
'2Td\ COS 2y cos (
n
nJ
, (6.61)
which does not contain exponential terms. Addition theorems for the polar n-dimensional cosexponential functions can be obtained from the relation exphi{y + z) = exphiy • exp/ii^, by substituting the expression of the exponentials as given in Eq. (6.51) for fc = 1, e^'y = gnoiy) + higniiy) + • • • + hn-ign,n-i{y),
9nk{y + z)=^ gno{y)9nk{z) + gni{y)gn,k-i{z) + • • •