A Derivation of n-Dimensional Spherical Coordinates

Author(s): L. E. Blumenson
Source: The American Mathematical Monthly, Vol. 67, No. 1 (Jan., 1960), pp. 63-66
Published by: Mathematical Association of America
Stable URL: http://www.jstor.org/stable/2308932 .
Accessed: 09/02/2015 03:56

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Mathematical Association of America is collaborating with JSTOR to digitize, preserve and extend access to
The American Mathematical Monthly.

http://www.jstor.org

This content downloaded from 128.235.251.160 on Mon, 9 Feb 2015 03:56:13 AM

All use subject to JSTOR Terms and Conditions
1960] CLASSROOM NOTES 63

Vi,R, I and D1$D' forall i. HenceD1ER(G) and D2ER(G). FinallyHR2= V

-R2H. But Hz Vi,Ri, D1, Dt, I and R2:Hi forall i. Hence HER(G). Thus
A C(G): I, R, R2,and R3.
The conversesof Theorem 2 and Theorem 4 are falseas Example 2 indicates.
Finally the A C(G) is, in general nontrivial,as is demonstratedby Example 3.
In a later paper, the author hopes to obtain results concerningG/AC(G),
some relationsbetween the centerand the anticenter,and to explore more fully
the effectof isomorphismand homomorphismon the anticenter.
References
1. R. D. Carmichael,Introductionto the Theoryof Groupsof FiniteOrder,Boston,1937.
2. A. Speiser,Theorieder Gruppenvon endlicherOrdnung,3d ed., Berlin,1937.
3. H. Zassenhaus,The Theoryof Groups(translatedby Saul Kravetz),New York, 1949.

CLASSROOM NOTES
EDITED BY C. 0. OAKLEY, HaverfordCollege
All material
forthisdepartment
shouldbesentto C. 0. Oakley,Department
ofMathe-
matics,Haverford
College,
Haverford,
Pa.

A DERIVATION OF n-DIMENSIONAL SPHERICAL COORDINATES

L. E. BLUMENSON, ColumbiaUniversity
ElectronicsResearchLaboratories
An instructiveexample in linear algebra is the derivation of n-dimensional
spherical coordinates without appealing to geometricintuition.The method of
derivation is based on concepts fromlinear algebra; namely, bases of a vector
space, scalar product, angle between vectors and projection of a vector onto a
subspace. Spherical coordinatesin n-dimensionsare a generalizationof the usual
three-dimensionalsphericalcoordinatesand are particularlyusefulin evaluating
certain integrals taken over the surface of an n-dimensionalsphere. Laterwe
shall give an example of such an integration.
Let En denote real n-dimensionaleuclidean space. Vectors in E. will be de-
noted by bold-faced letters. If x and y are two vectors in En with components
tj and n;, j =1, * , n, respectively,we definethe scalar product of x and y by
n
x.y = 2jvj.
j=1

The nonnegative number ||XI| = (x.x)"/2 is called the norm of x. The angle be-
tween x and y is definedby cos 4==x y/||x||||y||, where 4 is restrictedto the
range 0?< _7r. A set of vectors x1, * * , xn is an orthonormalset in En if
xi*xj=O or 1 accordinglyas i#j or i=j. Any set of n orthonormalvectorsforms
a basis forEn.

This content downloaded from 128.235.251.160 on Mon, 9 Feb 2015 03:56:13 AM

All use subject to JSTOR Terms and Conditions
64 CLASSROOM NOTES [January

Let el, , en be any orthonormalbasis in En. Let x be any vector on the

n-dimensionalsphereof radius r about the origin,that is, f|x||= r. If x = ei
then ffxff2Z = . If Giis the angle between x and ei then xi-xei==r cos Gi.
Hence x= r cos Oie1and x can be specifiedby givingits lengthr and the n
angles O;. But since r2=x-x=r2Z= i cos2sOwe see that the Oiare not independ-
ent of each other. Spherical coordinates in n-dimensionsshow us how to pick
out n -1 angles 01, - - *, On-2, 0 whichare independentof each otherand which,
when combined with the normr, completelydescribe the vector x with respect
to the given orthonormalbasis.
Derivation of the coordinates. Let el, , en and x be as above. Let (1 be
the angle between x and el, <0! < 7r.Then 1= x *el = r cos 01 and
n
x = r cos lel + E jej.
j=2

Now

r =ffxff2=rcos + E tj or Ei 2j= r sin

j=2 j-2

Setting 1=ajarsinc/i,j=2, * , n, we have

n
x = r cos4lei + r sin4' E ajej,
j=2

where ZJ=2 c42= 1. (If 4i is 0 or 7r,thenx = + re,.)

=1) in the
Let U2= a1=2ajej. The vectoru2 is a unitvector(that is fju211
directionof the projectionof x ontothe (n-1)-dimensionalsubspacespanned
by e2, * * *, en. If q52 iS the angle between u2 and e2 then cos c/22=u2e2=a2,
O <t2 < , and
n
U2 = COS 02e2 + E ajej.
je3
Hence,

1 11121 COS 02 + aj or a = sin 2.

j=3 ij3
If we setagj fj sin42,j=3, ,n, then
n
U2 = COS42e2+ sin 02 E f.ei,
j-3

where.103 j8 =1. Thus

n
x = r cos4lel + r sinq5lcos 42e2+ r sin (h sin 0>2 1 3jej.
j=3

This content downloaded from 128.235.251.160 on Mon, 9 Feb 2015 03:56:13 AM

All use subject to JSTOR Terms and Conditions
1960] CLASSROOM NOTES 65

In general, let uj be the unit vector in the directionof the projection of a

onto the space spanned by ej, ej+?, . . ., en,j=2, * * *, n-1 and let tj-i be
_ < oj < 7r,j=2, * , n-1. Then
the angle between Uj and ej,
n2 j$. l c r / n-2
x= r= II sinOAk)cos jej + r II sinOA;Un-l.
j==i k=1 k=1

Now un_l=8n_len_1+8nen, where 1 =!!uniII2 If now we define an

82_1+2.
angle 0 by cos 0= 85, sin 6= n-1 we see that 0?< ir willnotsuffice sincean 1
< ?r. In orderto include all possible combi-
can be negative and sin a _ 0 forO0?<
nations of (n-12 An)we must have 0<0<2r.
Thus if el, . . ., en is a given orthonormalbasis in E. and x is a vector of
normr with components{j with respect to this basis, then
= r cos 41,
j-1
j =r cos
cj IH sin Ok (j 2, . * -2),
k=1

(*) ~~~~~~~~~~n-2
(*) = r sin 0Il sin cI,
k=1
n-2

n r cos0 HI sin k,
k=1

where 0?kJ! 7, j=1, * * *, n-2; 060<22r; 0<r < oo.

Application
to integration.
Let f(t1,***, (n) be a continuous
real-valued
function inEnwhichmaybe written
defined in theform
2 2
fM(l . . .
n) = g(afitl + ***+ atntnt1 + ***+ tn)l
wheretheai areconstantsindependent
ofthes's. We wishto compute
thein-
tegral
off overthesurface
ofthen-dimensional
sphereofradiusrwiththeorigin
as center. If x is the vectorwithcoordinates{j and a the vectorwith coordinates
a1 (these coordinates being with respect to some given orthonormal basis
el,. . .* en) then

f+12
=_r X
t+tn 2 2( * *
.
{n)dS g(a
IIxl
1-
.x, J|xJ2)dS,

wheredS is the surfacedifferential.

Let a, =a/JIalI*and choosevectorsa2, * , an to complete an orthonormal
basis in En. Let the coordinates of x with respect to the basis a,, . . ., an be
*
I.. n. Then ai x= ,. Make the spherical coordinate transformation

* If a= 0, thenal maybe any unitvector.

This content downloaded from 128.235.251.160 on Mon, 9 Feb 2015 03:56:13 AM

All use subject to JSTOR Terms and Conditions
66 CLASSROOM NOTES [January

given by (*) with {j replaced by Rj,j= 1, * * *, n.

The Jacobian of the transformationis
n-2
J = rrn'1 H sink 4n-1-k
k=1

Also, a1 x - =r cos 41. Thus the integralbecomes

[ :
27rrn-'
n-8 { r

sin* n-l-kd]n-l-k
-

f r

g(|| alIr cos pl, r2) sinn-2 4idckl

2rnIg co()-1)/2 r J 5 d
r((n- 1)/2) Jg(lallr COS1 r2)
sinn-2oldol
Thus we have reduced the integral over the surface of an n-dimensional
sphere to a single integral on the real line. In particular, if f 1, we obtain
[27rn 2/P(n/2)
]rn-1 for the surface area of an n-dimensionalsphere of radius r
and, integratingfrom0 to r,we obtain [2iin12/(nr(n/2))]rnforthe volume of the
sphere.

MATRIX INTEGRATION OF xkexp (-B2x2)

RoscoE B. WHITE, Universityof Minnesota

Let V be the vector space of finitelinear combinationsof xkexp (-32x2),

fixedf, k = 0,1, * , withbasis { xk exp (-f32x2)}. Let D be a linear transforma-
tion on V which differentiates a vector belongingto V.
Since (xkexp (-f2x2))D kxk-l exp (-,2x2) - 232xk+l exp (-,2x2), the matrix
of D is
0 - 2p2 0
1 0 -2.2.2 .
2 0 -232 ...

* * 0 k 0 -22 . ..
_ * 0A
O * * *A

V is closed under D and the kernelof D consists of the zero vector alone. The
calculation of D-1 may be carriedout algebraically,givingan interestingequa-
tion forfxkexp (-32x2)dx.
Because of the nature of D, D-1 may be calculated in four independent
stepsdepending
on whetherk andj areevenor odd,whereIID-11I
=
I. Using
IIakJ
DD-1 = I, we obtain the followingexpressionsforakJ.
(1) j odd, k odd: aki = 0;

This content downloaded from 128.235.251.160 on Mon, 9 Feb 2015 03:56:13 AM

All use subject to JSTOR Terms and Conditions