Part I

n-dimension Spherical coordinates and the

volumes of the n-ball in Rn(by W en Shih)
1 Introduction
We know that !n ; the surface area of the unit ball in Rn ; gets involved in the fundamental solution for the Laplace
operator. We are also very familiar with the case in R2 and R3 : However, for high dimension case, n 4; it is not
easy to imagine. The purpose of this note is to discuss how to calculate !n for n 4:
R RR R
Let Vn (R) = BR (0) 1dx = 0 ( jxj=r 1dS)dr denote the volume of the ball with radius R in Rn and Sn (R)
d
R
denote the surface area of the ball with radius R in Rn : Then it is clear that Sn (R) = dR Vn (R) = jxj=R 1dS =
R R R d
R
Rn 1 jxj=1 1dS: Since Vn (R) = BR (0) 1dx = Rn B1 (0) 1dx as well, Sn (R) = dR Vn (R) = nRn 1 B1 (0) 1dx; and
n
R R
hence, the surface area of the unit ball in R is !n = Sn (1) = jxj=1 1dS = n B1 (0) 1dx = nVn (1): So, once we know
R
Vn (R); we can obtain !n immediately: However, to calculate Vn (R) = BR (0) 1dx; we not only need n-dimension
R
Spherical coordinates, but also 0 sinn d ; for n 2 N:

2 Observation for n-dimension Spherical coordinates

2.1 2D & 3D case:
In 2D case, we usually let x = r cos 1, y = r sin 1; where r 0 and 0 1 < 2 ; then dxdy = j @(x;y)
@(r; ) jdrd 1 and

@(x; y) cos 1 r sin 1

= det = r cos2 1 + r sin2 1 = r:
@(r; 1 ) sin 1 r cos 1

So; dxdy = rdrd 1 :

In 3D case, we usually let x = r sin 1 cos 2 ; y = r sin 1 sin 2; z = r cos 1; where r 0, 0 1 ; and
@(x;y;z)
0 2 < 2 ; then dxdydz = j @(r; 1 ; 2 ) jdrd 1 d 2 and

2 3
sin 1 cos 2 r cos 1 cos 2 r sin 1 sin 2
@(x; y; z)
= det 4 sin 1 sin 2 r cos 1 sin 2 r sin 1 cos 2
5 = r2 sin 1:
@(r; ; )
cos 1 r sin 1 0

So, dxdydz = r2 sin 1 drd 1 d 2 :

2.2 Generalization:
In 2D case, if we rewrite x ! x1 and y ! x2 ; then

x1 = r cos 1
;
x2 = r sin 1

where r 0 and 0 1 < 2 : Since dx1 dx2 = j @(x1 ;x2 )

@(r; ) jdrd 1 and

@(x1 ; x2 ) cos 1 r sin 1

= det = r cos2 1 + r sin2 1 = r; (1)
@(r; 1 ) sin 1 r cos 1

dx1 dx2 = rdrd 1 :

In 3D case, if we rewrite z ! x1 , x ! x2 ; y ! x3 (Note: the order satis…es right-hand rule), then let
8
< x1 = r cos 1 Note: r cos 1 2 spen(x1 ) and r sin 1 2 spen(x2 ; x3 )
x2 = r sin 1 cos 2 ; Then in x2 x3 -space; it is a 2D case; but r ! r sin 1
:
x3 = r sin 1 sin 2 instead. So, x2 = r sin 1 cos 2 & x3 = r sin 1 sin 2 .

1
where r 0, 0 1 ; and 0 2 < 2 : Since dx1 dx2 dx3 = j @(x1 ;x2 ;x3 )
@(r; 1 ; 2 ) jdrd 1 d 2 and
2 3
cos 1 r sin 1 0
@(x1 ; x2 ; x3 )
= det 4 sin 1 cos 2 r cos 1 cos 2 r sin 1 sin 2 5 (2)
@(r; 1 ; 2 )
sin 1 sin 2 r cos 1 sin 2 r sin 1 cos 2
cos 1 r sin 1 cos 1 r sin 1
= r sin 1 sin 2 det + r sin 1 cos 2 det
sin 1 sin 2 r cos 1 sin 2 sin 1 cos 2 r cos 1 cos 2

cos 1 r sin 1 cos 1 r sin 1

= r sin 1 sin2 2 det + r sin 1 cos2 2 det
sin 1 r cos 1 sin 1 r cos 1

= r sin 1 sin2 2 r + r sin 1 cos2 2 r (by (1))

2
= r sin 1: (3)
Similarly, in 4D case, if we let
8
>
> x1 = r cos 1 Note: r cos 1 2 spen(x1 ) and r sin 1 2 spen(x2 ; x3 ; x4 )
<
x2 = r sin 1 cos 2 Then in x2 x3 x4 -space; it is a 3D case; but r ! r sin 1
;
> x3 = r sin 1 sin 2 cos
> 3 instead. So, x2 = r sin 1 cos 2 ; x3 = r sin 1 sin 2 cos 3
:
x4 = r sin 1 sin 2 sin 3 & x4 = r sin 1 sin 2 sin 3 :

where r 0, 0 1 ;0 2 , and 0 3 < 2 : Since dx1 dx2 dx3 dx4 = j @(x1 ;x2 ;x3 ;x4 )
@(r; 1 ; 2 ; 3 ) jdrd 1 d 2 d 3 and

@(x1 ; x2 ; x3 ; x4 )
(4)
@(r; 1 ; 2 ; 3 )
2 3
cos 1 r sin 1 0 0
6 sin 1 cos 2 r cos 1 cos 2
r sin 1 sin 2 0 7
= det 6 4 sin 1 sin 2 cos
7 (5)
3 3 r cos 1 sin 2 cos
r sin 1 cos 2 cos 3 r sin 1 sin 2 sin 3 5
sin 1 sin 2 sin 3 3 r cos 1 sin 2 sin
r sin 1 cos 2 sin 3 r sin 1 sin 2 cos 3
2 3
cos 1 r sin 1 0
= r sin 1 sin 2 sin 3 det
4 sin 1 cos 2 r cos 1 cos 2 r sin 1 sin 2 5
sin 1 sin 2 sin 3 r cos 1 sin 2 sin 3 r sin 1 cos 2 sin 3
2 3
cos 1 r sin 1 0
+ r sin 1 sin 2 cos 3 det
4 sin 1 cos 2 r cos 1 cos 2 r sin 1 sin 2 5
sin 1 sin 2 cos 3 r cos 1 sin 2 cos 3 r sin 1 cos 2 cos 3
2 3
cos 1 r sin 1 0
= r sin 1 sin 2 sin
2
3 det
4 sin 1 cos 2 r cos 1 cos 2 r sin 1 sin 2 5
sin 1 sin 2 r cos 1 sin 2 r sin 1 cos 2
2 3
cos 1 r sin 1 0
+ r sin 1 sin 2 cos
2
3 det
4 sin 1 cos 2 r cos 1 cos 2 r sin 1 sin 2 5
sin 1 sin 2 r cos 1 sin 2 r sin 1 cos 2
= r sin 1 sin 2 sin2 3 r2 sin 1 + r sin 1 sin 2 cos2 3 r2 sin 1 (by (2) and (3))
3 2
= r sin 1 sin 2: (6)
In 5D case, if we let
8
>
> x1 = r cos 1
>
> Note: r cos 1 2 spen(x1 ) and r sin 1 2 spen(x2 ; x3 ; x4 ; x5 )
< x2 = r sin 1 cos 2
Then in x2 x3 x4 x5 -space, it is a 4D case, but r ! r sin 1
x3 = r sin 1 sin 2 cos 3 ;
>
> instead. So, x2 = r sin 1 cos 2 ; x3 = r sin 1 sin 2 cos 3 ;
>
> x 4 = r sin 1 sin 2 sin 3 cos 4
: x4 = r sin 1 sin 2 sin 3 cos 4 , x5 = r sin 1 sin 2 sin 3 sin 4
x5 = r sin 1 sin 2 sin 3 sin 4

where r 0, 0 1 ;0 2 ,0 3 ; and 0 4 < 2 : By (4), (5), (6), and the same process as above,
we obtain
@(x1 ; x2 ; x3 ; x4 ; x5 )
= r4 sin3 1 sin2 2 sin 3 : (7)
@(r; 1 ; 2 ; 3 ; 4 )
From the above observation, we can guess the n-dimension Spherical coordinates and the Jacobian should be (8)
and (9) in Theorem 1, respectively.

2
3 Observation for Vn (R)
Z
V1 (R) = 1dx1 = 2R
BR (0)
Z Z 2 Z R Z R Z 2
V2 (R) = 1dx1 dx2 = rdrd 1 = rdr 1d 1 (= R2 )
BR (0) 0 0 0 0
Z Z 2 Z Z R
V3 (R) = 1dx1 dx2 dx3 = r2 sin 1 drd 1 d 2 (by(3))
BR (0) 0 0 0
Z R Z Z 2
2 4 3
= r sin 1 dr sin 1d 1 1d 2 (= R )
0 0 0 3
Z Z 2 Z Z Z R
V4 (R) = 1dx1 dx2 dx3 dx4 = r3 sin2 1 sin 2 drd 1 d 2 d 3 (by (6))
BR (0) 0 0 0 0
Z R Z Z Z 2
3 2 1 2
= r dr sin 1d 1 sin 2d 2 1d 3 (= R4 )
0 0 0 0 2
Z Z 2 Z Z Z R
V5 (R) = 1dx1 dx2 dx3 dx4 dx5 = r4 sin3 1 sin2 2 sin 3 drd 1 d 2 d 3 d 4 (by (7))
BR (0) 0 0 0 0
Z R Z Z Z Z 2
8
= r4 dr sin3 1d 1 sin2 2d 2 sin 3d 3 1d 4 (= 2
R5 )
0 0 0 0 0 15
R n
From this observation, we can expect that we need 0
sin d ; for n 2 N to calculate Vn (R):

4 Main Results
By the above observations, we can obtain the following results.

Theorem 1 (n-dimension Spherical coordinates) Let

8
>
> x1 = r cos 1
>
>
>
> x 2 = r sin 1 cos 2
>
>
>
> x 3 = r sin 1 sin 2 cos 3
>
< x4 = r sin 1 sin 2 sin 3 cos 4
. ; (8)
> ..
>
>
>
>
> xn 2 = r sin 1 sin 2 sin 3
> sin n 3 cos n 2
>
>
>
> xn 1 = r sin 1 sin 2 sin 3 sin n 3 sin n 2 cos n 1
:
xn = r sin 1 sin 2 sin 3 sin n 3 sin n 2 sin n 1

where r 0; 0 i ; i = 1; 2; :::; n 2; and 0 n 1 < 2 ; then

@(x1 ; x2 ; ; xn 1 ; xn )
= rn 1
sinn 2
1 sinn 3
2 sin2 n 3 sin n 2: (9)
@(r; 1 ; 2 ; ; n 2; n 1)
R
Lemma 2 ( 0 sinn d ; for n 2 N) (1) If n is even, say, n = 2m for some m 2 N; then
Z
2m 1 2m 3 1
sin2m d ==
0 2m 2m 2 2
or Z
n 1n 3 1
sinn d = :
0 n n 2 2
(2) If n is odd, say, n = 2m 1 for some m 2 N; then
Z 2m 2 2m 4 42
2 ; m 2
sin2m 1 d == 2m 1 2m 3 53
0 2 ; m=1

3
or Z n 1n 3 42
2 ; n 3
sinn d = n n 2 53 :
0 2 ; n=1

Once we have Theorem 1, then

Z
Vn (R) = 1dx1 dx2 dxn
BR (0)
Z 2 Z Z Z Z R
= rn 1
sinn 2
1 sinn 3
2 sin2 n 3 sin n 2 drd 1 d 2 d n 2d n 1
0 0 0 0 0
Z R Z Z Z Z 2
= rn 1
dr sinn 2
1d 1 sinn 3
2d 2 sin n 2d n 2 1d n 1
0 0 0 0 0
Z Z Z
2 Rn
= sinn 2
1d 1 sinn 3
2d 2 sin n 2d n 2 (10)
n 0 0 0

and it follows from Lemma 2 that we have the following Theorem.

Theorem 3 (The volumes of the n-ball in Rn ) (1) If n is even, say, n = 2m for some m 2 N; then

R2m m
V2m (R) = :
m!
(2) If n is odd, say, n = 2m 1 for some m 2 N; then

2m m 1 R2m 1
V2m 1 (R) = :
(2m 1)(2m 3) 3 1

Remark 4 (Recall Gamma function) Gamma function is dedined as

Z 1
( )= y 1 e y dy for > 0;
0

and it has the following properties:

(1) ( + 1) = ( ) for > 0
(2) (n) = (n p 1)! for n 2 N :
(3) ( 21 ) =

Theorem 5 (Vn (R) in terms of Gamma function) For n 2 N;

n
2Rn 2
Vn (R) = :
n ( n2 )

5 Conclusion
Now we have known Vn (R) from Theorem 5; then we are able to obtain !n = nVn (1) immediately. See the following
table.
n=1 n=2 n=3 n=4 n=5 n
4 1 2 4 8 2 5 n n n
Vn (R) 2R R2 3 R 3
2 R 15 R 2R 2 =n ( )
2
4 1 2 8 2 n n
Vn (1) 2 3 2 15 2 2 =n ( 2 )
8 2 n
!n 2 2 4 2 2 3 2 2 = ( n2 )

6 Proofs
6.1 Proof of Theorem 1
It follows the same process in Section 2.2 and by indution.

4
6.2 Proof of Lemma 2
R
First, we compute sinn d by integration by parts,
Z Z Z
sinn d = sinn 1 d( cos ) = sinn 1 cos + (n 1) cos2 sinn 2
d
Z Z
= sinn 1
cos + (n 1) sinn 2 d (n 1) sinn d ;

and hence, Z Z
n n 1
n sin d = sin cos + (n 1) sinn 2
d
or Z Z
1 n 1
sinn d = sinn 1
cos + sinn 2
d : (11)
n n
Second, if n is even, say, n = 2m; for some m 2 N; then by (11)
Z Z
2m 1
sin2m d = sin2m 2 d
0 2m 0
Z
2m 1 1 2m 3
= ( sin2m 3 cos j0 + sin2m 4
d )
2m 2m 2 2m 2 0
Z
2m 1 2m 3
= sin2m 4 d
2m 2m 2 0
=
Z
2m 1 2m 3 1
= 1d
2m 2m 2 2 0
2m 1 2m 3 1
=
2m 2m 2 2
or Z
2m 1 2m 3 1 n 1n 3 1
sinn d = = :
0 2m 2m 2 2 n n 2 2
If n is odd, say, n = 2m 1; for some m 2 N; then by (11) for m 2,
Z Z
2m 2
sin2m 1 d = sin2m 3 d
0 2m 1 0
Z
2m 2 1 2m 4 2m 4
= ( sin cos j0 + sin2m 5
d )
2m 1 2m 3 2m 3 0
Z
2m 2 2m 4
= sin2m 5 d
2m 1 2m 3 0
=
Z
2m 2 2m 4 4
= sin3 d
2m 1 2m 3 5 0
Z
2m 2 2m 4 4 1 2
= ( sin2 cos j0 + sin d )
2m 1 2m 3 5 3 3 0
Z
2m 2 2m 4 42
= sin d
2m 1 2m 3 53 0
2m 2 2m 4 42
= 2;
2m 1 2m 3 53
for m = 1; Z
sin d = 2:
0

5
Or
Z 2m 2 2m 4 422
; m 2
sinn d = 2m 1 2m 3 531
0 2 ; m=1
n 1n 3 42
n n 2 53 2 ; n 3
= :
2 ; n=1

6.3 Proof of Theorem 3 & Theorem 5

n R R R
By (10), we have Vn (R) = 2 nR 0
sinn 2 1 d 1 0 sinn 3 2 d 2 0
sin n 2 d n 2 : For n is even, say, n = 2m
for some m 2 N; then by Lemma 2,
Z Z Z
2 R2m
V2m (R) = sin2m 2 1 d 1 sin2m 3 2 d 2 sin n 2 d n 2
2m 0 0 0
2 R2m 2m 3 2m 5 31 2m 4 2m 6 42 2m 5 31 2m 6 42 1
= ( )( 2)( )( 2) ( ) 2
2m 2m 2 2m 4 42 2m 3 2m 5 53 2m 4 42 2m 5 53 2
= R2m
mm 1m 2 1
R2m m
=
m!
R2m m
=
m (m 1)!
R2m m
=
m (m)
or n n
Rn 2 2Rn 2
Vn (R) = n = :
( 2 )! n ( n2 )
For n is odd, say, n = 2m 1 for some m 2 N; then
Z Z Z
2 R2m 1
V2m 1 (R) = sin2m 3 1 d 1 sin2m 5 2 d 2 sin n 2d n 2
2m 1 0 0 0
2 R2m 1 2m 4 2m 6 42 2m 5 31 2m 6 42 2m 7 31 2 1
= ( 2)( )( 2)( ) ( 2)( ) 2
2m 1 2m 3 2m 5 53 2m 4 42 2m 5 53 2m 6 42 3 2
2 R2m 1 2 2 2 2
=
2m 1 2m 3 2m 5 3 1
2m m 1 R2m 1
=
(2m 1)(2m 3) 3 1
or
2m m 1 R2m 1
Vn (R) =
(2m 1)(2m 3) 3 1
n+1 n 1
n
2 2 2 R
=
n(n 2) 3 1
n
n
2 2R 1
= p
n ( n2 1)( n2 2) 3
2
1
2
n
2 2Rn 1
=
n ( n2 1)( n2 2) 3
2
1
2 ( 21 )
n
2 Rn2
= :
n ( n2 )