Rose-Hulman Undergraduate Mathematics Journal

Volume 15 Article 14
Issue 1

The Volume of n-balls

Jake Gipple
University of Missouri, [email protected]

Follow this and additional works at: https://scholar.rose-hulman.edu/rhumj

Recommended Citation
Gipple, Jake (2014) "The Volume of n-balls," Rose-Hulman Undergraduate Mathematics Journal: Vol. 15 :
Iss. 1 , Article 14.
Available at: https://scholar.rose-hulman.edu/rhumj/vol15/iss1/14
Rose-
Hulman
Undergraduate
Mathematics
Journal

The Volume of n-balls

a
Jake Gipple

Volume 15, No. 1, Spring 2014

Sponsored by

Rose-Hulman Institute of Technology

Department of Mathematics
Terre Haute, IN 47803
Email: [email protected]
a University of Missouri
http://www.rose-hulman.edu/mathjournal
 Rose-Hulman Undergraduate Mathematics Journal
Volume 15, No. 1, Spring 2014

The Volume of n-balls

Jake Gipple

Abstract. In this short paper, we compute the volume of n-dimensional balls in

Rn . The computations rely on techniques from multivariable calculus and a few
properties of the gamma function.

Acknowledgements: I would like to thank my mentor, Professor Mike Munn, for his
guidance throughout this project , as well as my family for their lasting support.
Page 238 RHIT Undergrad. Math. J., Vol. 15, No. 1

1 Introduction
For a natural number n 1, an (n 1)-dimensional sphere of radius r is the set of all points
in Rn which are a fixed distance r from a given center point. Taking the center point to be
the origin, we denote by Sn 1 (r) the (n 1)-sphere of radius r in Rn ; that is
Sn 1 (r) = {(x1 , x2 , . . . , xn ) 2 Rn | x21 + x22 + · · · + x2n = r2 }.
When n = 1, the 0-sphere consists of just the two points on the real line R1 located at r and
r. For n = 2, S1 (r) is the subset of the plane given by
S1 (r) = {(x1 , x2 ) 2 R2 | x21 + x22 = r2 }.
Graphically, S1 (r) is simply a circle of radius r centered at the origin. Note that the
interior is not included. Taking n = 3, it follows that S2 (r) is a subset of R3 given by
S2 (r) = {(x1 , x2 , x3 ) 2 R3 | x21 + x22 + x23 = r2 },
which describes a sphere of radius r centered at the origin, again not including the interior.
In higher dimensions when n 4, for example a 3-dimensional sphere, Sn (r) is more difficult
to visualize. However, using our intuition of lower dimensional spheres described above, we
can get some idea of a description for higher dimensional spheres.
Here is one way to help visualize the 3-sphere. If you take a 0-sphere, which is the
endpoints of a line segment living in R1 , and rotate it about it’s center point (or the z-axis
extending out into R3 ), you will sweep out a 1-sphere in R2 , a circle. Similarly, if you take
a 1-sphere, a circle in R2 , and rotate every point about any axis going through its center
point and lying in the plane of R2 , you will sweep out a 2-sphere in R3 . Finally, if you take
a 2-sphere, a sphere in R3 , and rotate every point about any axis going through the center
point and lying in R3 , you will sweep out a 3-dimensional sphere in R4 . This is difficult to
see, but we can carefully define these rotations using some reduction techniques and group
actions of the special orthogonal group on subsets in Euclidean space.
Recall that orthogonal matrices represent linear transformations which preserve the dot
product of vectors. They represent isometries of Euclidean space and denote rotations or
reflections. We denote the group of orthogonal matrices in Rn by
O(n) = {A : AT A = I}.
By definition, orthogonal matrices have determinant ±1. The matrices in O(n) with de-
terminant +1 represent the rotations. These are called special orthogonal matrices and are
denoted by
SO(n) = {A : AT A = I; det A = 1}.
Consider the following rotation given as a square matrix in SO(n + 1).
2 3
Ij 1 0 0
Aj = 4 0 R 0 5 for 1  j  n,
0 0 In j
RHIT Undergrad. Math. J., Vol. 15, No. 1 Page 239


cos ✓ sin ✓
where R = is a 2 ⇥ 2 (counter-clockwise) rotation matrix, Ik is the k ⇥ k
sin ✓ cos ✓
identity matrix, and j specifies where the rotation matrix is placed. Note that Aj rotates
elements in the xj xj+1 -plane but leaves all other dimensions fixed. For instance A1 is the
(n + 1) ⇥ (n + 1)-matrix
2 3
 cos ✓ sin ✓ 0
R 0
A1 = = 4 sin ✓ cos ✓ 05 .
0 In 1
0 0 1

It is easy to see that for the general case Aj , the determinant will always be 1 and since
ATj = Aj 1 , it is true that Aj is in the special orthogonal group.
To help us see how these matrices Aj generate spheres in Rn+1 , we look at the case
with n = 3 to find a parameterization of a 3-sphere in R4 . We will start with the point
P = (1, 0, 0, 0) in R4 and inductively apply our rotations. Applying the rotation A1 to P for
all values of ✓ in 0  ✓ < 2⇡. We obtain
2 32 3
cos ✓ sin ✓ 0 0 1
6 sin ✓ cos ✓ 0 07 607
6 7 6 7 = (cos ✓, sin ✓, 0, 0).
4 0 0 1 05 4 05
0 0 0 1 0

This gives a familiar parameterization of the circle S 1 ⇢ R4 lying in the x1 x2 -plane. We

can then apply the rotation A2 to our circle in the x1 x2 -plane to obtain a two-dimensional
sphere living in x1 x2 x3 -space.
2 32 3
1 0 0 0 cos ✓
60 cos sin 07 6 7
6 7 6 sin ✓ 7 = (cos ✓, cos sin ✓, sin sin ✓, 0),
40 sin cos 0 5 4 0 5
0 0 0 1 0
where 0  ✓  ⇡ and 0  < 2⇡. Notice now that the parameterization resembles spherical
coordinates. Continuing in the same manner, letting the new variable range between 0 and
2⇡, and letting , ✓ range now from 0 to ⇡, we arrive at a parameterization of the 3-sphere
in R4 :

2 32 3
1 0 0 0 cos ✓
60 1 0 0 7 6 7
6 7 6cos sin ✓7 = (cos ✓, cos sin ✓, cos sin sin ✓, sin sin sin ✓).
40 0 cos sin 5 4 sin sin ✓ 5
0 0 sin cos 0

Thus we can see that rotations in higher dimensions can be realized as the action of a linear
transformation in which there is one free parameter. This parameter does a rotation in two
dimensions and leaves all other dimensions fixed.
Page 240 RHIT Undergrad. Math. J., Vol. 15, No. 1

Continuing this construction in higher dimensions, we can produce a parametrization of

the unit sphere Sn 1 ⇢ Rn given by

x1 = cos ✓1 (1)
x2 = sin ✓1 cos ✓2 (2)
x3 = sin ✓1 sin ✓2 cos ✓3 (3)
..
. (4)
xn 1 = sin ✓1 · · · sin ✓n 2 cos ✓n 1 (5)
xn = sin ✓1 · · · sin ✓n 2 sin ✓n 1 , (6)

where 0  ✓n 1 < 2⇡ and 0  ✓i  ⇡, for i = 1, 2, . . . , n 2. We will see later how these

coordinates can be used to simplify our computations.
In this paper, we are concerned with the volume of subsets of Rn bounded by Sn 1 (r),
i.e. n-dimensional balls in Rn . More precisely, define

Bn (r) = {(x1 , x2 , ..., xn ) 2 Rn | x21 + x22 + · · · + x2n  r2 }.

We will derive a well-known formula [1] to compute the volume of Bn (r) for any natural
number n. To simplify our computations, we begin by computing the volume of a unit n-ball;
i.e. Bn (1). Throughout this paper, we will denote V (n) = Vol (Bn (1)), the volume of the
unit n-ball. We begin by proving some computational lemmas which will be useful later.

2 Lemmas
As we will see, the volume of n-balls is closely related to the gamma function. In this
section we compute various quantities related to the gamma function which will aid our
computations later in the paper.
Z 1 p
x2 ⇡
Lemma 2.1. e dx = .
0 2
Z 1
2
Proof. Setting I = e x dx, note that by Fubini’s Theorem,
0
Z 1 Z 1 Z 1 Z 1 Z 1 Z 1
x2 y2 x2 y2 x2 + y 2
I2 = e dx e dy = e e dxdy = e dxdy.
0 0 0 0 0 0

By converting to polar coordinates, taking x = r cos ✓ and y = r sin ✓, we have by Change

⇡
of Variables with 0  r < 1, and 0  ✓ 
2
Z ⇡Z 1
2 2
2
I = re r drd✓.
0 0
RHIT Undergrad. Math. J., Vol. 15, No. 1 Page 241

Furthermore, using integration by substitution with u = r2 , we have

Z ⇡ Z 1
2
2 1 u ⇡
I = e = .
0 0 2 4
p r
⇡ ⇡
Thus, taking the square root on both sides we have I = = and we are done.
4 2
Note that a rotation similar to those described in the introduction was used to calculate I 2 .
This underlying theme of rotations will keep surfacing in the computations throughout the
paper.
The gamma function was first studied in the mid 18th century by Euler and Stirling. It
has since been used in many di↵erent areas of mathematics such as complex analysis, prob-
ability, statistics, and combinatorics. For our purposes, we will define the gamma function
as (s), for s > 0, by Z 1
(s) = ts 1 e t dt. (7)
0
✓ ◆
1 p
Lemma 2.2. = ⇡.
2
1
Proof. Plugging 2
into the gamma function, we have
✓ ◆ Z 1
1 1
= t 2 e t dt.
2 0

p 1 1
Using substitution with u = t, so du = t 2 dt we have
2
✓ ◆ Z 1
1 2
=2 e u du.
2 0

Using Lemma 2.1, and substituting in for the integral, we have

✓ ◆ ✓p ◆
1 ⇡
=2 .
2 2
✓ ◆
1 p
Finishing up, we have that = ⇡, and we are done.
2
Lemma 2.3. (s + 1) = s · (s).

Proof. Plugging s + 1 into the gamma function, we have

Z 1
(s + 1) = ts e t dt.
0
Page 242 RHIT Undergrad. Math. J., Vol. 15, No. 1

Using integration by parts with u = ts and dv = e t , we have

1
Z 1
t s
(s + 1) = e t ( s) ts 1 e t dt.
0 0

1 ts
Notice that e t ts 0
goes to zero because lim e t ts = lim
= 0, because excessive
t!1 t!1 et
applications of L’Hopitals Rule will show that the denominator gets exponentially large.
And by using equation (7), we are left with

(s + 1) = s · (s),

and we are done.

3 Recursion formula for V (n)

In this section, we develop a recursion formula for the volume of the unit n-ball. Denoting
the volume Vol (Bn (1)) as V (n), we can write
Z Z
V (n) = ··· dx1 dx2 · · · dxn . (8)
x21 +x22 +···+x2n 1

We have the following recursion formula for V (n) for all n 2.

2⇡
Proposition 3.1. V (n) = V (n 2) .
n
Proof. We begin with the understanding that
Z Z
V (n) = ··· dx1 dx2 · · · dxn .
x21 +x22 +···+x2n 1

Breaking this up into an iterated integral of an (n 2)-dimension integral and a 2-dimensional

integral, we have
0 1
ZZ Z Z
B C
V (n) = @ ··· dx3 dx4 · · · dxn A dx1 dx2 ,
x21 +x22 1 x23 +···+x2n 1 x21 x22

which can also be written as

ZZ ✓ ✓q ◆◆
V (n) = Vol B n 2
1 x21 x22 dx1 dx2 . (9)
x21 +x22 1
RHIT Undergrad. Math. J., Vol. 15, No. 1 Page 243

We will now simplify the above equation by using the following

Claim: For r > 0, Vol(Bn (r)) = rn · V (n).
Proof of Claim. Recall that in any dimension, the volume of a ball of radius r can be
written as Z Z
n
Vol(B (r)) = · · · (1) dx1 dx2 dx...dxn .
x21 +···+x2n r2

Using the change of variables2formula, taking 3 x1 = ru1 , x2 = ru2 , and xn = run , we see that
r 0 ···
60 r 0 · · · 7
6 7
the Jacobian is given by det 6 .. ..
. 7 = rn .
4. 0 5
..
. r
Thus, we now have
Z Z
n
Vol(B (r)) = · · · (1)|rn | du1 du2 du...dun .
u21 +···+u2n 1

Therefore, by equation (8), and since r > 0, we have justified the claim and indeed

Vol(Bn (r)) = rn · V (n). (10)

By utilizing (10), we can now simplify the integrand of (9). Namely,

ZZ ✓ ✓q ◆◆
V (n) = Vol B n 2 2
1 x1 x2 2
dx1 dx2
x21 +x22 1
ZZ ✓q ◆n 2
= 1 x21 x22 V (n 2) dx1 dx2
x21 +x22 1
ZZ ✓q ◆n 2
= V (n 2) 1 x21 x22 dx1 dx2 .
x21 +x22 1

By converting to polar coordinates, taking x1 = r cos ✓ and x2 = r sin ✓ with 0  r  1, and

0  ✓  2⇡, we can rewrite the double integral above to yield
Z 2⇡ Z 1 n 2
V (n) = V (n 2) 1 r2 2 r dr d✓.
0 0

Now using substitution with u = 1 r2 we have

Z Z
1 2⇡ 1 n 2
V (n) = V (n 2) u 2 du d✓.
2 0 0
Page 244 RHIT Undergrad. Math. J., Vol. 15, No. 1

Treating n as a constant, and evaluating the integral, we find that

Z Z 1
1 2⇡ n 2 1 2 n 1 2⇡
V (n) = V (n 2) · d✓ · u 2 = V (n 2) · · 2⇡ · u 2 = V (n 2) .
2 0 0 2 n 0 n

3.1 The Recursion Formula for V (n) via Spherical Coordinates.

It is worth noting how the computation made above (performed using rectangular coor-
dinates) can also be done using the coordinate system developed in the introduction; i.e.
“spherical” or hyperspherical coordinates.
Recalling (1), it is possible to parameterize the n-dimensional ball Bn (1) ⇢ Rn by
x1 = r cos ✓1
x2 = r sin ✓1 cos ✓2
x3 = r sin ✓1 sin ✓2 cos ✓3
..
.
xn 1 = r sin ✓1 · · · sin ✓n 2 cos ✓n 1
xn = r sin ✓1 · · · sin ✓n 2 sin ✓n 1 ,
taking
0  r  1,
0  ✓i  ⇡, for i = 1, 2, . . . , n 2
0  ✓n 1 < 2⇡.
It then follows from the Change of Variables formula that the rectangular volume element
dV = dx1 dx2 · · · dxn can be written in spherical coordinates as
✓ ◆
@xi
dV = det drd✓1 · · · ✓n 1
@(r, ✓j ) 1in
1jn 1
n 1 n 2
=r sin (✓1 ) sinn 3 (✓2 ) · · · sin(✓n 2 ) drd✓1 · · · ✓n 1 .
Thus,
Z Z
V (n) = Vol (Bn (1)) = ··· dx1 dx2 · · · dxn ,
x21 +x22 +···+x2n 1
Z 2⇡ Z ⇡ Z ⇡ Z 1
= rn 1 sinn 2 (✓1 ) · · · sin(✓n 2 ) drd✓1 · · · d✓n 2 d✓n 1 ,
···
0 0 0 0
Z 2⇡ Z ⇡ Z 1 Z ⇡
n 1 n 2
= d✓n 1 r dr sin ✓ d✓ · · · sin ✓ d✓,
0 0 0 0
Z Z ⇡
2⇡ ⇡ n 2
= sin ✓ d✓ · · · sin ✓ d✓.
n 0 0
RHIT Undergrad. Math. J., Vol. 15, No. 1 Page 245

In the third line we used Fubini’s Theorem and dropped the index dependence of the ✓i ’s
after splitting the intreated integral into a product of integrals. In the last line, we evaluated
the first two integrals to arrive at 2⇡n
.
Going forward, keep in mind the useful relation
Z ⇡ Z ⇡
n 2 n
sin ✓ d✓ · · · sin ✓ d✓ = V (n).
0 0 2⇡
In particular (and relevant to our later calculations), since n 4 = (n 2) 2,
Z ⇡ Z ⇡
n 4 n 2
sin ✓ d✓ · · · sin ✓ d✓ = V (n 2). (11)
0 0 2⇡
R
In order to justify the final recursion formula, the following integral formula for sinm ✓ d✓
will be helpful to us. For any integer m 2, we have
Z ⇡ ✓=⇡ Z ⇡
m sinm 1 ✓ cos ✓
sin ✓ d✓ = + sinm 2 ✓d✓
0 m ✓=0 0
Z ⇡
m 2
= sin ✓d✓.
0

Note that when m is even, say m = 2k, then

Z ⇡
2k 1 2k 3 3 1
sin2k ✓ d✓ = · · · · · · ⇡. (12)
0 2k 2k 2 4 2
Similarly, when m is odd, say m = 2k + 1, then
Z ⇡
2k 2k 2 4 2
sin2k+1 ✓ d✓ = · · · · · · 2. (13)
0 2k + 1 2k 1 5 3
We are now in a position to prove the recursion formula. Combining these facts above,
and assuming without loss of generality that n is even, we get
Z Z ⇡ Z ⇡ Z ⇡
2⇡ ⇡ n 2
V (n) = sin ✓ d✓ sinn 3 ✓ d✓ sinn 4 ✓ d✓ · · · sin ✓ d✓,
n 0 0 0 0
Z Z ⇡
2⇡ ⇡ n 2 n 2
= sin ✓ d✓ sinn 3 ✓ d✓ · V (n 2), by (11),
n 0 0 2⇡
✓ ◆ ✓ ◆
2⇡ n 3 1 n 4 2 n 2
= ··· · ⇡ · ··· · 2 · V (n 2), by (12) and (13),
n n 2 2 n 3 3 2⇡
✓ ◆
2⇡ n 2 n 3 n 4 3 2 2⇡
= · · ··· · · · V (n 2), by rearranging terms
n n 2 n 3 n 4 3 2 2⇡
2⇡
= V (n 2),
n
which is precisely the conclusion of Proposition 3.1 above.
Page 246 RHIT Undergrad. Math. J., Vol. 15, No. 1

4 Computing V (n) as a sequence

Consider the sequence defined by

⇡ n/2
f (n) = 1
, for n being any natural number.
2
n+1

Proposition 4.1. The sequence f (n) satisfies the same recursion formula as V (n). Namely,

2⇡
f (n) = f (n 2) .
n
⇡ n/2
Proof. Plugging into both sides of the recursion formula, we have
( 12 n+1)

⇡ n/2 2⇡ ⇡ (n 2)/2

1
= 1
.
2
n +1 n 2
(n 2) + 1

Further simplifying the right side, we have

n
⇡ n/2 2⇡ 2 1
1
= n ,
2
n+1 ( )
2
and simplifying the left side using Lemma 2.3, we have
n
⇡ n/2 2⇡ 2 1
1 1
= n .
2
n 2
n+1 ( )
2
Further simplifying of the left side gives us
n n
2⇡ 2 1 2⇡ 2 1
n = n .
( ) ( )
2 2
Thus proving that the initially given sequence f (n) satisfies the same recursion formula as
V (n).

We can now combine what we have done to compute the volume of an n-dimensional
ball.

Proposition 4.2. For any natural number n 1 and any real number r > 0,

⇡ n/2
Vol (Bn (r)) = 1
rn .
2
n + 1
RHIT Undergrad. Math. J., Vol. 15, No. 1 Page 247

Proof. To begin, note that

V (1) = Vol B1 (1) = 2,

and

⇡ 1/2 ⇡ 1/2 ⇡ 1/2

f (1) = = ✓ ◆ = = 2.
1
+1 1 1 1 1/2
2 · ⇡
2 2 2

Since both V (n) and f (n) satisfy the same recursion formula (as verified in Proposition 4.1
above), we have that V (n) = f (n) for all n 1. Thus,

⇡ n/2
V (n) = 1
, for n being any natural number.
2
n +1

Furthermore, by equation (10), we have, for any natural number n 1,

⇡ n/2
Vol (Bn (r)) = rn · V (n) = rn · f (n) = 1
rn .
2
n + 1

5 Calculations

Here we list calculated values for V (n) which give volume of the interior of a unit-sphere in
dimensions n = 1 through n = 10. Note: to calculate Vol (Bn (r)), a ball of radius r, we need
only add an rn to the calculations below.
Page 248 RHIT Undergrad. Math. J., Vol. 15, No. 1

n V (n) = Vol (Bn (1))

1 2=2

2 ⇡ ⇡ 3.14
4
3 ⇡ ⇡ 4.19
3
1 2
4 ⇡ ⇡ 4.93
2
8 2
5 ⇡ ⇡ 5.26
15
1 3
6 ⇡ ⇡ 5.16
6
16 3
7 ⇡ ⇡ 4.72
105
1 4
8 ⇡ ⇡ 4.10
24
32 4
9 ⇡ ⇡ 3.29
945
1 5
10 ⇡ ⇡ 2.54
120

It is interesting to note that the volume begins to decrease after n = 5.

References
[1] Apostol, T. Calculus. Vol. II: Multi-variable Calculus and Linear Algebra, With Applications To Di↵erent
And Probability. Second Edition Blaisdell Publishing Co. 1969.

[2] Weeks, Je↵rey R. The Shape of Space, Second Edition. Monographs and Textbooks in
Pure and Applied Mathematics, 249. Marcel Dekker, Inc., New York, 2002.