This is the exercise of 1.18 on page 62 of pattern recognition and machine learning.
We can use the result ($\int_{-\infty}^\infty \exp(-\frac{x^2}{2\sigma^2} \, \mathrm d x) = (2\pi\sigma^2)^{1/2}$) to derive an expression for the surface area $S_D$, and the volume $V_D$, of a sphere of unit radius in $D$ dimensions. To do this, consider the following result, which is obtained by transforming from Cartesian to polar coordinates:
(1) $\prod_{i=1}^D \int_{-\infty}^\infty e^{-x_i^2} \, dx_i = S_D\int_0^\infty e^{-r^2} r^{D-1} \, dr$
Using the definition ($\Gamma(x)=\int_0^\infty u^{x-1}e^{-u} \, du$), together with ($\int_{-\infty}^\infty \exp(-\frac{x^2}{2\sigma^2} \, \mathrm d x)=(2\pi\sigma^2)^{1/2}$), evaluate both sides of the equation, and hence show that:
(2) $S_D=\frac{2\pi^{D/2}}{\Gamma(D/2)}$
Next, by integrating with respect to radius from $0$ to $1,$ show that the volume of the unit sphere in $D$ dimensions is given by:
(3) $V_D=\frac{S_D}{D}$
I have completed all the proofs but am confused about the meaning of these proofs:
- What is the meaning of equation (1), and how to convert it from left to right?
- By requiring to integrating equation (1) on both sides to obtain volume, I guess the left side of (1) is volume, but why volume is not $\pi^{D/2}$ then?
- Can someone give me the hint of the idea here to solve the problem of volumes of $n$-ball please?
Thank you very much in advance.