I'm trying to prove this formula from here, that the volume of a n-ball with radius 1 (let's call it $B_n$) is: $$\frac{\pi^{n/2}}{\Gamma(\frac{n}{2}+1)}$$
However, I come to the wrong result and I cannot find the mistake.
The intersection of an n-ball with a hyperplane is an $(n − 1)$-ball. Therefore:
$$\text{vol}(B_{n+1}) = \int_{-1}^1 \text{vol}((B_{n} \,\,| \,\,\text{Radius = }\sqrt{1-x^2})) \,\,dx =$$
$$2 \int_{0}^1 \text{vol}((B_{n} \,\,| \,\,\text{Radius = }\sqrt{1-x^2})) \,\,dx$$
Now I want to find $\text{vol}((B_{n} \,\,| \,\,\text{Radius = }\sqrt{1-x^2}))$. That is per definition $\int_{-1}^1 \,\,1_K$ where $K = \{(x_1, ..., x_{n}) \,\,|\,\, x_1^2 + ...+ x_n^2 \leq \sqrt{1-x^2} \}$.
It holds: $ x_1^2 + ...+ x_n^2 \leq \sqrt{1-x^2} \Leftrightarrow \sum \frac{x_i^2}{\sqrt{1-x^2}} \leq 1 $
Now I do the transformation $x_i \mapsto x_i \cdot (\sqrt{1-x^2})^{1/2}$. The determinant of the Jacobian is $(\sqrt{1-x^2})^{n/2}$.
So I get:
$$2 \int_{0}^1 \text{vol}((B_{n} \,\,| \,\,\text{Radius = }\sqrt{1-x^2})) \,\,dx = $$
$$2 \int_{0}^1 \text{vol}(B_{n}) \cdot (\sqrt{1-x^2})^{n/2} \,\,dx = $$
$$2 \int_{0}^1 \text{vol}(B_{n}) \cdot (1-x^2)^{n/4} \,\,dx$$
Here I would use induction now.
But above formula isn't correct. For $n=2$ it's correct, but for $n=3$ it would state that $$\text{vol}(B_{3}) = 2 \int_{0}^1 \text{vol}(B_{2}) \cdot (1-x^2)^{3/4} \,\,dx = 2 \pi \cdot \int_{0}^1 (1-x^2)^{3/4} \,\,dx$$
which according to Wolfram Alpha doesn't equal $\frac{3}{4}\pi$, which would be the correct answer.
You have two errors. The formula for volume is wrong by a factor of 2. The volume is proportional to $r^n$, not $r^{n/2}$ $$\text{vol}((B_{n} \,\,| \,\,\text{Radius = }\sqrt{1-x^2}))$$ The second error is that if you want to calculate the volume of the 3D sphere, the $n$ on the right hand side is $2$, not $3$. So $$\text{vol}(B_{3}) = 2 \int_{0}^1 \text{vol}(B_{2}) \cdot (1-x^2)^{2/2} \,\,dx=2\pi\frac23$$