0
$\begingroup$

Let $[a, b] \subset \Bbb R, a, b \in \Bbb R, a < b$, and let $f: [a, b] \rightarrow [0, \infty)$ be a continuous function. Define:

$K := \{(x, y, z) \in [a, b] \times \Bbb R^2 : y^2 + z^2 \le (f(x))^2 \} \subset \Bbb R^3$.

Prove that $\lambda^3(K) = \pi \int_a^b (f(x))^2 dx$.

After receiving an answer with examples on how $K$ looks like (https://math.stackexchange.com/a/2087291/3445309), I tried to solve the problem like this:

First, we note that $f$ is continuous on a compact interval, thus, it is bounded, thus, $K$ is compact itself. Therefore, we are allowed to use the principle of Cavalieri here. In order to do so, we need to define a cut surface and work in the $2$-dimensional space, so we need to handle circle surfaces right now. We define:

$K_t := K_2(\sqrt{(f(x))^2 - t^2})$ for $|t| \le (f(x))^2, \emptyset$ otherwise.

I define it like this because the radius is obviously given by $(f(x))^2$. But how do I need to apply Cavalieri now? I started with something like this:

$\lambda^3(K) = \int_a^b \lambda^2(K_t) dt = \theta_2 \int_a^b (f(x))^2 - t^2) dt$.

Note that $\theta_n$ is identical with the Lebesgue-measure of the $n$-dimensional ball with radius $1$, so $\theta_2$ is just $\pi$, which is part of the solution. But the integral there doesn't work out correctly. It would work out correctly if I would choose $t = 0$, and I think that this even makes sense when one regards that $(f(x))^2$ already depends on $x \in [a, b]$, so I would automatically get different radians. Is it true though?

This might be relevant:

https://en.wikipedia.org/wiki/Disc_integration

1 Answers 1

1

If you cut the solid $K$ at $x = x_t$ you are left with a surface, let's call it $K_t$, which is a circle of radius $f(x)$. This happens for all $x, a \leq x \leq b$. Therefore you have

$$\lambda^3(K) = \int_a^b\lambda^2(K_t)$$

But it is easy to verify that

$$\lambda^2(K_t) = \pi f(x)^2$$

given that $K_t$ is a circle of radius $f(x)$.

Therefore the desired result should be trivially verified, substituting $\lambda^2(K_t)$ into the first integral.

  • 0
    Thanks! Just realized my mistake before I took a look at your answer - wuhu and proved it the same way as you - wuhu! :D2017-01-07