2
$\begingroup$

$1$. How to prove that for $n\in\mathbb{N}, a\in(0,1)$ one have $$f(a,0):=\int_{0<x_1,\cdots,x_n<1,\ 0<(x_1\cdots x_n)^{\frac{1}{n}}<a}dx_1\cdots dx_n=a^n \sum_{k-0}^{n-1}\frac{(-n\log(a))^k}{k!}$$ This identity arises from probability theory, but I wonder if it's solvable using calculus alone.

$2$. Moreover, for $p\in \mathbb{R}$, can we give a closed form to the generalized $$f(a,p):=\int_{0<x_1<1,\ \cdots,\ 0<x_n<1,\ 0<\left(\frac1n \sum _{i=1}^n x_i^p\right)^\frac{1}{p}<a}dx_1\cdots dx_n$$ This one is rather open. Thanks in advance!

$\endgroup$
1
1
$\begingroup$
  1. Let $f_n(a)=\idotsint\limits_{\substack{0<x_1,\ldots,x_n<1\\x_1\cdots x_n<a}}dx_1\cdots dx_n$ for $a>0$ (for $a<1$ it is your $f(a^{1/n},0)$; for $a\geqslant 1$ it is $1$). Then $$x_n=x\implies(x_1\cdots x_n<a\iff x_1\cdots x_{n-1}<a/x),$$ thus $f_n(a)=\int_0^1 f_{n-1}(a/x)\,dx=a+\int_a^1 f_{n-1}(a/x)\,dx$ for $0<a<1$. (Now induction works.)
  2. This time, for $g_n(a)=\idotsint\limits_{\substack{0<x_1,\ldots,x_n<1\\x_1^p+\ldots+x_n^p<a}}dx_1\cdots dx_n$ (and $p>0$), we get $g_1(a)=\min\{a^{1/p},1\}$ and $$g_n(a)=\int_0^{a}g_{n-1}(a-x^p)\,dx,$$ which doesn't look explicitly solvable ($n=2$ already looks clumsy).
$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.