1
$\begingroup$

Q. Determine the values of p and q for which the improper integral converges \begin{equation} \int_0^{\infty}x^p[\ln(1+x)]^q\,dx \end{equation} There is already a question regarding this problem (Here), but I think there is a small glitch which needs to be modified. I agree with his argument for the case $p = -1$ or $p > -1$. When proving that this integral diverges for $p < -1$, however, he used$$\lim_{x \to 0} x^cln(1+x)^q = \infty \space \text{when} \space c < 0$$ But for the case $ c = -1$ and $q = 1$ , corresponding limit approaches $1$, not $\infty$. So I think we need a modification for the case $p < -1$. Maybe I'm missing some context, too. Summarizing, what I'm curious about is as follows: 1. Is his argument correct for the case $p = -1$ and $p > -1\ ?$ 2. How can we prove for the case $p < -1 $ correctly?

$\endgroup$
1
$\begingroup$

It's probably best to organize our thoughts by considering a $pq$ plane of values. Note that both $x$ and $\ln(1+x)$ have zeros at $x=0$.

If both $p$ and $q$ are positive, this integral diverges horribly due to the infinite bound, so our entire first quadrant is out. Similarly, if $p$ and $q$ were both negative, then the integral would diverge on the other bound at $0$, so our third quadrant is gone, too. The last easy rule-outs are the axes, where either $p=0$ or $q=0$.

Split the integral into three parts:

$$= \int_0^\epsilon x^p \ln^q(1+x) dx + \int_\epsilon^T x^p \ln^q(1+x) dx + \int_T^\infty x^p \ln^q(1+x) dx$$

for some $\epsilon$ and $T$ small enough and large enough to apply our approximations, respectively. Near $0$, $\ln(1+x)\approx x$, so the integrand looks like $\frac{1}{x^{-(p+q)}}$. The integral test near $0$ tells us that $$-(p+q) < 1 \implies p+q > -1$$

The integral in the middle is always finite so we can ignore it. On the other hand, for large $x$, $\ln(1+x) \approx \ln(x)$. So consider the the right integral with that asymptotic substitution and the change of variable $u = \ln(x)$:

$$\int_T^\infty x^p \ln^q(x) dx = \int_{\ln T}^\infty u^q e^{(p+1)u}du$$

The integral only converges either if $p+1 < 0$, or if $p+1 = 0$ and $q < -1$ because the exponential dominates the behavior of the integral. However, the second case is impossible because those points $(-1,q)$ exist in the third quadrant of the $pq$ plane.

Combining our inequalities, we definitively have that $p < -1$ and $p+q > -1$. Here is a graph from Desmos showing the region of allowed values (with the horizontal axis being $p$ and the vertical axis being $q$):

1]

$\endgroup$
  • $\begingroup$ Sorry for being late!! Thanks for the answer~ I'd really appreciate it! $\endgroup$ – Kim Oct 19 at 4:42

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.