I have tried this question by drawing the graphs of the above two functions and found that both are giving asymptote at $Y=0$. But surprisingly integration of $\ln{(x)}$ from $0$ to $1$ is defined but integration of $\text{pow}(x,\ln{(x)})$ is undefined. What can be the reason for this?
-
The same happens with $\int _0 ^1 \frac{1}{x^p}$ when one compares the cases $p=1$ and $p=1/2.$ – coffeemath Oct 17 '22 at 04:33
-
But what can be the reason for this? – DEB SANKAR ROY Oct 17 '22 at 04:51
-
It's a geometric fact. The total amount of signed area given by the integral just happens to be sometimes finite, sometimes infinite depending on the integrand. There are various tricks one can use in simple cases to see which thing happens. I don't know such tricks for complicated integrands involving logs... – coffeemath Oct 17 '22 at 04:54
1 Answers
The issue is that $x^{\ln{(x)}}$ has a pole at $x=0$, so the function is not integrable (in the standard sense) over any domain containing a neighborhood of $0$. This basically means the function isn't integrable on $(0,1]$. To clarify, if $\epsilon > 0$, then you can rewrite $\int_{0}^{1}x^{\ln\left(x\right)}dx$ as
$$\int_{0}^{1}x^{\ln\left(x\right)}dx = \lim_{\epsilon\to 0} \int_{\epsilon}^{1}x^{\ln\left(x\right)}dx,$$
which diverges to $\infty$, and you can use the substitution $x=e^u$ to convince yourself the integral equals $\infty$.
But for $\ln{(x)}$, even though it's not defined at $0$, it's integrable on $(0,1]$. For $\epsilon > 0$, we have
$$\int_0^1 \ln{(x)} d x = \lim_{\epsilon\rightarrow 0}\int_{\epsilon}^1 \ln{(x)} d x = \lim_{\epsilon\rightarrow 0}\left[x \ln{(x)} - x\right]_{\epsilon}^1= \lim_{\epsilon\rightarrow 0}(-1-\epsilon\ln{(\epsilon)} +\epsilon ) = -1,$$
where you can prove that $\displaystyle \lim_{\epsilon \to 0} \epsilon \ln{(\epsilon)} = 0$.
- 4,923
