Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have an integral:



$1<\frac{1}{B_{1}}\leq2 , 2\leq\frac{1}{B_{2}}$

I need to find some finite result. The result should look like ~$(\ln\frac{B_{1}}{B_{2}}+i\pi)^{2}$ or something like that. I am not really sure. Please help. I am stuck with this for a really long time (weeks). Need some hints of complex analysis I think... B1 and B2 are dependent on each other: $B_{12}=\frac{1\pm\sqrt{1-4\tau}}{2}$ , where $0<\tau<1/4$ . There must be the way to get a beautiful and finte answer.

share|cite|improve this question

If $0<x<1$, then, assuming the principal branch of the logarithm, $$ \ln(x-1)=\ln|x-1|+\pi\,i. $$ Since none of the integrals $$ \int_0^1\frac{\ln|x-1|}{x}\,dx\ ,\quad\int_0^1\frac{1}{x}\,dx $$ is convergent, the integrals in your question are not defined.

share|cite|improve this answer

From a formal point of view we have : $$f(z)=\int \frac{\log(z-1)}z dz=\log(z)\log(z-1)+\int \frac{\log(z)}{1-z}dz$$ $$f(z)=\log(z)\log(z-1)-\log(z)\log(1-z)-\int \frac{\log(1-z)}z dz$$ $$f(z)=\log(z)\left(\log(z-1)-\log(1-z)\right)-\mathrm{Li}_2(z)$$

( $\log(z-1)-\log(1-z)$ is simply $+\pi i$ for $\Im(z)>0$ and $-\pi i$ for $\Im(z)<0$ and we could have obtained the same result by integrating $\frac{\log(z-1)}z=\frac{\log\left((1-z)e^{\pm \pi i}\right)}z$ ).

Here are the pictures of the real and imaginary parts of this function :

real partenter image description here

Both parts are very smooth near $2$ but the behavior of the imaginary part near $0$ is divergent (because of the $\pm \log(z) \pi i$ part).

At this point I see only two ways to make sense of your formula $\displaystyle I(B_1,B_2)=\int_{0}^{\frac{1}{B_{1}}}\frac{dx}{x}\ln(x-1)+\int_{0}^{\frac{1}{B_{2}}}\frac{dx}{x}\ln(x-1)\ $ :

  • subtract the integrals instead of adding them with the advantage of getting different signs for $B_1$ and $B_2$ (and possibly a term $\log\left(\frac{B_1}{B_2}\right)$). You should get : $$I(B_1,B_2)=-\mathrm{Li}_2\left(\frac 1{B_1}\right)+\mathrm{Li}_2\left(\frac 1{B_2}\right)$$

  • keep your addition of integrals but 'shift' their paths in the complex plane for example by $+i\epsilon$ and $-i\epsilon$ so that both 'regularized' infinite contributions will cancel at $0$ and your sum will become simply (neglecting the $\epsilon$ term for $z \approx 2$ by replacing $\mathrm{Li}_2(z\pm i \epsilon)$ with $\mathrm{Li}_2(z)$) :

$$I(B_1,B_2)=-\mathrm{Li}_2\left(\frac 1{B_1}\right)-\mathrm{Li}_2\left(\frac 1{B_2}\right)$$

In this case : $$I\left(\frac 12,\frac 12\right)=-2\mathrm{Li}_2(2)=-\frac{\pi^2}2+2\pi i\log(2)= \frac{\pi^2}2-\log(2)^2+\left(\log(2)+i\pi\right)^2$$ looks a little like you wished but closed forms for dilogarithms are sparse (see functions.Wolfram).

In your special case $B_{12}=\frac{1\pm\sqrt{1-4\tau}}{2}$ (so that $B_1\cdot B_2=\tau$) I got these special values :

$ \begin{array} {c|l} \tau & I(B_1,B_2) \\ \hline \\ 0 & -\frac{\pi^2}6 \\ \frac 14 & -\frac{\pi^2}6+\log(2)^2 \\ \frac 12 & -5\frac{\pi^2}{48}+\frac{\log(2)^2}4 \\ 1 & -\frac{\pi^2}{18}\\ \end{array} $

UPDATE: Since $B_1+B_2=1$ let's use one of the dilogarithm identities to get a closed form (for $z>1$) : $$-\mathrm{Li}_2\left(\frac 1{\frac 1z}\right)-\mathrm{Li}_2\left(\frac 1{1-\frac 1z}\right)=\frac{\log(z-1)^2}2-\frac{\pi^2}2-i\pi\log\left(\frac{z-1}{z^2}\right)$$

Now consider $\frac 1z= B_1$ and $1-\frac 1z= B_2$ to get (I think) :

$$-\mathrm{Li}_2\left(\frac 1{B_1}\right)-\mathrm{Li}_2\left(\frac 1{B_2}\right)=\frac{\log\left(\frac {B_2}{B_1}\right)^2}2-\frac{\pi^2}2-i\pi\log\left(B_1 B_2\right)$$

Hoping this helped,

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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