How do I compute the integration for $a>0$,

$$ \int_0^\pi \frac{x\sin x}{1-2a\cos x+a^2}dx? $$

I want to find a complex function and integrate by the residue theorem.

  • 2
    $\begingroup$ Are you interested in alternative solutions that don't use complex variables? $\endgroup$ – David H Jun 6 '14 at 3:59

Since it hasn't been specifically objected to yet, here is a solution that doesn't rely on complex variable methods.

We shall make use of the Fourier sine series,

$$\frac{a\sin x}{1-2a\cos x+a^2}=\begin{cases} \sum_{n=1}^{\infty}a^{n}\sin{(nx)},~~~\text{for }|a|<1,\\ \sum_{n=1}^{\infty}\frac{\sin{(nx)}}{a^{n}},~~~\text{for }|a|>1. \end{cases}$$

These series may readily be derived by taking the imaginary parts of the complex geometric series $\sum_{n=1}^{\infty}\left(ae^{ix}\right)^n$ and $\sum_{n=1}^{\infty}\left(\frac{1}{a}e^{ix}\right)^n$, respectively.

By expanding the integrand in terms of these series and then swapping the order of integration and summation, a closed form may be obtained. For the $|a|>1$ case,

$$\begin{align}I(a)&=\int_0^\pi \frac{x\sin x}{1-2a\cos x+a^2}\mathrm{d}x\\ &=\int_{0}^{\pi}x\sum_{n=1}^{\infty}\frac{\sin{(nx)}}{a^{n+1}}\mathrm{d}x\\ &=\sum_{n=1}^{\infty}\frac{1}{a^{n+1}}\int_{0}^{\pi}x\sin{(nx)}\mathrm{d}x\\ &=\sum_{n=1}^{\infty}\frac{1}{a^{n+1}}\left(\frac{\sin{(n\pi)}}{n^2}-\frac{\pi\cos{(n\pi)}}{n}\right)\\ &=\pi\sum_{n=1}^{\infty}\frac{(-1)^{n+1}}{n\,a^{n+1}}\\ &=\frac{\pi}{a}\log{\frac{1+a}{a}}. \end{align}$$

For the $0<|a|<1$ cases, we can make use of the fact that if $0<|a|<1$, then $\frac{1}{|a|}>1$, thus allowing us to invoke the previous result. Hence,

$$\begin{align}I(a)&=\int_0^\pi \frac{x\sin x}{1-2a\cos x+a^2}\mathrm{d}x\\ &=\frac{1}{a^2}\int_0^\pi \frac{x\sin x}{a^{-2}-2a^{-1}\cos x+1}\mathrm{d}x\\ &=\frac{1}{a^2}\frac{\pi}{a^{-1}}\log{\frac{1+a^{-1}}{a^{-1}}}\\ &=\frac{\pi}{a}\log{(1+a)}. \end{align}$$

  • $\begingroup$ Comparing your result with Random Variable's solution, there seems to be a mistake in the argument of $\log$. $\endgroup$ – Fabian Jun 6 '14 at 4:52
  • 1
    $\begingroup$ @Fabian David H's answer is for when $a>1$. Nice answer, BTW. +1 $\endgroup$ – Random Variable Jun 6 '14 at 4:55
  • $\begingroup$ @RandomVariable Thanks for clarifying. I should have specified the condition on $a$. $\endgroup$ – David H Jun 6 '14 at 4:59
  • $\begingroup$ Very nice solution! (+1) I am interested in knowing how you came up with that series or where can I find similar ones. Thanks! $\endgroup$ – Pranav Arora Jun 6 '14 at 6:23
  • 1
    $\begingroup$ @PranavArora Random Variable's answer is for the $0<a<1$ case. Mine is for the $1<a$ case. The answers are supposed to be different. ;) $\endgroup$ – David H Jun 6 '14 at 7:31

I'll first consider the case $0<a<1$.

Let $\displaystyle f(z) = \frac{z}{a-e^{-iz}}$ and integrate around a rectangle with vertices at $\pm \pi$ and $\pm \pi + iR$.

The function $f(z)$ has poles where $a-e^{-iz}=e^{\ln a + 2 \pi i n}-e^{-iz}= 0$.

That is, when $z= i \ln a - 2 \pi n$.

If $ 0<a< 1$, all of those points are in the lower half-plane.

Letting $ R \to \infty$ and going counterclockwise around the contour,

$$ \int_{- \pi}^{0} \frac{z}{a-e^{-ix}} \ dx + \int_{0}^{\pi} \frac{x}{a-e^{-ix}} \ dx + i \int_{0}^{\infty}f(\pi + i t) \ dt + \lim_{R \to \infty} \int_{-\pi}^{\pi} f(t + iR) \ dt $$

$$+ \ i \int_{\infty}^{0} f(-\pi+it) \ dy =0$$

Combing the first two integrals,

$$ \begin{align} \int_{- \pi}^{0} \frac{x}{a-e^{-ix}} \ dx + \int_{0}^{\pi} \frac{x}{a-e^{-ix}} \ dx &= -\int_{0}^{\pi} \frac{u}{a-e^{iu}} \ du + \int_{0}^{\pi} \frac{x}{a-e^{-ix}} \ dx \\ &= \int_{0}^{\pi} \frac{-x(a-e^{-ix})+ x(a-e^{ix}) }{(a-e^{ix})(a-e^{-ix})} \ dx \\ &= - 2 i \int_{0}^{\pi} \frac{x \sin x}{1-2a \cos x +a^{2}} \ dx \end{align}$$

Combining the third and fifth integrals,

$$ \begin{align} i \int_{0}^{\infty} f(\pi + it) \ dt - i \int_{0}^{\infty} f(-\pi + it) \ dt &= i \int_{0}^{\infty} \frac{\pi + it}{a-e^{-i(\pi + it)}} \ dt - i \int_{0}^{\infty} \frac{- \pi + it}{a-e^{-i( -\pi + it)}} \ dy \\ &= i \int_{0}^{\infty} \frac{\pi + it}{a +e^{t}} \ dt - i \int_{0}^{\infty} \frac{- \pi + it}{a+ e^{t}} \ dt \\ &= 2 \pi i \int_{0}^{\infty} \frac{1}{a+e^{t}} \ dt \\ &= 2 \pi i \int_{0}^{\infty} \frac{e^{-t}}{1+ae^{-t}} \ dt \\ &= - \frac{2 \pi i}{a} \ln (1+ae^{-t}) \Big|^{\infty}_{0} \\ &= \frac{2 \pi i}{a} \ln(1+a) \end{align}$$

And the fourth integral vanishes.

So we have

$$ - 2i \int_{0}^{\pi} \frac{x \sin x}{1-2a \cos x +a^{2}} \ dx + \frac{2 \pi i}{a} \ln(1+a) = 0$$


$$ \int_{0}^{\pi } \frac{x \sin x}{1-2a \cos x +a^{2}} \ dx = \frac{\pi \ln (1+a)}{a}$$

The case for when $a >1$ is similar.

The only difference is that there is now a pole inside of the contour at $z=i \ln a$ with residue $ \displaystyle \frac{\ln a}{a}$.

And therefore,

$$- 2i \int_{0}^{\pi} \frac{x \sin x}{1-2a \cos x +a^{2}} \ dx + \frac{2 \pi i}{a} \ln(1+a) = 2 \pi i \frac{\ln a}{a} $$


$$ \int_{0}^{\pi} \frac{x \sin x}{1-2a \cos x+a^{2}} \ dx = \frac{\pi \ln (1+a)}{a} - \frac{\pi \ln a}{a} = \frac{\pi \ln \left(\frac{1+a}{a} \right)}{a}$$

  • $\begingroup$ I tried to answer this question by considering this integral. If I differentiate it with respect to $n$ and taking $b=1$, I get the same integral as this question but why the result in RHS is not similar with yours nor David's answer? $\endgroup$ – Tunk-Fey Jun 6 '14 at 12:56
  • $\begingroup$ After Differentiating both sides with respect to $n$, I got $$ \int_0^\pi\frac{x\sin nx}{a^2-2ab\cos x+b^2}\, dx=\left\{ \begin{array}{l l} \left(\frac{a}{b}\right)^{n}\cfrac{\pi}{a^2-b^2}\ln\left(\frac{a}{b}\right) &; \quad \text{if }0<a<b,\\ \\ \left(\frac{b}{a}\right)^{n}\cfrac{\pi}{b^2-a^2}\ln\left(\frac{b}{a}\right) &; \quad \text{if }0<b<a. \end{array} \right. $$ But it's not equal to your answer. Do you know why? $\endgroup$ – Tunk-Fey Jun 6 '14 at 13:05
  • 1
    $\begingroup$ @Tunk-Fey In that other problem $n$ is an integer, not a real number. And all of the answers treat $n$ as an integer. So you can't differentiate with respect to $n$. $\endgroup$ – Random Variable Jun 6 '14 at 13:56
  • $\begingroup$ Ohh, I didn't know that fact. Thanks for pointing it out. +1 for your answer. $\endgroup$ – Tunk-Fey Jun 6 '14 at 15:25
  • $\begingroup$ @Tunk-Fey Thanks. Checking with Wolfram Alpha, the answer to the other integral is only valid if $n$ is a positive integer. $\endgroup$ – Random Variable Jun 6 '14 at 16:12

$\newcommand{\+}{^{\dagger}} \newcommand{\angles}[1]{\left\langle\, #1 \,\right\rangle} \newcommand{\braces}[1]{\left\lbrace\, #1 \,\right\rbrace} \newcommand{\bracks}[1]{\left\lbrack\, #1 \,\right\rbrack} \newcommand{\ceil}[1]{\,\left\lceil\, #1 \,\right\rceil\,} \newcommand{\dd}{{\rm d}} \newcommand{\down}{\downarrow} \newcommand{\ds}[1]{\displaystyle{#1}} \newcommand{\expo}[1]{\,{\rm e}^{#1}\,} \newcommand{\fermi}{\,{\rm f}} \newcommand{\floor}[1]{\,\left\lfloor #1 \right\rfloor\,} \newcommand{\half}{{1 \over 2}} \newcommand{\ic}{{\rm i}} \newcommand{\iff}{\Longleftrightarrow} \newcommand{\imp}{\Longrightarrow} \newcommand{\isdiv}{\,\left.\right\vert\,} \newcommand{\ket}[1]{\left\vert #1\right\rangle} \newcommand{\ol}[1]{\overline{#1}} \newcommand{\pars}[1]{\left(\, #1 \,\right)} \newcommand{\partiald}[3][]{\frac{\partial^{#1} #2}{\partial #3^{#1}}} \newcommand{\pp}{{\cal P}} \newcommand{\root}[2][]{\,\sqrt[#1]{\vphantom{\large A}\,#2\,}\,} \newcommand{\sech}{\,{\rm sech}} \newcommand{\sgn}{\,{\rm sgn}} \newcommand{\totald}[3][]{\frac{{\rm d}^{#1} #2}{{\rm d} #3^{#1}}} \newcommand{\ul}[1]{\underline{#1}} \newcommand{\verts}[1]{\left\vert\, #1 \,\right\vert} \newcommand{\wt}[1]{\widetilde{#1}}$ $\ds{\int_{0}^{\pi}{x\sin\pars{x} \over 1-2a\cos\pars{x} + a^{2}}\,\dd x:\ {\large ?}.\qquad{\large a > 0}}$.

\begin{align}&\color{#c00000}{\int_{0}^{\pi} {x\sin\pars{x} \over 1-2a\cos\pars{x} + a^{2}}\,\dd x} =\half\int_{-\pi}^{\pi}{x\sin\pars{x} \over \pars{a - \expo{\ic x}}\pars{a - \expo{-\ic x}}}\,\dd x \\[3mm]&=\half\int_{-\pi}^{\pi}{x\sin\pars{x} \over 2\ic\sin\pars{x}} \,\pars{{1 \over a - \expo{\ic x}} -{1 \over a - \expo{-\ic x}}}\,\dd x =\half\,\Im\int_{-\pi}^{\pi}{x \over a - \expo{\ic x}}\,\dd x \\[3mm]&=\half\Im \int_{\verts{z}\ =\ 1 \atop {\vphantom{\Huge A}\verts{{\rm Arg}\pars{z}}\ <\ \pi}} {-\ic\ln\pars{z} \over a - z}\,{\dd z \over \ic z} \\[3mm]&={1 \over 2a}\,\Im\bracks{% \int_{\verts{z}\ =\ 1 \atop {\vphantom{\Huge A}\verts{{\rm Arg}\pars{z}}\ <\ \pi}} {\ln\pars{z} \over z - a}\,\dd z\ -\ \overbrace{% \int_{\verts{z}\ =\ 1 \atop {\vphantom{\Huge A}\verts{{\rm Arg}\pars{z}}\ <\ \pi}} {\ln\pars{z} \over z}\,\dd z}^{\ds{=\ 0}}} \end{align}

$$ \color{#c00000}{\int_{0}^{\pi} {x\sin\pars{x} \over 1-2a\cos\pars{x} + a^{2}}\,\dd x} ={1 \over 2a}\,\Im \int_{\verts{z}\ =\ 1 \atop {\vphantom{\Huge A}\verts{{\rm Arg}\pars{z}}\ <\ \pi}} {\ln\pars{z} \over z - a}\,\dd z $$

\begin{align}&\color{#c00000}{\int_{0}^{\pi} {x\sin\pars{x} \over 1-2a\cos\pars{x} + a^{2}}\,\dd x} \\[3mm]&={1 \over 2a}\Im\bracks{% 2\pi\ic\ln\pars{a}\Theta\pars{1 - a} -\int_{-1}^{0}{\ln\pars{-x} + \ic\pi \over x - a}\,\dd x -\int_{0}^{-1}{\ln\pars{-x} - \ic\pi \over x - a}\,\dd x} \\[3mm]&={\pi \over a}\bracks{\ln\pars{a}\Theta\pars{1 - a} -\int_{-1}^{0}{\dd x \over x - a}} ={\pi \over a}\bracks{\ln\pars{a}\Theta\pars{1 - a} + \ln\pars{1 + a \over a}} \end{align}

$$\color{#66f}{\large% \left.\int_{0}^{\pi}{x\sin\pars{x}\,\dd x \over 1-2a\cos\pars{x} + a^{2}} \right\vert_{\ds{\color{#c00000}{\,a > 0}}}} =\color{#66f}{\large% \left\lbrace\begin{array}{lcrcl} {\pi \over a}\,\ln\pars{1 + a} & \mbox{if} & a & < & 1 \\[3mm] {\pi \over a}\,\ln\pars{1 + a \over a} & \mbox{if} & a & > & 1 \end{array}\right.} $$


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.