I have been trying to evaluate this integral:$$\int_0^\infty xe^{-x^2}\sin(\xi x)\ dx$$ But I seem to be a little bit stuck on how to do this. I have tried partial integration by taking derivatives of $x\sin(\xi x)$ or $xe^{-x^2}$, but both times I arrived at an integral that seemed even harder to solve. An idea that kind of worked was to use the series representation of $\sin(\xi x)$ and then interchanging the summation with the integral. In that case, I would not know if the interchange is viable though ( I could not really find an integrable upper bound to use dominated convergence).

Any hints on how the interchange with summation would work or any other approach would be greatly appreciated!

  • $\begingroup$ hint: your integral is the derivative of the real part of the fourier transform of a gaussian with respect to frequency $\endgroup$ – tired Nov 25 '17 at 10:56
  • $\begingroup$ @tired Actually, thats how I arrived here since I wanted to evaluate the fourier transform. Was that a step backwards? $\endgroup$ – Jack4t3 Nov 25 '17 at 10:57
  • $\begingroup$ the searched resuld should be this here $$\frac{1}{4} \sqrt{\pi } e^{-\frac{\xi ^2}{4}} \xi$$ $\endgroup$ – Dr. Sonnhard Graubner Nov 25 '17 at 10:57
  • $\begingroup$ @Jack4t3 the FT of a Gaussian is usually shown using the completition of the square in the exponent. and yes i think your approach might go a little bit in a not so useful direction ;) $\endgroup$ – tired Nov 25 '17 at 11:01

With integrating by parts \begin{align} \int_0^\infty xe^{-x^2}\sin(\xi x)\ dx &= \dfrac{\xi}{2}\int_0^\infty e^{-x^2}\cos(\xi x)\ dx \\ &= \dfrac{\xi}{2}{\bf Re}\int_0^\infty e^{-x^2}e^{i\xi x}\ dx \\ &= \dfrac{\xi}{2}{\bf Re}\int_0^\infty e^{-(x-\frac12\xi i)^2-\frac14\xi^2}\ dx \\ &= \dfrac{\xi}{2}e^{-\frac14\xi^2}{\bf Re}\int_0^\infty e^{-(x-\frac12\xi i)^2}\ dx \\ &= \dfrac{\xi}{2}e^{-\frac14\xi^2}\dfrac{\sqrt{\pi}}{2} \end{align}


Here is a real way of arguing, that is a bit longer, but still. If you define $$ f(\xi)=-\int_0^{+\infty}e^{-x^2}\cos(\xi x)\,dx $$ then your integral is $f'(\xi)$.

But using parity arguments, we find that $$ \begin{align} f'(\xi)&=\int_0^{+\infty} xe^{-x^2}\sin(\xi x)\,d x\\ &=\frac{1}{2}\int_{-\infty}^{+\infty} xe^{-x^2}\bigl(\sin(\xi x)+\cos(\xi x)\bigr)\,dx\\ &=\frac{1}{2}\int_{-\infty}^{+\infty}\biggl[\frac{d}{dx}\Bigl(-\frac{1}{2}e^{-x^2}\bigl(\sin(\xi x)+\cos(\xi x)\bigr)\Bigr)+\frac{\xi}{2}e^{-x^2}\bigl(\cos(\xi x)-\sin(\xi x)\bigr)\biggr]\,dx\\ &=\frac{\xi}{4}\int_{-\infty}^{+\infty}e^{-x^2}\bigl(\cos(\xi x)-\sin(\xi x)\bigr)\,dx\\ &=\frac{\xi}{4}\int_{-\infty}^{+\infty}e^{-x^2}\cos(\xi x)\,dx\\ &=-\frac{\xi}{2}f(\xi). \end{align} $$ Hence $$ f(\xi)=Ce^{-\xi^2/4} $$ for some constant $C$. But $$ C=f(0)=-\int_0^{+\infty}e^{-x^2}\,dx=-\frac{\sqrt{\pi}}{2}, $$ so $$ f(\xi)=-\frac{\sqrt{\pi}}{2}e^{-\xi^2/4}. $$ Differentiating, we find that $$ f'(\xi)=\frac{\sqrt{\pi}}{4}\xi e^{-\xi^2/4}. $$


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.