1
$\begingroup$

I would like to evaluate the following integral:

$$\int_0^\infty \frac{\sin(ax)}{x} \frac{\sin(bx)}{x}e^{-kx}dx$$

I have made several attempts using some crude methods, the workout is way too long to format and type out, so I will make it as brief as possible.

Firstly, I evaluated the integral:

$$\int_0^\infty \frac{\sin(ax)}{x} \frac{\sin(bx)}{x}dx$$

Using integration by parts and the properties of definite integrals:

Setting $u=\sin(ax)\sin(bx)$ and $dv=x^{-2}$

Using:

  • $\int_{-\infty}^\infty \frac{g(ax)}{x}dx = \int_{-\infty}^\infty \frac{g(x)}{x}dx$
  • $\int_{-\infty}^\infty f(x)dx=\frac 1 2\int_{0}^\infty f(x)dx$
  • $2\sin a\cos b=\sin(a+b)+\sin(a-b)$

I obtained:

$$\frac1 2b\pi$$

I tried the same approach with the original question, and obtained 3 sets of integrals:

  1. $\frac{1}{2}(a+b)\int_{0}^\infty \operatorname{sinc}xe^{-kx}dx = \frac{1}{2}(a+b)\arctan(k^{-1})$

  2. $\frac{1}{2}(b-a)\int_{0}^\infty \operatorname{sinc} xe^{-kx}dx = \frac{1}{2}(b-a)\arctan(k^{-1})$

  3. $(-\frac{1}{2}k)\int_{0}^\infty \frac {\sin(ax)\sin(bx)e^{-kx}}{x}dx$ (These 3 are all summed up)

For (1) and (2), the term $a$ cancels out as expected.

I have no idea how (1) and (2) resulted in $\arctan(k^{-1})$, I just put the intrgral in wolfram alpha. I think inverse Fourier Transforms were involved here. I also don't think $\int_{-\infty}^\infty \frac{g(ax)}{x}dx = \int_{-\infty}^\infty \frac{g(x)}{x}dx$ can be used here, since we have an $e^{-kx}$ term. I went with it just to see where I can get to.

I am unable to solve integral (3).

My second attempt involved dominant and monotone convergence theorems, which was a mess. I don't even know what I was trying to do, hence I will not type it out here. I'm 100% sure it is totally wrong.

We were given some hints:

There's a common and smart trigonometric identity involved, then invoke convergence theorems. The final answer looks very neat.

I suspect the "smart trigonometric identity" is the one listed above, and my answer for the integral without the exponent did indeed look very neat. However using the same method, I wasn't able to obtain the result I needed. It should be a lot easier if convergence theorems are used, at least I assume it would.

I appreciate any help or hints. I'd like to know how this was obtained:

$\int_{0}^\infty \operatorname{sinc} xe^{-kx}dx = \arctan(k^{-1})$

Also please pick out any mistakes I've made, as I am certain my method for the question was highly fallacious and/or mathematically unrigorous. I think I'm on the right track though.
Thanks in advance!

$\endgroup$
  • $\begingroup$ Do you know Laplace transform ? $\endgroup$ – C. Dubussy May 14 '18 at 10:51
  • $\begingroup$ Yes, I do know Laplace transforms. $\endgroup$ – Hypergeometry May 14 '18 at 10:55
  • $\begingroup$ For the last one, the antiderivative involves the exponential integral function. $\endgroup$ – Claude Leibovici May 14 '18 at 10:56
  • $\begingroup$ I've never studied exponential integral functions, and I think according to the hint, it isn't required. The answer should look as neat as $\frac{1}{2}b\pi$. I'll look into that function and see what I can spin out of it though. $\endgroup$ – Hypergeometry May 14 '18 at 11:00
3
$\begingroup$

A common strategy is to write $x^{-n}=\frac{1}{\Gamma (n)}\int_0^\infty y^{n-1}e^{-xy}dy$. For example, $$\int_0^\infty\frac{\sin x}{x}e^{-kx}dx=\int_0^\infty dy\Bigg[\int_0^\infty e^{-(k+y)x}\sin x dx\Bigg]=\int_0^\infty\frac{dy}{(k+y)^2+1}.$$

$\endgroup$
  • $\begingroup$ Wow thanks, I didn't think about invoking the gamma function. I'll give it a go. Edit: That worked out nicely. $\endgroup$ – Hypergeometry May 14 '18 at 13:10
  • $\begingroup$ I computed the integral and obtained $\frac{\pi b}{2}$. Still don't know how to solve the third integral in the top post question. $\endgroup$ – Hypergeometry May 14 '18 at 14:28
  • $\begingroup$ @Hypergeometry The integral representation of $1/x$ should work. $\endgroup$ – J.G. May 14 '18 at 15:31
  • $\begingroup$ I used dominant convergence theorem and solved that part. Easy but annoyingly long. $\endgroup$ – Hypergeometry May 15 '18 at 4:41
1
$\begingroup$

I've solved the integral, the answer looks odd, and it is very very long. I'll post the results later. It's 5am the I have a class that begins at 10.

Basically, these are all wrong:

  1. $\frac{1}{2}(a+b)\int_{0}^\infty \operatorname{sinc}xe^{-kx}dx = \frac{1}{2}(a+b)\arctan(k^{-1})$

  2. $\frac{1}{2}(b-a)\int_{0}^\infty \operatorname{sinc} xe^{-kx}dx = \frac{1}{2}(b-a)\arctan(k^{-1})$

  3. $(-\frac{1}{2}k)\int_{0}^\infty \frac {\sin(ax)\sin(bx)e^{-kx}}{x}dx$ (These 3 are all summed up)

For 1 and 2, $(a+b)$ cannot be removed from the integral. The result has the inverse tangent function.

For 3, I used dominant convergence theorem which allowed me to differentiate within the integral. (I checked all conditions and they were satisfied) The result was a set of functions involving $ln(x)$.

Here's the the result looks like:

$$\int_0^\infty \frac{\sin(ax)}{x} \frac{\sin(bx)}{x}e^{-kx}dx=\frac{1}{2(a+b)}(\frac{\pi}{2}-arctan(\frac{k}{a+b})+\frac{1}{2(b-a)}(arctan(\frac{k}{b-a})-\frac{\pi}{2})+\frac{barctan(\frac{a-b}{k})}{k}+\frac{1}{2}ln((a+b)^2+k^2)-\frac{1}{2}ln((a-b)^2+k^2)$$

Which is just ridiculous. I assume it can be simplified into something nicer, or I just made a massive mistake somewhere in my 13 pages of writing. I'll know the result later today anyway. I'll post a proper solution if mine is wrong.

$\endgroup$
  • $\begingroup$ Update: we were given a hint, and the answer I obtained is likely correct. The due date has been delayed. $\endgroup$ – Hypergeometry May 15 '18 at 4:17

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.