Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Very tricky integral

  1. Aug 17, 2009 #1
    Hi, can someone give me a hand with this "little" integral please.

    [tex]\int (\cos{k})^{t-s+1}(b-a\sin^2k)^{s/2}e^{-ik(n-1)}dk[/tex]

    t is the time, which is discrete
    s is between 0 and t
    k has domain [-pi, pi]
    n is a natural number
    a, b are complex constants

    actually this integral is the binomial expansion of this other integral

    [tex]\int (\cos{k}+\sqrt{b-a\sin^2k})^t\cos k \quad e^{-ik(n-1)}dk[/tex]

    there are some other constants, which I omitted. Also, I am omitting the coefficients and summation of the binomial expansion in the first integral.

    I'm looking for a closed form, and I tried using asymptotic approximations but it doesn't work because in general to solve it using asymptotic techniques you need to write it in this form:


    and I can't do that. Maybe there are some other asymptotic techniques that I don't know, maybe you can also give me a hand on this.

    Thanks in advance!!!
  2. jcsd
  3. Aug 18, 2009 #2
    That looks painfull. Sorry, but I have no ideas.
  4. Aug 18, 2009 #3
    Maybe Laplaces's method could work. It requires an integral with the form:

    [tex]I_n=\int_a^b \phi(k)[f(x)]^tdk[/tex]

    Take a maximum point [tex]\xi[/tex] of the function [tex]f[/tex], then the integral is:

    [tex]I_n\sim \phi(\xi)[f(\xi)]^{t+1/2}\left[\frac{-2\pi}{tf{''}(\xi)}\right]^{1/2}\textrm{ as }t\to\infty[/tex]

    I think that this method only requires [tex]f[/tex] and [tex]\phi[/tex] to be analytic.
  5. Aug 20, 2009 #4
    I am not fully sure right now how to incorporate it, but it seems as though you have "[URL [Broken] identity[/URL] in your original equation. Perhaps you can rewrite [tex]e^{-ik(n-1)}[/tex] in terms of complex numbers (cosine and sine). You would have [tex]e^{-ik(n-1)} = e^{i(1k- kn)} = \frac{e^{ik}}{e^{kn}} = \frac{cos(1k)+isin(1k)}{cos(kn)+isin(kn)}.[/tex] If that seems like a good idea, maybe someone else could go into this integration in more detail (using euler's identity).

    If that helps at all, that would be great. Otherwise, forgive me, and you can always check answers to various integrations online- Wolfram offers a website that does nearly all integrations.

    I did out your"binomial expansion" equation and the results can be found:


    Last edited by a moderator: May 4, 2017
  6. Aug 20, 2009 #5
    I tried to decompose it but the integral stays ugly :frown:

    Also, I've checked the link, but the input was incorrect, there was one parenthesis missing and solved a different integral. I fixed it, and the answer I got was "Mathematica could not find a formula for your integral. Most likely this means that no formula exists" :cry:

    Thank you very much for the reply
    Last edited by a moderator: May 4, 2017
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook