Results 1 to 3 of 3

Math Help - Residue Theorem evaluation of a Real Definite Integral

  1. #1
    Member
    Joined
    Oct 2009
    Posts
    195

    Residue Theorem evaluation of a Real Definite Integral

    \int_{0}^{\infty}\frac{\sqrt{x}dx}{x^2+2x+5}<br />

    Step 1: Write in terms of z, and define the closed curve to integrate over.

    f(z) = \frac{\sqrt{z}dz}{z^2+2z+5}<br />

    Let D be the Complex Plane with the line x \ge 0 deleted.

    \gamma_1 = te^{i\delta} with \varepsilon \le t \le R
    \gamma_2 = Re^{i\theta} with \delta \le \theta \le (2\pi - \delta )
    \gamma_3 = te^{i(2\pi -\delta )} with  -R \le \theta \le -\varepsilon
    \gamma_4 = Re^{i\theta} with  (2\pi - \delta ) \le \theta \le \delta

    Step 2: Find the poles of f(z) and calculate their residues.

    z_0 = -1+2i
    Res(f;z_0) = \frac{\sqrt{(-1+2i)}}{4i} = \frac{\sqrt{5}e^{i\theta_0}}{4i}
    where \theta_0 = Arg(-1 + 2i) = \pi - tan^{-1}(2)

    z_1 = -1-2i
    Res(f;z_0) = \frac{\sqrt{(-1+2i)}}{-4i} = \frac{\sqrt{5}e^{i\theta_1}}{-4i}
    where \theta_1 = Arg(-1 - 2i) = tan^{-1}(2) - \pi

    Step 3: Apply the Reside Theorem

    Okay... so let \Gamma = \gamma_1 + \gamma_2 + \gamma_3 + \gamma_4, then

    \int_{\Gamma}f(z)dz = 2i\pi \sum_{i=0}^{1}Res(f;z_i) = \frac{\sqrt{5}\pi}{2}(e^{i\theta_0}-e^{i\theta_1})

    and since \theta_0 = -\theta_1

    (e^{i\theta_0}-e^{i\theta_1}) = (cos\theta_0 + i sin\theta 0) - (cos\theta_0 - i sin\theta_0)

    cos(\pi -tan^{-1}(2)) + i sin(\pi -tan^{-1}(2)) - cos(\pi -tan^{-1}(2)) + i sin(\pi -tan^{-1}(2)) =

    2i sin(\pi -tan^{-1}(2)) = -sin(-tan^{-1}(2)) = sin(tan^{-1}(2)) = \frac{4}{\sqrt{5}}

    Therefore,

    \int_{\Gamma}f(z)dz = 2\pi

    But this answer is very wrong.
    Follow Math Help Forum on Facebook and Google+

  2. #2
    Newbie
    Joined
    Mar 2010
    Posts
    8
    I haven't checked your residue calculations, I'll assume those are correct, but you aren't done yet. The contour integral over \Gamma isn't supposed to be equal to the real integral you're trying to evaluate... you haven't dealt with the branch cut yet. You have two pieces of the contour that lie along the real axis, the one just above and the one just below; the integral over both of them should be equal to 2\pi, but what you're trying to evaluate is just the integral along \gamma_1 as R goes to infinity. You should be able to show that the integral on \gamma_1 is some constant factor times the integral on \gamma_2, so you can solve for the integral on \gamma_1 using the fact that their sum is 2\pi.
    Follow Math Help Forum on Facebook and Google+

  3. #3
    Super Member
    Joined
    Aug 2008
    Posts
    903
    That's a multi-residue calculation because the integrand is multivalued. We could write the residue calculations as:

    \mathop\text{mRes}\limits_{z=-1+2i}\left\{\frac{\sqrt{z}}{z^2+2z+5}\right\}=\pm1/4i\sqrt{-1+2i}

    \mathop\text{mRes}\limits_{z=-1-2i}\left\{\frac{\sqrt{z}}{z^2+2z+5}\right\}=\pm1/4i\sqrt{-1-2i}

    That makes sense right? Since the square root has two sheets then the residue is going to depend on which sheet you use to integrate around the pole. So if we start the calculation on the principal sheet right above (actually right on but that's a little confusing for beginners) so right above the positive real axis (your \gamma_1), and then go around in the positive sense, then it encloses the first pole at z_0=-1+2i, \sqrt{z_0}=r^{1/2}e^{i/2 \text{Arg}\,z_0} where Arg is is principal argument. However, as we cross the negative real axis, we leave the principal branch square root by the construction of your branch cut on the positive real axis and therefore, the residue calculation at z_1=-1-2i then must use \sqrt{z_1}=r^{1/2}e^{i/2(\text{Arg}\, z_1+2\pi)}.

    This problem should be approached from the perspective of the underlying Riemann surfaces (multisheet): draw the multisheet, identify it's principal component, the contour of integration over it, the poles, the branch-cut, and why the residue calculation requires two different residue components.
    Last edited by shawsend; March 26th 2010 at 09:40 AM. Reason: changed first to principal
    Follow Math Help Forum on Facebook and Google+

Similar Math Help Forum Discussions

  1. Real integral via the Residue Theorem
    Posted in the Math Challenge Problems Forum
    Replies: 3
    Last Post: March 30th 2010, 10:24 AM
  2. definite integral, evaluation
    Posted in the Calculus Forum
    Replies: 2
    Last Post: March 3rd 2010, 06:36 AM
  3. Need help with definite integral evaluation
    Posted in the Calculus Forum
    Replies: 3
    Last Post: February 16th 2010, 12:19 PM
  4. [SOLVED] residue theorem (definite real integrals)
    Posted in the Differential Geometry Forum
    Replies: 6
    Last Post: March 8th 2009, 10:24 AM
  5. Replies: 10
    Last Post: June 9th 2007, 12:28 PM

Search Tags


/mathhelpforum @mathhelpforum