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

Recurrence relation for an integral equation

  1. Aug 1, 2011 #1
    Hi guys, would appreciate any help with this problem. It's from a graduate school entrance examination which I am practicing.

    problem statement

    When n is a natural number, the indefinite integral [itex]I_n[/itex] is defined as


    Here, a is a constant real number, and not equal to zero. Answer the following questions.

    1. Express [itex]I_{n+1}[/itex] as a recurrence equation with [itex]I_n[/itex].
    2. Derive [itex]I_1[/itex] and [itex]I_2[/itex]

    where I'm at
    I can solve part 2 relatively easily by using the trigonometric substitution [itex]x=a \tan\theta[/itex] which gives the result (feel free to ask for working if you don't believe me):



    [itex]I_{n+1}=\frac{1}{a^{2n+1}}\int(\cos\theta)^{2n} d\theta[/itex]

    these are fairly easy to solve for fixed values of n using basic integration and backwards substitutions such as [itex]\tan\theta=(x/a)[/itex] etc, giving:


    [itex]I_2=\arctan(x/a)/(2a^3)+\frac{x}{2a^2(x^2+a^2)} [/itex]

    but I don't know how to generalize to the final recurrence equation.

    attempted solution
    I tried to solve the original equation using integration by parts by writing:


    then setting:
    [itex]u=\frac{1}{(x^2+a^2)}[/itex], [itex]v'=\frac{1}{(x^2+a^2)^n}[/itex]

    [itex]u'=\frac{-2x}{(x^2+a^2)^2}[/itex], [itex]v=I_n[/itex]

    this gives:
    [itex]I_{n+1}=\frac{I_n}{(x^2+a^2)}+\int\frac{I_n 2x}{(x^2+a^2)^2}dx[/itex]

    but this doesn't agree with the calculations from above with fixed n, so it must be wrong. (I presume [itex]v=I_n[/itex] is invalid for some reason?).

    Unfortunately I don't really have any ideas on where to go from here...?
  2. jcsd
  3. Aug 1, 2011 #2


    User Avatar
    Staff Emeritus
    Science Advisor
    Homework Helper
    Gold Member

    Try the following: (I haven't checked it out myself, so it may be bogus.)

    [itex]\displaystyle I_n=\int\frac{1}{(x^2+a^2)^n}dx[/itex]
    [itex]\displaystyle =\int\frac{x^2+a^2}{(x^2+a^2)^{(n+1)}}dx[/itex]

    [itex]\displaystyle =a^2I_{n+1}+\int (x)\,\frac{x}{(x^2+a^2)^{(n+1)}}dx[/itex]

    Now try integration by parts: u = x, [itex]\displaystyle dv=\frac{x}{(x^2+a^2)^{(n+1)}}dx[/itex]​
    Last edited: Aug 1, 2011
  4. Aug 1, 2011 #3
    That is very clever, it worked great thank you!
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook