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

Question on erf(x) function

  1. Apr 28, 2010 #1
    If we consider the error function [tex]\mathrm{erf}(x)=\int_{0}^{x}e^{-t^2}dt[/tex]

    How can I re-express the following in terms of the erf function?

    [tex]f(x)=\int_{ib}^{x+ib}e^{-t^2}dt = \\ ?[/tex]

    I have troubles with this kind of integrals. How should I treat an integral with complex bounds?
  2. jcsd
  3. Apr 28, 2010 #2


    User Avatar
    Science Advisor

    Let s = t-ib, then the complex function will appear in the exponent, wile the s integral is from 0 to x.
  4. Apr 28, 2010 #3
    How about if you consider a closed path from zero straight up to the point ib, straight across to the point x+ib, and then diagonally down back to the origin. The integral over that path is zero since the integrand is entire. We can then write [itex]\int_0^{ib} f(z)dz+\int_{ib}^{x+ib} f(z)dz+\int_{x+ib}^0 f(z)dz=0[/itex] or [itex]\int_{ib}^{x+ib}f(z)dz=\int_0^{x+ib} f(z)dz-\int_0^{ib}f(z)dz=erf(x+ib)-erf(ib)[/itex] given your definition of erf and [itex]f(z)=e^{-z^2}[/itex].
    Last edited: Apr 28, 2010
  5. Apr 28, 2010 #4
    ...then you get


    I cannot relate that integral to the erf function.
  6. Apr 28, 2010 #5
    thanks jackmell!
    your solution is interesting!
    Unfortunately I have some troubles interpreting the erf function with a complex argument: in particular, could you explain how the limit of [itex]2(erf(x+ib)-erf(ib))[/itex] for [itex]x\rightarrow \infty[/itex] gives the result of the gaussian integral, which is [itex]\sqrt{\pi}[/itex] ?
  7. Apr 28, 2010 #6
    Hi. I'm not an expert at this but I believe this is correct:

    [itex]\lim_{x\to\infty} \int_{ib}^{x+ib} e^{-t^2}dt=\lim_{x\to\infty}\left\{\int_0^{x+ib} e^{-t^2}dt-\int_{0}^{ib} e^{-t^2}dt\right\}[/itex]

    [itex]=\lim_{x\to\infty}\left\{\int_0^x +\int_x^{x+ib}-\int_0^{ib}\right\}[/itex]

    and as [itex]x\to\infty [/itex], the center integral goes to zero. Then:

    [itex]\lim_{x\to\infty} \int_{ib}^{x+ib} e^{-t^2}dt=\frac{\sqrt{\pi}}{2}-erf(ib)[/itex]

    If you like you can parameterize the path for the second integral from the origin, straight up to the point ib by letting t=iy and dt=idy then:

    [itex]erf(ib)=\int_0^{ib}e^{-t^2}dt=i\int_0^b e^{y^2}dy[/itex]
  8. Apr 29, 2010 #7
    thanks a lot!
    now it is clear, and your solutions looks correct.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook