MathGroup Archive 2004

[Date Index] [Thread Index] [Author Index]

Search the Archive

Polylogarithm Integration


I've been told to try Mathematica, that is supposed
to be more efficient for integrations.
Here is my very first try with Mathematica 5:

  Integrate[  PolyLog[2, Exp[I*(x - y)]], {y, 0, 2*Pi} ]

Instead of 0 (at least if | Exp[I*x] | <= 1),
after a long time, I get an uggly:

  \!\(If[x >= 2\ \[Pi] ||
       x <= 0, \[Pi]\^3 - \[Pi]\ 
Log[\(-\[ExponentialE]\^\(\(-\[ImaginaryI]\)\ \
x\)\)]\^2 +
       2\ \[ImaginaryI]\ \[Pi]\^2\ Log[
           1 - \[ExponentialE]\^\(\(-\[ImaginaryI]\)\ x\)] -
       2\ \[ImaginaryI]\ \[Pi]\^2\ Log[
           1 - \[ExponentialE]\^\(\[ImaginaryI]\ x\)],
       PolyLog[2, \[ExponentialE]\^\(\[ImaginaryI]\ \((x - y)\)\)], {y, 0,
         2\ \[Pi]}, Assumptions -> \(! \((x >= 2\ \[Pi] || x <= 0)\)\)]]\)

If I replace x by any numerical real value, it works.
But if I put an imaginary numerical, for example x=I,
I get:

   I*( 4*Pi^2 - PolyLog[2,1/e] + PolyLog[3,1/e] )

that, I believe, is wrong.

Is it a bug in Mathematica 5?
Or, did I misunderstood something about polylogarithms?
How can I workaround the problem?

Thanks for your help.

  • Prev by Date: Re: simplifying first-order diff eq solution
  • Next by Date: Re: import table in mathematica
  • Previous by thread: Re: InputAutoReplacements With "-marks
  • Next by thread: Re: Mathematica fonts in LaTeX