MathGroup Archive 1992

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

Search the Archive

help on integration

Dear Mathgroup,

I have the following function that I wish to integrate over (0,q), 

In[1]:= f[q_]=(L^a/Gamma[a])^2 Exp[-L(c s-q^2)/(c-q)](q(s-q))^(a-1)

            2 a  -1 + a         -1 + a
           L    q       (-q + s)
Out[1]= ----------------------------------
         (L (-q  + c s))/(c - q)         2
        E                        Gamma[a]

In[2]:= Integrate[f[q],{q,0,s}]

Out[2]= Integrate[
       -(L c) - L q + (-(L c ) + L c s)/(-c + q)  2 a  -1 + a         -1 + a
      E                                          L    q       (-q + s)
>     ----------------------------------------------------------------------, 
>    {q, 0, s}]

Mma not surprisingly cannot perform this integration directly.
I was wondering if anyone can recognise this as being of a standard 
form that I can coerce Mma to perform.  Any tips would be most

Incidentally, The application arose from the convolution of two
gamma random variables (correlated, not independent).  I noticed
that the ContinuousDistributions.m Package uses an unusual 
definition of the parameter lambda which needs to be inverted to
obtain the more usual definition for the density, i.e.

In[3]:= <<DataAnalysis/ContinuousDistributions.m

In[4]:= Density[GammaDistribution[alpha,1/lambda],t]

              alpha  -1 + alpha
        lambda      t
Out[4]= -----------------------
         lambda t
        E         Gamma[alpha]

Peter Thomson,

School of Mathematics and Statistics,
University of Sydney  NSW  2006

email: thomson_p at

  • Prev by Date: Re: What should Mma be?
  • Next by Date: Editable Postscript
  • Previous by thread: Font problem and Postscript problem
  • Next by thread: help on integration