Wrong result for definite integral

*To*: mathgroup at smc.vnet.net*Subject*: [mg109954] Wrong result for definite integral*From*: Matthias <ichschmidt at gmx.net>*Date*: Wed, 26 May 2010 07:09:17 -0400 (EDT)

Dear all, I get a wrong, complex value for the following definite integral: In[27]:= Integrate[(1/x - 1/x/E^x)* PDF[NormalDistribution[1, 2], x], {x, -Infinity, Infinity}] Out[27]= -((I Sqrt[\[Pi]])/(Sqrt[2] E^(1/8))) Numerical Integration gives the following correct result: In[25]:= NIntegrate[(1/x - 1/x/E^x)* PDF[NormalDistribution[1, 2], x], {x, -100, 100}] Out[25]= 1.25243 This result does not depend on the integration bounds. Does anybody have an idea what's going wrong? Thanks, Matthias