strange results of integration

*To*: mathgroup at yoda.physics.unc.edu*Subject*: strange results of integration*From*: win at wisdom.weizmann.ac.il (Wiener Zvi)*Date*: Fri, 7 Jan 94 12:28:31 +0200

Some strange results (isn't it a bug?): (* a normal probability distribution *) In[1]:= p[x_, t_] = 1/(Sqrt[2Pi t])*Exp[-x^2/(2t)]; In[2]:= Integrate[ (Exp[s]-1)*p[s,h], {s,-Infinity, Infinity}] On::none: Message SeriesData::csa not found. Out[2]= 0 In[3]:= Integrate[ Exp[s]*p[s,h]-p[s,h], {s,-Infinity, Infinity}] h/2 1 Out[3]= (-1 + E ) Sqrt[-] Sqrt[h] h In[2] and In[3] are actually the same functions but the results are different!! I also do not understand the Message. Does anybody knows a simple way to get rid of Sqrt[1/h]*Sqrt[h] assuming that h is real and positive, without Unprotect and redefining Sqrt? Zvi Wiener, Dept.of Theor. Math. The Weizmann Inst. of Sc. Rehovot, 76100, Israel.