On Mon, 22 Nov 1999, Christian Bauer wrote:
normal(q10^2*(I*Pi^2*(q10^2-q21^2)^(-1/2)*q10^2*(-(1/2)*(q10^2-q21^2)^(-1/2)*q10^2+(1/2)*q21^2*(q10^2-q21^2)^(-1/2))+1/2*I*Pi^2*q10^2)); q10^2*(1/2*I*Pi^2*q21^2*(q10^2-q21^2)^(-1)*q10^2-1/2*I*Pi^2*(q10^2-q21^2)^(-1)*q10^4+1/2*I*Pi^2*q10^2)
normal("); 0
This is caused by (x^(1/2))^2 being treated (during normal()) as foobar^2 and not being further evaluated until foobar gets re-inserted in the very end. A workaround is to call expand() before normal().
Hmm, not very transparent. Since contrary to (x^2)^(1/2), (x^(1/2))^2 can be safely transformed into x maybe something like this should be done internally by normal()? On the other hand, maybe this will become too complex. I don't know... What are we gonna do? Bye -richy. -- Richard Kreckel <Richard.Kreckel@Uni-Mainz.DE> <http://wwwthep.physik.uni-mainz.de/~kreckel/>