If I submit the expression
with the assumption j integer and m integer the result is always 1. Does this strike you?
This bug seems to be corrected in Maple 6.
Whow! I verified the "simplification" in Rel. 5.1; Maple 6.01 yields the correct value
(-1)^(j-m) however.
Strange enough: (-1)^(2+j-m+n) is computed correctly by Rel. 5.1, (-1)^(2+j-m+n+r) is
wrong again. Probably one can generalize the bug to an odd and an even number of
assumed variables...