I'm currently using Maple to plot a 2-D (x,y) function F(I) that, at each point, is a function of a nasty integral, I(x,y,t)dt between -infinity and some constant. Maple does a good job of plotting the data and the resulting plot looks fine. The trouble is that when I ask Maple to just evaluate F for a single (x,y) rather than over the full range of (x,y) it tells me that it is not analytically tractable and it cannot produce a result.

If this is true, then how on EARTH...