If true, that seems very misleading. It would seem to me that the closest
integer could be much further off from the maximum than some other integer
that may not be as close to where the best solution is, depending upon the
nature of the curve.

Or am I missing something?

Dean

"Bill Renaud" wrote in message
. ..
After experimenting some more with the hypothetical problem that I posted
earlier, it appears that Solver ignores any Integer constraints on a
variable, then simply rounds the answer off to the nearest integer after
the convergence has finished.

For example, if I set my parabola to peak at 4.49, then Solver returns 4
as
the answer. If I set the peak at 4.51, then Solver returns 5 as the
answer.

--
Regards,
Bill Renaud