I have developed a mid-sized minimization model (176 variables, 4180 constraints).
When I execute the model on different non-linear solvers I get results, which I plug it into an excel spreadsheet for analysis.
My problem is that my more complicated constraints are not honored (i.e. when i insert the values of x1, y1, x56 and y56 into the equation it is LESS THAN 54). Example
How do I instruct the solver to honor the constraints exactly as they are written (a hard constraint) and if there is no solution, simply to say it is infeasible.
Yes. I’m using the NEOS server at the University of Wisconsin. There are a number of solvers and I test them all. Several literally come back saying optimized. I know because others say “too many iterations”, so I take the final values and use them as initial values in the next run.
Unfortunately no. I tried changing the constraints from quadratics to absolute values to simplify them even though a less precise solution would result. Similar output.