I'm running Excel Solver using the Nonlinear method (because the problem is much too nonlinear to suggest using any other method). There are over 40 variables, so there are over 80 constraints - some binary and most numerical > or<.
I have implemented a few "different" solver criteria for the end point in a time series including:
- maximize the value,
- maximize the percentage gain in that same value,
- maximize the compound gain in that same value,
- maximize the sum of the percentage gain and the compound gain (which makes no real sense but introduces a "difference" to the Solver). These are attempts at "kicking" the problem so that a broader range of possibilities might be found. And, it does work .. to a point. Of course, this is because Solver is "canned" with no internal access. And, I can't venture away from Excel for now. I'm asking if there are other methods to force Solver Nonlinear problems to move around the solution space?