Optimization wrongly converging to the boundary

I have a highly non-linear objective function with bounds that I am trying to optimize using 2 design variables. No matter which optimizer I use, PSO or fmincon, it seems to always converge to the upper bound of one of the variables. I am positive it is not correct because I have test results of the problem which was manually solved. The design space is convex, with the minimum well away from the boundaries.

Answer 1

[Alan Weiss]

You have given no specifics, so my answer is completely generic. Solvers converge to local optima, not necessarily global optima. See Local vs. Global Optima for definitions and suggestions.

If you want more specific advice, give more specific information, such as the objective function you have, options used, constraints, initial point or points tried, and your function call.

Alan Weiss

MATLAB mathematical toolbox documentation