Reduced parameter optimization for optimization toolbox

Hi I use the optimization toolbox (mostly fmincon) for optimization but since I usually have rather many variables in my model, I usually make a several sub optimization (do not know if this is the proper word) there I choose a number of variable while keeping the other constant. This work fine but requires that I restart the optimization for each sub optimization. To save time I am looking for a function that can handle this for me, do someone know if there might be such a function on the File Exchange forum? I now that the ga optimization is more suitable for lager problems and I uses this toolbox to, but it is not what I am looking for here. Thankful for answers! /Thomas

Answer 1

[Matt J]

It seems impossible for any such function to do this automatically for you, or at least not in a way that would truly benefit you. In order to benefit from this, your "sub-optimizations" would have to consist of different objective functions, each one a simplification of your original objective function. Deriving the reduced sub-objective requires analytical effort on your part. I can't really see how any but a very artificially intelligent machine could be taught to do the analysis for you.