|
I have a problem where the dimensionality can be very large (on the order of thousands) and the function has multiple minima. On the plus side, it is a very smooth function and I can compute the derivatives easily. I am looking for a method to find the global minimum. It doesn't have to super fast but it does need to be robust. I have tried a population based methods (DE and variants) but I have not been able to get them to work too well. They often get trapped in a local minimum. I am wondering if there are any methods that might be particularly suited to these issues. More generally, I have not come across many global optimization methods which make use of gradient information. I can see why it might be of limited use but nevertheless it would be good to know if there are any such methods. Thanks for any suggestions/comments. |
