Finding global minimum with fseminf

Hello

I'm trying to solve optimization problems with fseminf, but as we know it finds local minimum. My question is: Is it a good strategy to call fseminf lots of times with different starting values? Is finding the/one global minimum depends only on the starting point?

Sometimes it's possible that my optimization problem has infinite number of global minimum(maybe always, not sure about that, the constraints contain sin and cos function).

In my optimization problem I have a decision variable that is between 0 and 2*pi, the rest of the variables can have any positive value.

The constraints are smooth functions most likely.

Answer 1

[Matt J]

Is it a good strategy to call fseminf lots of times with different starting values?

It's not a bad strategy, but no strategy is guaranteed.

Is finding the/one global minimum depends only on the starting point?

Yes.

The constraints are smooth functions most likely.

They had better be. It is a requirement of fseminf.