Random Walk Matrix Optimisation

Does any one know how to perform a random walk method of optimization with matrices in MATLAB?

I have an initial matrix Q, an eigenvalue matrix L and an eigen vector V such that R=VLV'. From these I need to solve the following problem:

minimize : |Q-R| (wrt to L) such that 1. Tr(L)=a (a some constant) 2. Det(L)>0 3. R (which is VLV') is Positive semidefinite (or eig®>=0)

If anyone has any clue how I can go about solving this please let me know.

Thanks, Stefan.

Answer 1

[Torsten]

help fmincon

Best wishes

Torsten.

[Stefan Cutajar]

Thanks Torsten for your reply.

I have been trying to figure out how to apply fmincon to matrix optimisation but it doesn't seem to be feasible could you please elaborate how I could insert matrices as variables, constraints etc?

[Torsten]

You will have to break everything down to the level of the matrix entries (which are scalars).

Best wishes

Torsten.