How to deal with a complex matrix in cvx

1

both in the whole thing i want to know the trace of the question, but i notice that the w0 are complex matrix
but for the matrx w0’*w0, i want its trace, and the trace are positive.And the whole question is a convex question, how can i transport it to meet the demand of the CVX? thanks alot

image
image1265×597 53.2 KB

falut cvx/pow_cvx
Disciplined convex programming error:
Illegal operation: pow_p( {complex affine}, {-1} )

wrong cvx/inv_pos (line 5)
y = pow_cvx( x, -1, ‘pow_p’ );

wrong do2 (line 118)
(omiga2 / m) * trace( inv_pos(rg^-1 + (1 / seitafang) * rx + (ldt / seitafang)(w0w0’)) ) )

2

trace(w'*w) can be reformulated as square_pos(norm(w,'fro'))

variables can be declared as complex or hermitian. See the CVX Users Guide.

However, it appears you may have other difficulties. have you proven this optimization problem is convex?

3

Yes, it is a convex question. actually i just repetition it from a paper :smile:, but meet some difficulties like the complex matrix :smiling_face_with_tear:

4

Many people claim their so-called convex problem is from a paper. Yet, in many cases, the problem they presented here is not actually claimed to be a convex optimization problem in the paper, even though the paper might have a (different) convex optimization problem.

Please show us a convex optimization problem, along with your proof of convexity. Please see the link in my previous post.