Hi all,
I am trying to minimize an objective function where one of the components is trace_inv(G.’*G), where .’ denotes transpose (using matlab syntax) and I get the following back
"Only scalar quadratic forms can be specified in CVX"
I understand that CVX only takes specific functions and doesn’t accept other expressions not in DCP ruleset. I have also read the following post Trace of $S^{-2}$ in cvx which is very similar to my question but it doesn’t seem to help me much. I know I have to formulate my current expression into one that CVX can tolerate, but I am not sure how tho.
I was able to show that trace_inv(G.’*G) = ||vec(G^-1)||^2 but this expression is also not CVX friendly due to the inverse. I am not here for a solution to my problem (that would be nice tho ) but guidance to what my next step should be. I have been at this issue for a while and have started to become frustrated.
Help and comments are much appreciated.
EDIT: I should have mentioned G is a matrix.
Thank you,
Perplexabot