Hi,

I currently have an objective function: maximize log(A’*X*A), where X is a PSD square matrix size NxN, A is a known matrix size Nx1. The function is concave but it is currently violate the DCP rule (Illegal operation: log ({convex}). I am wondering how to express this in cvx?

Thank you so much for the assistance

Something must not be as you say. As you have described it, I interpret `X`

as the only variable, in which case `A'*X*A`

is affine, and `log(A'*X*A)`

is concave and is accepted by CVX, and therefore `maximize log(A'*X*A)`

is accepted by CVX.

Based on the error message, perhaps you declared `A`

as a variable, and provided `X`

as input data? if so, and with `X`

being psd, that would not be a convex optimization problem.