Mosek warned about near zero elements in the input data, and wasn’t able to reliably conclude it found the optimum. In such cases, CVX considers the solver to have failed and returns NaN for CVX variablea and optimal objective value.
Either you have bad numerical scaling to begin with, or the output of CVX problems becomes wilder and wilder, which leads to numerically terrible optimization problems. Read my posts on this forum pertaining to the perils of (unsafeguarded) Successive Convex Approximation (SCA).
This seems similar to your previous question. Hello,how to deal with it?Status: Infeasible Optimal value (cvx_optval): -Inf