Disciplined convex programming error: Invalid numeric values (NaNs) may not be used in CVX expressions.

[Mark_L_Stone]

This means that one or more input data values is NaN. So look at the input data values and find the offending data.

Are you using an iterative process in which one or more optimal variable values from a CVX problem are used to generate the input data for the next CXV problem? If so, if a problem is not solved to optimality (for example, infeasible, unbounded, or solver encounters numeric difficulties), CVX will populate the variable values as NaN, which could then result in NaN in the input data for the next CVX problem. You need to look at the solution status, and perhaps also be aware of the possibility of wilder and wilder solotuons eventually leading to CVX problem failure for some reason.

Related: See

The result obtained using the SeDuMi solver is inaccurate. The optimal solution is not as good as before the optimization.

Unsafeguarded (no line search or trust region)Successive Convex Approximation (SCA) or generally, alternating variables optimization are unreliable. It might not descend (for minimization problem), i.e., iterates could get worse. It might not converge to anything; and if it does converge, it might not be to a local optimum of the original problem, let alone a global optimum. The solution of successive iterations, and therefore subproblem inputs, can become wilder and wilder, until at some point …