Why use cvx to solve the feasibility problem, there is an error reporting + inf, the first order Taylor expansion is used in this feasibility problem.
To be honest, I don’t know how many times the sca needs to iterate in the feasibility issue.
And the initial point of Taylor’s expansion is certainly feasible, so I really can’t understand the reason for no solution.
When unsafeguarded SCA is run, an infeasible CVX problem can be generated. That will result in
Optimal value (cvx_optval): +Inf
You can read many posts by me on the perils of unsafeguarded SCA. Or just read:
oh my god!!!This problem is really a headache, if not sca, then is there a more appropriate way in the feasibility problem?
You can use a non-convex optimization solver under YALMIP with no objective function. If a local solver finds a solution, you know it’s feasible. If not, you can try a global optimizer.