My optimization problem is shown in the following figure. CVX can be solved, but when verifying the feasibility of the solution, I found that this solutions cannot satisfy the last constraint. May I ask why this situation occurs?
Remove quiet
. Then you will see the solver and CVX output, including whether CVX claims the problem was solved. Even if solved, constraints are only satisfied to within a feasibility tolerance. If any of the non-zero input data is very extreme in magnitude (large or small), that could exacerbate the amount by which a constraint appears to not be exactly feasible.