The problem was solved.
Nevertheless, for best reliability, use Mosek as solver if you have it available. Otherwise, consider following the directions at CVXQUAD: How to use CVXQUAD's Pade Approximant instead of CVX's unreliable Successive Approximation for GP mode, log, exp, entr, rel_entr, kl_div, log_det, det_rootn, exponential cone. CVXQUAD's Quantum (Matrix) Entropy & Matrix Log related functions .
First of all, thank you very much for your reply, and secondly, I have one more question that I would like to have your help with, and how to fix this situation for the expired Mosek solver license. Also, for the case where the objective function in sdp is log det(W), where W is the matrix, how should it be modified without a mosek solver, so that the results of the problem are more accurate
Follow the instructions in the link om y previous post. However, for this particular problem, CVX claims to have solved your problem accurately (but perhaps might not in problem instances having different input data).
As for Mosek license, please seek help at https://groups.google.com/g/mosek or send a help request to Mosek.
Find the installation package for CVXQUAD.Tahnk you very much.