I want to know how to solve the problem with this error message on the linux server, and it shows that the successive convex approximation method still be employed to solve the problem
What version of CVX and Mosek are you running on Linux? To avoid CVX’s Successive Approximation method, you should be running CVX 2,2 with Mosek 9.x (not 8.x or 7,x).
Also, CVX warns of near zero elements. it appears that you are getting away with that. But the same near zero elements, perhaps due to bad scaling, could be too much for the Successive Approximation method to sucessfully handle, even if Mosek is the underling solver called by CVX.
The output shows CVX 2.1. You need to use CVX 2.2 in order to use Mosek’s native exponential cone capability and avoid CVX"s Successive Approximation method.