the subject says it all; csdp is a Sage package that can provide a more efficient than cvxopt backend. Also, another solver that
can be hooked up is DSDP,
which has a licence that makes it possible to bundle it with Sage.

Moreover, DSDP may be interfaced via cvxopt, so this should be very easy indeed.

• Build cvxopt with dsdp #35202

All of these (as well as cvxopt) backends would have a precision limitation coming from the fact that they all work with RDF numbers. Implementing a robust (in the sense of precision) SDP solver is doable, but the precision has to be controlled, as explained in de Klerk & Vallentin (2016).

• Add SDP backend using CVXPY #31962 Package cvxpy, add SDP backend

CC: @mkoeppe

Component: numerical

Issue created by migration from https://trac.sagemath.org/ticket/22218