Sometimes, when solving very large GCS problems, it's necessary to use a first-order method like SCS even when a solver like Mosek or Gurobi is available. However, this also leads to SCS being used for the preprocessing step -- these smaller optimization problems can still be handled by a second-order method even if the overall problem is too large. For those steps, it would be very helpful if we could specify a different solver (much like one can specify separate solvers for the relaxation and rounding stages).
Sometimes, when solving very large GCS problems, it's necessary to use a first-order method like SCS even when a solver like Mosek or Gurobi is available. However, this also leads to SCS being used for the preprocessing step -- these smaller optimization problems can still be handled by a second-order method even if the overall problem is too large. For those steps, it would be very helpful if we could specify a different solver (much like one can specify separate solvers for the relaxation and rounding stages).