From 1703d33ed1b35c2f22dad3df547fb386aed14a3b Mon Sep 17 00:00:00 2001 From: Gleb Belov Date: Thu, 30 May 2024 13:17:06 +1000 Subject: [PATCH] Conic doc: cvt:socp, cvt:socp2qp, solvers list #192 --- doc/source/modeling-expressions.rst | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) diff --git a/doc/source/modeling-expressions.rst b/doc/source/modeling-expressions.rst index bdf89c819..aaf4b4c47 100644 --- a/doc/source/modeling-expressions.rst +++ b/doc/source/modeling-expressions.rst @@ -508,7 +508,8 @@ Conic optimization ************************************** Some solvers can handle conic constraints with tailored algorithms: -Mosek, Gurobi, COPT. Note that general non-linear solvers accept them too, +Mosek, Gurobi, COPT, SCIP, CPLEX, Xpress. +Note that general non-linear solvers accept them too, but might not provide any specialized methods. See `conic examples `_ at Google Colab. @@ -544,10 +545,10 @@ complaining:: The problem contains both conic and nonlinear constraints. -In this case, setting :ref:`option ` ``cvt:socp=0`` results in second-order conic -constraints being passed to the solver as quadratics, even if -the solver has native SOCP API. This gives the solver a chance to recognize -SOCP forms in preprocessing. +In this case, change the model to use standard SOCP forms, +or consider playing with +the :ref:`options ` ``cvt:socp`` and ``cvt:socp2qp``. + Exponential cones ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^