Hongkai Dai

@jwnimmer-tri I tried the new Mosek 10 and one of the test `TestInitialGuess` fails. I think there is some behavioral change on the Mosek side. I sent an email to...

@jwnimmer-tri I am not sure if this is the right protocal for collaborating on dev branch, but I just rebased this branch on top of master and force pushed the...

Sounds good, I will only do merge commits then.

@drake-jenkins-bot mac-big-sur-clang-bazel-experimental-everything-release please @drake-jenkins-bot linux-focal-clang-bazel-experimental-everything-address-sanitizer please @drake-jenkins-bot linux-focal-clang-bazel-experimental-everything-valgrind-memcheck please @drake-jenkins-bot linux-focal-unprovisioned-gcc-wheel-experimental-snopt-release please.

@drake-jenkins-bot mac-big-sur-clang-bazel-experimental-everything-release please

@drake-jenkins-bot mac-big-sur-clang-bazel-experimental-everything-release please

Mathematically here is what happens: 1. since V_sym = x[0]**4 - 2 * x[0] * u[0], by calling AddSosConstraint(V_sym), we first compute the monomial basis for V_sym, which is x[0]**2....

I agree the problem is resolved here. Still curious why the stackoverflow post mentions about empty constraint, but we should resolve it in a different issue.

Thanks for reporting this with the reproducible code. I believe the problem is in drake's code in sdpa_free_format.cc file. With the default argument of drake::solvers::CsdpSolver it generates the wrong SDPA...

I just looked into this example code ``` from pydrake.all import (MathematicalProgram, Variables, Solve) prog0 = MathematicalProgram() x = prog0.NewIndeterminates(1, 'x') y = prog0.NewIndeterminates(1, 'y') l_deg = 2 l_11 =...