Paul Goulart

Thank you for this, it is very helpful. We will have a look and clean up the edge cases. It seems that you do not test QP problems directly. Is...

Thanks again. We have made some stability improvements and added exp and pow cones in v0.3.0. I have made a PR to ConvexTests.jl with this change.

I am happy to implement this, but maybe there is also some confusion here about defining cone constraints. Assuming that you are modelling all linear inequalities, you don't need to...

Yes, the above would also work with equalities, i.e. you can have `[Clarabel.ZeroConeT(m), Clarabel.NonnegativeConeT(n)]` or similar. You can also have a mix or multiple instances of these cones in arbitrary...

Thanks. We are aware of a number of similar issues that effect solver convergence at high tolerance. We have implemented some improvements for v0.3 that solve the problem you provided,...

It's possibly the same underlying problem. The main issue is that v.0.2 is not doing enough to protect against poor conditioning at high tolerances. So in the first case it...

@harshangrjn We have released v0.3.0 with some stability improvements. Can you give a try again and let us know if the problems are resolved?

You don't have MKL on your platform, so compile with `-DENABLE_MKL_PARDISO=FALSE`.

Yes, just set the lower and upper bounds equal. The solver will recognise that as an equality and act accordingly.

Version 1.0 of OSQP will have improved functionality to support other linear solver types, including GPU-based solvers. It is not released yet though. Some info can be found here: https://github.com/oxfordcontrol/osqp/issues/30