Ian McInerney

This has actually been on my todo list to implement for some time now, since I have some systems I want to simulate with MPC. I will probably be able...

Yes, I have managed to successfully get OSQP to generate code through Simulink Coder and run on an Opal-RT platform. The branch I used at the time is https://github.com/imciner2/osqp-matlab/tree/im/simulink_cg (the...

Sorry, just saw this discussion (I have been busy with the C library changes recently). I definitely can't guarantee it would have worked on 2015b unfortunately (although I do claim...

Good catch. yes, the workspace should in general have valid data at all times so that you can call the solver and have it properly operate. We have actually combined...

There is already a final stage in the algorithm that can be enabled - the polishing stage. That will identify the active set using the multipliers, and then solve an...

The conflicts have been fixed in the `develop-1.0` branch, which is the current development version of the 1.0 release. Note that this contains several breaking API changes (including the namespace...

Try lowering the primal infeasibility tolerance to something like 1e-6 or smaller and see if it returns a feasible solution them.

@vineetbansal I wonder if this failed because of the rename pointed out in https://github.com/osqp/osqp/issues/387. It appears that Intel added an soversion to the DLL/SO files, so that reporter says that...

Specifically, I wonder if we have to update https://github.com/osqp/osqp/blob/develop-1.0/configure/cmake/MKLConfig.cmake#L570 somehow to handle both a versioned/unversioned MKL library.

Because upstream QDLDL also now includes a way to query the version at compile time, I have added that into the linear solver name: ``` ----------------------------------------------------------------- OSQP v0.0.0 - Operator...