Fabian Schramm

Now it should be fixed. I will manually squash my commits and force push.

Hey @jameswzhu , It seems you're not building Pinocchio with SDF support . Make sure to be on the `pinocchio3-preview` branch and enable the cmake option `BUILD_WITH_SDF_SUPPORT`. I hope this...

For now, the `BFGSStrategy` works on Euclidian space. As a first step, I checked it therefore on the `ur5-ik` example and compared it against an identity hessian. As a next...

With the dense solver backend and the following settings (notably rho = 1e-5) ``` backend = dense, eps_abs = 2e-08 eps_rel = 0 eps_prim_inf = 0.0001, eps_dual_inf = 0.0001, rho...

Hello @LiuisStupid, thank you for your interest in using ProxQP. It would be very helpful if you could share a more complete code snippet to understand your problem setup. Then...

Hey @singya, currently, our default settings for infeasibility detection are very low, see [here](https://github.com/Simple-Robotics/proxsuite/blob/main/include/proxsuite/proxqp/settings.hpp#L252C1-L253C29) ```cpp T eps_primal_inf = 1.E-12, T eps_dual_inf = 1.E-12, ``` and do not correspond to what's...

Hi @yyylllaaafff , thanks for your interest in `proxsuite`. Indeed, we have an option for warm starting the solver: ``` qp = proxsuite.proxqp.dense.QP(nz, neq, nineq) qp.settings.initial_guess = qp.settings.initial_guess = proxsuite.proxqp.WARM_START_WITH_PREVIOUS_RESULT...

Thanks for providing this example, and yes I see your point. I think that currently there is no easy way around. The reason is that we reuse the `qp` inside...

Note that currently, the pipeline `[conda:macos-latest:Debug:c++17]` is running with the option `CHECK_RUNTIME_MALLOC` which fails for several tests, meaning we still have eigen memory allocation in the dense solver. To not...

The allocations [here](https://github.com/fabinsch/proxsuite/commit/4c16380c24df96c5950599fdc92c7953eb5e0577#diff-0976b59fc541928cfb09b65187ec42f0e977e16360add3c2ec72b4c82d575938L230) and [here](https://github.com/fabinsch/proxsuite/commit/4c16380c24df96c5950599fdc92c7953eb5e0577#diff-657d936dd41e91ecfa6d5998b1d1e4744481a6b11b468a57e8424bfdef98bb63L275) can be avoided by using again `layzProduct` instead of the `operator*`. In our unittests, memory alloctions happen just above a dimension of the problem. A...