Louis Bethune

I took a look; I will be honest: I couldn't understand the error. I don't know if it's purely numeric, a failure of our implementation, or a failure of yours....

Your benchmark has `900` primal variables and more than `10,000` constraints without obvious sparsity pattern to exploit. Ours unit tests do not cover instances this big. We should have conduct...

> Using `init_params = qp.init_params(params, ...)` helps so I feel this is just an issue with the default (zero) initialization. It would be nice to be able to give an...

Hi Basant1861 When you differentiate a Jaxopt solver it will attempt - whenever possible - to differentiate with implicit differentiation. Implicit Differentiation is only possible if the argument you are...

It is certainly possible, but what is the advantage of doing so compared to just using `BoxOSQP` instead ? I'm afraid that the fix is not straighforwarsd since it may...

Hi Illviljan Sorry for the cryptic error message. The error comes from the fact that the matrix `A_ = jnp.array([[]], dtype=float).T` is not a valid linear operator. If you don't...

That's true ; but using OSQP when you don't have constraints is overkill. In this case OSQP algorithm degenerates toward an inefficient way to solve a linear system. As argued...

Can you try to promote the `params_ineq=(-1, 1)` tuple to `float32` by default? Tell me how it's going.

Would you mind sharing your minimal (not) working example in Colab? Thanks in advance.

You did not gave me a Colab link. So, I copy/pasted the code in Colab, add a few imports, and in Colab, it works! There are no errors... which version...