Julian Hall
Julian Hall
I still don't see what you mean, or appreciate the value of having "a clean C# code", as I'm not a C# user. Specifically, I don't see the difference between...
I'll leave the issue open and retain the branch, but I believe that what I've done works, and @LatonCypher wants to use it.
Now that I see that the QP-related (like the integrality parameter) can be made optional by passing null as default, I'm happy with the extension of `Highs_passModel`. > I also...
> Thank you for all your contribution since yesterday. I have not (**now?**) completed and integration and tested all examples provided by Matlab here Quadratic programming - MATLAB quadprog (mathworks.com)...
@LatonCypher Thanks for your file, but it breaks the API in several ways, so we can't use it. Note that `passHessian` will always be necessary since an LP can be...
Good point. I'll review your PR in due course. Thanks
Thanks, I'll have a look at where they occur. Note that we did remove one memory leak in going to v1.7.2
That's very helpful. We didn't write cuPDLP-C, so I guess we should have checked it for memory leaks. Note that since cu-PDLP-C is using only a CPU implementation, its performance...
This is very interesting. I've yet to see an instance where PDLP is meaningfully faster - not that I've done any systematic benchmarking. At the moment it's not possible to...
> By the way it would be nice and helpful to have a rough explanation of the solver in the [documentation](https://ergo-code.github.io/HiGHS/dev/) with details like which anti cycling rules are in...