@dance858 let's use this issue to communicate 😆 .
I actually just received your WhatsApp messages but am unable to reply.
Nice work on the permuted dense matrix abstraction, I don't quite understand it yet, but will look over some code over the next days. I am in full trust of your work of course!
I had a few questions: is this permuted dense code being added in the DNLP sync to cvxpy? I saw that you added some commits to master and made a new release so just wanted to make sure. It seems like you did everything correctly though, as far as I am aware.
Also, I had a question about the bugs that Claude discovered in the PR.. I am just curious how we didn't catch them earlier? Like the param quad-form breaking for some reason. Are these bugs that Claude discovered by itself and they weren't pre-existing tests for them?
p.s. I just did the Copt talk and it went pretty well, I got to also visit their office and meet one of their solver developers in person :).
Thanks for your time and keep in touch!
@dance858 let's use this issue to communicate 😆 .
I actually just received your WhatsApp messages but am unable to reply.
Nice work on the permuted dense matrix abstraction, I don't quite understand it yet, but will look over some code over the next days. I am in full trust of your work of course!
I had a few questions: is this permuted dense code being added in the DNLP sync to cvxpy? I saw that you added some commits to master and made a new release so just wanted to make sure. It seems like you did everything correctly though, as far as I am aware.
Also, I had a question about the bugs that Claude discovered in the PR.. I am just curious how we didn't catch them earlier? Like the param quad-form breaking for some reason. Are these bugs that Claude discovered by itself and they weren't pre-existing tests for them?
p.s. I just did the Copt talk and it went pretty well, I got to also visit their office and meet one of their solver developers in person :).
Thanks for your time and keep in touch!