-
Notifications
You must be signed in to change notification settings - Fork 15
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
bit of cleanup of relative entropy cone #838
Conversation
This needs @chriscoey to take a look. I can't really review this competently, other than noting that the tests pass. |
Actually @lkapelevich is the one who knows about this particular cleanup. |
Thanks @araujoms! In general we put lots of effort into optimizing the oracle code for most of the cones, but not so much for epitrrelentropytri, because of the complexity. |
Co-authored-by: Chris Coey <chriscoey@users.noreply.github.com>
Co-authored-by: Chris Coey <chriscoey@users.noreply.github.com>
Thanks for the review Chris. This is indeed a nasty cone. I'll slowly optimize it. |
Nothing from me to add! |
Well since everyone approves can someone merge it? |
Oh I assumed you were able to merge. Is that not the case? |
No, I don't have write access. |
remove
mul!
s that were multiplyingHermitian
withAdjoint
, as this is an order of magnitude slower than usual. It doesn't have a significant impact in overall performance, but drastically reduces the allocations ofdder3()
(in my test from 819.07 k to 23.77 k).