Shane Barratt

The way we've programmed it, the threads in the pool should be non-blocking for the computationally intensive operations. So you should be able to get some benefits from parallelism using...

@akshayka

Yeah, that's exactly it. I based it off of https://papers.nips.cc/paper/8910-differentiable-ranking-and-sorting-using-optimal-transport.pdf . I'll add some references so we can connect it to prior work. On Tue, Apr 14, 2020 at 11:47...

Note: Might want to not double-count cores (https://stackoverflow.com/questions/40217873/multiprocessing-use-only-the-physical-cores).

I'll defer to @akshayka, who is the resident Tensorflow expert.

Have a few follow up questions: Is the architecture, OS, etc... of both machines the same (eg AMD vs Intel, and Ubuntu vs Mac OS)? How different are the gradients?...

I changed python 3.5 to 3.8 in travis. hopefully it passes.

Diffcp already supports warm starting: https://github.com/cvxgrp/diffcp/blob/master/diffcp/cone_program.py#L54 So you just have to save the solution when the forward ( https://github.com/cvxgrp/cvxpylayers/blob/master/cvxpylayers/torch/cvxpylayer.py#L119) method is called, and then when it's called again, pass that...

CVXPY already extracts the dual variables associated with CVXPY constraints: https://github.com/cvxgrp/cvxpy/blob/715e7ad1b1e32de7a718225edbae2c96ea6dd00f/cvxpy/reductions/solvers/conic_solvers/scs_conif.py#L218. Since this mapping is also linear, we could surely differentiate this. Should this be a part of cvxpylayers, though,...

ghcd sounds good to me. People also have the option to alias it. On Fri, Feb 14, 2020 at 4:14 PM MichaelDimmitt wrote: > ghcd sounds cool; a mashup of...