sweichwald

Thanks! This is something we should take into account for the backend change that is to come :-)

Hi there -- we are always welcoming pull requests! I would expect it to be quite doable. It should be possible to draw inspiration from our Stiefel manifold implementation as...

Hi there and thanks for your interest in contributing! We are happy to provide help and guidance. The workaround you mention should work. Regarding the addition of a new stochastic...

Thanks! For clarification: Does this supersede #119 #121 #122 or is there something in those PRs that did not make it into this PR?

Hey @antoinecollas – sorry for the long turnover and thans for your contributions! I remember @semlen spotted a bug in the psd inner product #128 – does this carry over...

Stopping by to leave [this cross-ref](https://github.com/pymanopt/pymanopt/pull/126#issuecomment-907583067) here; maybe that's interesting @yuvalH9? It appears @Raph-AI is using a GPU/pymanopt/tensorflow2 combo :upside_down_face:

Hi. We have implemented the above manifold, currently only supporting the autograd autodiff backend. But theano and tensorflow should be fixed soon (~1-2 weeks). You can install this updated version...

FYI #126 is merged now (and part of pymanopt 0.2.6rc1). As @Raph-AI mentions, with the TF2 back end comes some GPU support. Regarding pytorch/GPU/pymanopt – we are very welcoming of...

@chovine You are very welcome to turn this into a WIP PR for reviewing, discussing about tests, ...

Hey @mraveri – yes, this is a reasonable suggestion, thanks! Currently, this solver only stops if the number of iterations or costevals is exhausted: https://github.com/pymanopt/pymanopt/blob/master/pymanopt/solvers/particle_swarm.py#L119 In fact, the maxtime stopping...