Fabian Hoppe

Useful functionality; easy to implement (?) if #914 has been resolved. Could be first part of a student project involving ASSET #376 (Reviewed within #1109)

In principle very useful, but maybe not of highest priority (but only since there is some workaround). (Reviewed within #1109)

This issue has been inactive for almost one year; so far there does not seem to be an analogue in PyTorch for the proposed linear operator functionality. Therefore, I close...

In principle relevant, although not of highest priority because this problem can be solved by increasing the number of processes usually. (Reviewed within #1109 ) My question @Markus-Goetz: should this...

Stil active and ongoing work. (Reviewed within #1109 )

For tiling to reduce size of potentially too large messages: see #456 including the question raised there.

In principle interesting, but for the moment out of scopy. Therefore, closed within #1109

In the future, it will be easier to manage the topic of this issue as part of other issues (e.g. #1175) in which particular test cases have to be designed...

**Suggestion for a "prototype"** A good start could be to implement sth like `diffable_Allreduce_SUM`, i.e. a AD-compatible counterpart of `ht.comm.Allreduce( ... , op=MPI.SUM)` ... this is maybe the simplemost case...

Resolved by merging of #1204