Fabian Hoppe

In the branch [1031-support-distribution-of-xarray](https://github.com/helmholtz-analytics/heat/tree/1031-support-distribution-of-xarray) there is now available: * a class `DXarray` implementing sth similar as `xarray` but using `DNDarrays` as underlying objects * balace and resplit for `DXarray` *...

(just updated this PR to find out whether possibly a change in other dependencies may have resolved our issues here)

Stil open and relevant; related PR is ongoing work. (Reviewed within #1109)

@github-actions bot: it is planned, because someone is actively working on it!

@ClaudiaComito I think K-Means/Medians and Batch-parallel K-Means/Medians should be handled as separate functions since from an algorithmic point of view they are quite different. However, it's possible (and I have...

Things to be addressed: * all combinations of splits for A and b; b either vector or matrix * performance (scaling) * unit tests to be added

Things to be done: * which cases have been covered so far? (which combinations of `A.split` $\in$ {`0`, `1,`, `None`}, `b.split` $\in$ {`0`, `1,`, `None`}) * tests need to be...

With a view on #1104 #1320 it might also be usefull to include the case of batched inputs (however, for the moment it should suffice to cover the simple case...

There seems to be an error that is not due to the (meanwhile) resolved QR errors etc.

There seems to be a small bug somewhere regarding the device, so probably you habe created a tensor somewhere without specifying the device and thus the default (cpu) has been...