search

helmholtz-analytics / heat

Distributed tensors and Machine Learning framework with GPU and MPI acceleration in Python
https://heat.readthedocs.io/
MIT License
211 stars 54 forks source link

vectorized sorting #363

Open TheSlimvReal opened 5 years ago

TheSlimvReal commented 5 years ago

Related When specifying an axis for the unique function, the array is searched for unique vectors along this axis. However a sorted result is difficult to compute when split != axis. Because this feature might also be interesting on its own it would make sense to implement it as a standalone function.

Feature functionality Compared to the classic sort the vectorized sorting should keep the values in one vector consistent and only sort the vectors along the sorting axis. The sorting should give the highest priority for the first value, if on this level there are equal values the next value should be compared and so on.

Additional context Numpy uses some kind of vectorized sorting in their unique function when specifying the axis argument.

ClaudiaComito commented 1 year ago

Will review this feature request when #749 is finished.

Reviewed within #1109