Revert "Merge branch 'master' into release/1.1.x"

This reverts commit 99f80f0, reversing
changes made to 2343d00.

CHANGELOG.md

@@ -1,4 +1,3 @@
-
 # v1.1.1
 - [#864](https://github.com/helmholtz-analytics/heat/pull/864) Dependencies: constrain `torchvision` version range to match supported `pytorch` version range.
 
@@ -211,8 +210,6 @@ Example on 2 processes:
 - [#664](https://github.com/helmholtz-analytics/heat/pull/664) New feature / enhancement: distributed `random.random_sample`, `random.random`, `random.sample`, `random.ranf`, `random.random_integer`
 - [#666](https://github.com/helmholtz-analytics/heat/pull/666) New feature: distributed prepend/append for `diff()`.
 - [#667](https://github.com/helmholtz-analytics/heat/pull/667) Enhancement `reshape`: rename axis parameter
-- [#678](https://github.com/helmholtz-analytics/heat/pull/678) New feature: distributed `tile`
-- [#670](https://github.com/helmholtz-analytics/heat/pull/670) New Feature: `bincount()`
 - [#674](https://github.com/helmholtz-analytics/heat/pull/674) New feature: `repeat`
 - [#670](https://github.com/helmholtz-analytics/heat/pull/670) New Feature: distributed `bincount()`
 - [#672](https://github.com/helmholtz-analytics/heat/pull/672) Bug / Enhancement: Remove `MPIRequest.wait()`, rewrite calls with capital letters. lower case `wait()` now falls back to the `mpi4py` function

README.md

@@ -63,7 +63,7 @@ Requirements
 ------------
 
 Heat requires Python 3.7 or newer.
-Heat is based on [PyTorch](https://pytorch.org/). Specifically, we are exploiting
+Heat is based on [PyTorch](https://pytorch.org/). Specifially, we are exploiting
 PyTorch's support for GPUs *and* MPI parallelism. For MPI support we utilize
 [mpi4py](https://mpi4py.readthedocs.io). Both packages can be installed via pip
 or automatically using the setup.py.

heat/core/_operations.py

@@ -419,10 +419,7 @@ def __reduce_op(
     else:
         output_shape = x.gshape
         for dim in axis:
-            if not (
-                partial.shape.numel() == 0 and partial_op.__name__ in ("local_max", "local_min")
-            ):  # no neutral element for max/min
-                partial = partial_op(partial, dim=dim, keepdim=True)
+            partial = partial_op(partial, dim=dim, keepdim=True)
             output_shape = output_shape[:dim] + (1,) + output_shape[dim + 1 :]
         if not keepdim and not len(partial.shape) == 1:
             gshape_losedim = tuple(x.gshape[dim] for dim in range(len(x.gshape)) if dim not in axis)
@@ -442,7 +439,7 @@ def __reduce_op(
             balanced = True
             if x.comm.is_distributed():
                 x.comm.Allreduce(MPI.IN_PLACE, partial, reduction_op)
-        elif axis is not None and not keepdim:
+        elif axis is not None:
             down_dims = len(tuple(dim for dim in axis if dim < x.split))
             split -= down_dims
             balanced = x.balanced