fix #925: ht.nonzero() returns tuple of 1-D arrays instead of n-D arrays

.github/workflows/ci.yaml

@@ -0,0 +1,44 @@
+name: ci
+
+on:
+  pull_request_review:
+    types: [submitted]
+
+jobs:
+  approved:
+    if: github.event.review.state == 'approved'
+    runs-on: ubuntu-latest
+    strategy:
+      fail-fast: false
+      matrix:
+        py-version:
+          - 3.7
+          - 3.8
+        mpi: [ 'openmpi' ]
+        install-options: [ '.', '.[hdf5,netcdf]' ]
+        pytorch-version:
+          - 'torch==1.7.1+cpu torchvision==0.8.2+cpu torchaudio==0.7.2'
+          - 'torch==1.8.1+cpu torchvision==0.9.1+cpu torchaudio==0.8.1'
+          - 'torch==1.9.0+cpu torchvision==0.10.0+cpu torchaudio==0.9.0'
+
+
+    name: Python ${{ matrix.py-version }} with ${{ matrix.pytorch-version }}; options ${{ matrix.install-options }}
+    steps:
+      - name: Checkout
+        uses: actions/checkout@v2
+      - name: Setup MPI
+        uses: mpi4py/setup-mpi@v1
+        with:
+          mpi: ${{ matrix.mpi }}
+      - name: Use Python ${{ matrix.py-version }}
+        uses: actions/setup-python@v2
+        with:
+          python-version: ${{ matrix.py-version }}
+          architecture: x64
+      - name: Test
+        run: |
+          pip install pytest
+          pip install ${{ matrix.pytorch-version }} -f https://download.pytorch.org/whl/torch_stable.html
+          pip install ${{ matrix.install-options }}
+          mpirun -n 3 pytest heat/
+          mpirun -n 4 pytest heat/

CITATION.cff

@@ -0,0 +1,68 @@
+cff-version: 1.2.0
+message: "If you use this software, please cite it as below."
+authors:
+- family-names: "Götz"
+  given-names: "Markus"
+- family-names: "Debus"
+  given-names: "Charlotte"
+- family-names: "Coquelin"
+  given-names: "Daniel"
+- family-names: "Krajsek"
+  given-names: "Kai"
+- family-names: "Comito"
+  given-names: "Claudia"
+- family-names: "Knechtges"
+  given-names: "Philipp"
+- family-names: "Hagemeier"
+  given-names: "Björn"
+- family-names: "Tarnawa"
+  given-names: "Michael"
+- family-names: "Hanselmann"
+  given-names: "Simon"
+- family-names: "Siggel"
+  given-names: "Martin"
+- family-names: "Basermann"
+  given-names: "Achim"
+- family-names: "Streit"
+  given-names: "Achim"
+title: "Heat - Helmholtz Analytics Toolkit"
+version: 1.1.0
+date-released: 2021-09-21
+url: "https://github.com/helmholtz-analytics/heat"
+preferred-citation:
+  type: conference-paper
+  authors:
+  - family-names: "Götz"
+    given-names: "Markus"
+  - family-names: "Debus"
+    given-names: "Charlotte"
+  - family-names: "Coquelin"
+    given-names: "Daniel"
+  - family-names: "Krajsek"
+    given-names: "Kai"
+  - family-names: "Comito"
+    given-names: "Claudia"
+  - family-names: "Knechtges"
+    given-names: "Philipp"
+  - family-names: "Hagemeier"
+    given-names: "Björn"
+  - family-names: "Tarnawa"
+    given-names: "Michael"
+  - family-names: "Hanselmann"
+    given-names: "Simon"
+  - family-names: "Siggel"
+    given-names: "Martin"
+  - family-names: "Basermann"
+    given-names: "Achim"
+  - family-names: "Streit"
+    given-names: "Achim"
+  title: "HeAT -- a Distributed and GPU-accelerated Tensor Framework for Data Analytics"
+  year: 2020
+  collection-title: "2020 IEEE International Conference on Big Data (IEEE Big Data 2020)"
+  collection-doi: 10.1109/BigData50022.2020.9378050
+  conference:
+      name: 2020 IEEE International Conference on Big Data (IEEE Big Data 2020)
+      date-start: 2020-12-10
+      date-end: 2020-12-13
+  start: 276
+  end: 287

heat/core/dndarray.py

@@ -872,7 +872,7 @@ def __getitem__(self, key: Union[int, Tuple[int, ...], List[int, ...]]) -> DNDar
             output_split = None
 
         # data are not distributed or split dimension is not affected by indexing
-        if not self.is_distributed or key[self.split] == slice(None):
+        if not self.is_distributed() or key[self.split] == slice(None):
             return DNDarray(
                 self.larray[key],
                 gshape=output_shape,
@@ -1654,7 +1654,7 @@ def __set(arr: DNDarray, value: DNDarray):
             raise Exception("Advanced indexing is not supported yet")
 
         split = self.split
-        if not self.is_distributed or key[split] == slice(None):
+        if not self.is_distributed() or key[split] == slice(None):
             return __set(self[key], value)
 
         if isinstance(key[split], slice):

heat/core/indexing.py

@@ -13,13 +13,13 @@
 __all__ = ["nonzero", "where"]
 
 
-def nonzero(x: DNDarray) -> DNDarray:
+def nonzero(x: DNDarray) -> Tuple:
     """
-    Return a :class:`~heat.core.dndarray.DNDarray` containing the indices of the elements that are non-zero.. (using ``torch.nonzero``)
+    Return a Tuple of :class:`~heat.core.dndarray.DNDarray`s, one for each dimension of a,
+    containing the indices of the non-zero elements in that dimension. (using ``torch.nonzero``)
     If ``x`` is split then the result is split in the 0th dimension. However, this :class:`~heat.core.dndarray.DNDarray`
     can be UNBALANCED as it contains the indices of the non-zero elements on each node.
-    Returns an array with one entry for each dimension of ``x``, containing the indices of the non-zero elements in that dimension.
-    The values in ``x`` are always tested and returned in row-major, C-style order.
+    The values in ``x`` are always tested and returned in column-major, F-style order.
     The corresponding non-zero values can be obtained with: ``x[nonzero(x)]``.
 
     Parameters
@@ -32,10 +32,8 @@ def nonzero(x: DNDarray) -> DNDarray:
     >>> import heat as ht
     >>> x = ht.array([[3, 0, 0], [0, 4, 1], [0, 6, 0]], split=0)
     >>> ht.nonzero(x)
-    DNDarray([[0, 0],
-              [1, 1],
-              [1, 2],
-              [2, 1]], dtype=ht.int64, device=cpu:0, split=0)
+    (DNDarray([0, 1, 1, 2], dtype=ht.int64, device=cpu:0, split=None),
+        DNDarray([0, 1, 2, 1], dtype=ht.int64, device=cpu:0, split=None))
     >>> y = ht.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]], split=0)
     >>> y > 3
     DNDarray([[False, False, False],
@@ -48,6 +46,8 @@ def nonzero(x: DNDarray) -> DNDarray:
               [2, 0],
               [2, 1],
               [2, 2]], dtype=ht.int64, device=cpu:0, split=0)
+    (DNDarray([1, 1, 1, 2, 2, 2], dtype=ht.int64, device=cpu:0, split=None),
+        DNDarray([0, 1, 2, 0, 1, 2], dtype=ht.int64, device=cpu:0, split=None))
     >>> y[ht.nonzero(y > 3)]
     DNDarray([4, 5, 6, 7, 8, 9], dtype=ht.int64, device=cpu:0, split=0)
     """
@@ -56,14 +56,14 @@ def nonzero(x: DNDarray) -> DNDarray:
     except AttributeError:
         raise TypeError("Input must be a DNDarray, is {}".format(type(x)))
 
+    lcl_nonzero = torch.nonzero(input=local_x, as_tuple=False)
+
     if x.split is None:
         # if there is no split then just return the values from torch
-        lcl_nonzero = torch.nonzero(input=local_x, as_tuple=False)
         gout = list(lcl_nonzero.size())
         is_split = None
     else:
         # a is split
-        lcl_nonzero = torch.nonzero(input=local_x, as_tuple=False)
         # adjust local indices along split dimension
         _, displs = x.counts_displs()
         lcl_nonzero[..., x.split] += displs[x.comm.rank]
@@ -73,20 +73,18 @@ def nonzero(x: DNDarray) -> DNDarray:
         gout[0] = x.comm.allreduce(gout[0], MPI.SUM)
         is_split = 0
 
-    if x.ndim == 1:
-        lcl_nonzero = lcl_nonzero.squeeze(dim=1)
-    for g in range(len(gout) - 1, -1, -1):
-        if gout[g] == 1:
-            del gout[g]
-
-    return DNDarray(
-        lcl_nonzero,
-        gshape=tuple(gout),
-        dtype=types.canonical_heat_type(lcl_nonzero.dtype),
-        split=is_split,
-        device=x.device,
-        comm=x.comm,
-        balanced=False,
+    lcl_nonzero = lcl_nonzero.transpose(0, 1)
+
+    return tuple(
+        DNDarray(
+            lcl_nonzero,
+            gshape=tuple(gout),
+            dtype=types.canonical_heat_type(lcl_nonzero.dtype),
+            split=is_split,
+            device=x.device,
+            comm=x.comm,
+            balanced=False,
+        )
     )
 
 

heat/core/tests/test_indexing.py

@@ -9,18 +9,18 @@ def test_nonzero(self):
         a = ht.array([[1, 2, 3], [4, 5, 2], [7, 8, 9]], split=None)
         cond = a > 3
         nz = ht.nonzero(cond)
-        self.assertEqual(nz.gshape, (5, 2))
-        self.assertEqual(nz.dtype, ht.int64)
-        self.assertEqual(nz.split, None)
+        self.assertEqual(len(nz), 2)
+        self.assertEqual(len(nz[0]), 5)
+        self.assertEqual(nz[0].dtype, ht.int64)
 
         # split
         a = ht.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]], split=1)
         cond = a > 3
         nz = cond.nonzero()
-        self.assertEqual(nz.gshape, (6, 2))
-        self.assertEqual(nz.dtype, ht.int64)
-        self.assertEqual(nz.split, 0)
-        a[nz] = 10.0
+        self.assertEqual(len(nz), 2)
+        self.assertEqual(len(nz[0]), 6)
+        self.assertEqual(nz[0].dtype, ht.int64)
+        a[nz] = 10
         self.assertEqual(ht.all(a[nz] == 10), 1)
 
     def test_where(self):

scripts/tutorial.ipynb

@@ -1044,38 +1044,6 @@
     "a + b"
    ]
   },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "The example below will show that it is also possible to use operations on tensors with different split and the proper result calculated. However, this should be used seldomly and with small data amounts only, as it entails sending large amounts of data over the network."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": 34,
-   "metadata": {},
-   "outputs": [
-    {
-     "data": {
-      "text/plain": [
-       "(0/2) tensor([[9., 9., 9., 9., 9., 9.],\n",
-       "(0/2)         [9., 9., 9., 9., 9., 9.]])\n",
-       "(1/2) tensor([[9., 9., 9., 9., 9., 9.],\n",
-       "(1/2)         [9., 9., 9., 9., 9., 9.]])"
-      ]
-     },
-     "execution_count": 34,
-     "metadata": {},
-     "output_type": "execute_result"
-    }
-   ],
-   "source": [
-    "a = ht.full((4, 6,), 8, split=0)\n",
-    "b = ht.ones((4, 6,), split=1)\n",
-    "a + b"
-   ]
-  },
   {
    "cell_type": "markdown",
    "metadata": {},