Fix #963: PETSc 3.20.0 broke the world (#982)

Fixes #963
(Actually, PETSC 3.19.0 broke the world.)

This PR:
- Assembles before `multTranspose` to prevent newly added exception
- Renames `bandwidth` to `nonZerosPerRow` and removes `sizeHint`

  The two were confusingly redundant:
  - PySparse takes `sizeHint`, the number of non-zeros in the matrix.
  - PyTrilinos takes `NumEntriesPerRow`.
  - petsc4py didn't used to be clear what it took, but is now
    documented as number of non-zeros per row (of the local portion
    of the matrix, but we'll ignore that part).
  - scipy doesn't preallocate.
  - Linear algebra
    [defines bandwidth](https://en.wikipedia.org/wiki/Band_matrix#Bandwidth)
    as "number $k$ such that $a_{i,j}=0$ if $|i-j| > k$", which is
    roughly half the number of non-zeros per row (and only applies
    to a band-diagonal matrix).
    Better to be explicit about what we really mean.

  Now all take same parameter and PySparse adjusts as needed.

  `sizeHint` was introduced in @a15d696 (in 2006!) to
  "allow smaller preallocations", but it was never used that way.
  Now, `nonZerosPerRow` can take an array_like to specify row-by-row
  preallocations, which are directly supported by PyTrilinos and petsc4py,
  and can be simulated for PySparse.

  Added `exactNonZeros`, which may have performance benefits for
  PyTrilinos and petsc4py. Currently unused.
- Uses `Term`'s knowledge of own stencil to preallocate more effectively.
  Still doesn't do a good job with vector equations, but that's a deeper change
  (the resolution of which might help #920).
- Fixes(?) conda/mamba installs
- Fixes(?) race condition

.azure/pipelines.yml

@@ -258,7 +258,7 @@ stages:
         condition: startsWith(variables.image, 'windows')
 
       - bash: |
-          mamba create --quiet --name wheelEnvironment --channel conda-forge python=3.10 mamba
+          conda create --quiet --name wheelEnvironment --channel conda-forge python=3.10
           source activate wheelEnvironment
           mkdir tmp
           cd tmp

.azure/templates/install.yml

@@ -31,24 +31,11 @@ steps:
   - bash: |
       conda config --set always_yes yes --set changeps1 no
       conda config --remove channels defaults
-      if [[ ${{ parameters.python_version }} == "2.7" ]]; then
-        # mamba doesn't work in Py2.7
-        conda create --quiet --name myEnvironment --channel conda-forge python=${{ parameters.python_version }}
-      else
-        # mamba needs to be installed in the Py3k environment for some reason
-        conda create --yes --quiet --name myEnvironment --channel conda-forge python=${{ parameters.python_version }} mamba
-      fi
+      conda config --set solver libmamba
     displayName: Create Anaconda environment
 
   - bash: |
-      if [[ ${{ parameters.python_version }} == "2.7" ]]; then
-        # mamba doesn't work in Py2.7
-        conda install --quiet --name myEnvironment --channel conda-forge python=${{ parameters.python_version }} ${{ parameters.conda_packages }}
-      else
-        # do mamba installs from environment on Py3k
-        source activate myEnvironment
-        mamba install --quiet --channel conda-forge python=${{ parameters.python_version }} ${{ parameters.conda_packages }}
-      fi
+      conda create --yes --quiet --name myEnvironment --channel conda-forge python=${{ parameters.python_version }} ${{ parameters.conda_packages }}
     displayName: Install Anaconda packages
 
   - bash: |

.en-custom.txt

@@ -271,6 +271,9 @@ pre-assembled
 pre-built
 pre-installed
 Pre-Installed
+preallocated
+preallocation
+preallocating
 preconditioner
 preconditioners
 Pusztai

fipy/boundaryConditions/boundaryCondition.py

@@ -57,8 +57,8 @@ def _buildMatrix(self, SparseMatrix, Ncells, MaxFaces, coeff):
         Ncells : int
             Size of matrices
         MaxFaces : int
-            Maximum number of faces per cell (determines bandwidth of
-            :math:`\mathsf{L}`)
+            Maximum number of faces per cell (determines number of
+            non-zeros per row of :math:`\mathsf{L}`)
         coeff : list
             Contribution due to this face
         """

fipy/boundaryConditions/fixedValue.py

@@ -51,15 +51,15 @@ def _buildMatrix(self, SparseMatrix, Ncells, MaxFaces, coeff):
         Ncells : int
             Size of matrices
         MaxFaces : int
-            Maximum number of faces per cell (determines bandwidth of
-            :math:`\mathsf{L}`)
+            Maximum number of faces per cell (determines number of
+            non-zeros per row of :math:`\mathsf{L}`)
         coeff : list
             Contribution to adjacent cell diagonal and
             :math:`\mathsf{b}` vector by this exterior face
         """
         faces = self.faces.value
 
-        LL = SparseMatrix(mesh=self.faces.mesh, sizeHint=len(self.faces), bandwidth=1)
+        LL = SparseMatrix(mesh=self.faces.mesh, nonZerosPerRow=1)
         LL.addAt(coeff['cell 1 diag'][faces], self.adjacentCellIDs, self.adjacentCellIDs)
 
         ## The following has been commented out because

fipy/matrices/offsetSparseMatrix.py

@@ -14,16 +14,41 @@ class OffsetSparseMatrixClass(SparseMatrix):
         equationIndex = 0
         varIndex = 0
 
-        def __init__(self, mesh, bandwidth=0, sizeHint=None,
-                     numberOfVariables=numberOfVariables, numberOfEquations=numberOfEquations):
-            SparseMatrix.__init__(self, mesh=mesh, bandwidth=bandwidth, sizeHint=sizeHint,
-                                  numberOfVariables=numberOfVariables, numberOfEquations=numberOfEquations)
+        def __init__(self, mesh, nonZerosPerRow=1, exactNonZeros=False,
+                     numberOfVariables=numberOfVariables,
+                     numberOfEquations=numberOfEquations):
+            if hasattr(nonZerosPerRow, "__iter__"):
+                # nonZerosPerRow is an iterable for each row.
+                # need to pad rows for other equations with zeros.
+                # can't compare to collections.abc.Iterable because PySparse.
+                before = self.equationIndex
+                after = numberOfEquations - self.equationIndex - 1
+                N = len(nonZerosPerRow)
+                nonZerosPerRow = numerix.concatenate([[0] * N * before,
+                                                      nonZerosPerRow,
+                                                      [0] * N * after])
+            else:
+                nonZerosPerRow //= numberOfEquations
+            SparseMatrix.__init__(self,
+                                  mesh=mesh,
+                                  nonZerosPerRow=nonZerosPerRow,
+                                  exactNonZeros=exactNonZeros,
+                                  numberOfVariables=numberOfVariables,
+                                  numberOfEquations=numberOfEquations)
 
         def put(self, vector, id1, id2):
-            SparseMatrix.put(self, vector, id1 + self.mesh.numberOfCells * self.equationIndex, id2 + self.mesh.numberOfCells * self.varIndex)
+            N = self.mesh.numberOfCells
+            SparseMatrix.put(self,
+                             vector=vector,
+                             id1=id1 + N * self.equationIndex,
+                             id2=id2 + N * self.varIndex)
 
         def addAt(self, vector, id1, id2):
-            SparseMatrix.addAt(self, vector, id1 + self.mesh.numberOfCells * self.equationIndex, id2 + self.mesh.numberOfCells * self.varIndex)
+            N = self.mesh.numberOfCells
+            SparseMatrix.addAt(self,
+                               vector=vector,
+                               id1=id1 + N * self.equationIndex,
+                               id2=id2 + N * self.varIndex)
 
         def addAtDiagonal(self, vector):
             if type(vector) in [type(1), type(1.)]: