Add linalg.block_diag and sparse equivalent (#576)

* Copy `block_diag` and support functions from `pymc.math`

* Evaluate output in sphinx code example

Co-authored-by: Ricardo Vieira <28983449+ricardoV94@users.noreply.github.com>

* Test type equivalence with `isinstance` instead of `==`

Co-authored-by: Ricardo Vieira <28983449+ricardoV94@users.noreply.github.com>

* Typo in test function

* Split `block_diag` into sparse and dense version

Closely follow scipy function signature for `block_diag`

* Use `as_sparse_or_tensor_variable` in `SparseBlockDiagonalMatrix` to allow sparse matrix inputs to `pytensor.sparse.block_diag`

* Test sparse and dense inputs to `pytensor.sparse.block_diag`

* Add numba overload for `pytensor.tensor.slinalg.block_diag`

* add jax overload for `pytensor.tensor.slinalg.block_diag`

* Move stand-alone `block_diag_grad` function into `grad` method

* Add `format` prop to `SparseBlockDiagonalMatrix`

* Use `compare_numba_and_py` in `numba\test_slinalg.py::test_block_diag`

* Add support for Blockwise to `slinalg.block_diag`

* Add gradient test

Remove `Matrix` from `BlockDiagonal` and `SparseBlockDiagonal` `Op` names

Correct errors in docstrings

Move input validation to a shared class method

* Remove `gufunc_signature` from `__props__`

Co-authored-by: Ricardo Vieira <28983449+ricardoV94@users.noreply.github.com>

* Implement correct `__props__` for subclasses of `BaseBlockMatrix`

---------

Co-authored-by: Ricardo Vieira <28983449+ricardoV94@users.noreply.github.com>

Author: jessegrabowski

pytensor/link/jax/dispatch/slinalg.py

@@ -1,7 +1,7 @@
 import jax
 
 from pytensor.link.jax.dispatch.basic import jax_funcify
-from pytensor.tensor.slinalg import Cholesky, Solve, SolveTriangular
+from pytensor.tensor.slinalg import BlockDiagonal, Cholesky, Solve, SolveTriangular
 
 
 @jax_funcify.register(Cholesky)
@@ -45,3 +45,11 @@ def solve_triangular(A, b):
         )
 
     return solve_triangular
+
+
+@jax_funcify.register(BlockDiagonal)
+def jax_funcify_BlockDiagonalMatrix(op, **kwargs):
+    def block_diag(*inputs):
+        return jax.scipy.linalg.block_diag(*inputs)
+
+    return block_diag

pytensor/link/numba/dispatch/slinalg.py

@@ -9,7 +9,7 @@
 
 from pytensor.link.numba.dispatch import basic as numba_basic
 from pytensor.link.numba.dispatch.basic import numba_funcify
-from pytensor.tensor.slinalg import SolveTriangular
+from pytensor.tensor.slinalg import BlockDiagonal, SolveTriangular
 
 
 _PTR = ctypes.POINTER
@@ -273,3 +273,25 @@ def solve_triangular(a, b):
         return res
 
     return solve_triangular
+
+
+@numba_funcify.register(BlockDiagonal)
+def numba_funcify_BlockDiagonal(op, node, **kwargs):
+    dtype = node.outputs[0].dtype
+
+    # TODO: Why do we always inline all functions? It doesn't work with starred args, so can't use it in this case.
+    @numba_basic.numba_njit(inline="never")
+    def block_diag(*arrs):
+        shapes = np.array([a.shape for a in arrs], dtype="int")
+        out_shape = [int(s) for s in np.sum(shapes, axis=0)]
+        out = np.zeros((out_shape[0], out_shape[1]), dtype=dtype)
+
+        r, c = 0, 0
+        for arr, shape in zip(arrs, shapes):
+            rr, cc = shape
+            out[r : r + rr, c : c + cc] = arr
+            r += rr
+            c += cc
+        return out
+
+    return block_diag

pytensor/sparse/basic.py

@@ -7,6 +7,7 @@
 TODO: Automatic methods for determining best sparse format?
 
 """
+from typing import Literal
 from warnings import warn
 
 import numpy as np
@@ -47,6 +48,7 @@
     trunc,
 )
 from pytensor.tensor.shape import shape, specify_broadcastable
+from pytensor.tensor.slinalg import BaseBlockDiagonal, _largest_common_dtype
 from pytensor.tensor.type import TensorType
 from pytensor.tensor.type import continuous_dtypes as tensor_continuous_dtypes
 from pytensor.tensor.type import discrete_dtypes as tensor_discrete_dtypes
@@ -60,7 +62,6 @@
 
 sparse_formats = ["csc", "csr"]
 
-
 """
 Types of sparse matrices to use for testing.
 
@@ -183,7 +184,6 @@ def as_sparse_variable(x, name=None, ndim=None, **kwargs):
 
 as_sparse = as_sparse_variable
 
-
 as_sparse_or_tensor_variable = as_symbolic
 
 
@@ -1800,7 +1800,7 @@ def infer_shape(self, fgraph, node, shapes):
         return r
 
     def __str__(self):
-        return f"{self.__class__.__name__ }{{axis={self.axis}}}"
+        return f"{self.__class__.__name__}{{axis={self.axis}}}"
 
 
 def sp_sum(x, axis=None, sparse_grad=False):
@@ -2775,19 +2775,14 @@ def comparison(self, x, y):
 
 greater_equal_s_d = GreaterEqualSD()
 
-
 eq = __ComparisonSwitch(equal_s_s, equal_s_d, equal_s_d)
 
-
 neq = __ComparisonSwitch(not_equal_s_s, not_equal_s_d, not_equal_s_d)
 
-
 lt = __ComparisonSwitch(less_than_s_s, less_than_s_d, greater_than_s_d)
 
-
 gt = __ComparisonSwitch(greater_than_s_s, greater_than_s_d, less_than_s_d)
 
-
 le = __ComparisonSwitch(less_equal_s_s, less_equal_s_d, greater_equal_s_d)
 
 ge = __ComparisonSwitch(greater_equal_s_s, greater_equal_s_d, less_equal_s_d)
@@ -2992,7 +2987,7 @@ def __str__(self):
         l = []
         if self.inplace:
             l.append("inplace")
-        return f"{self.__class__.__name__ }{{{', '.join(l)}}}"
+        return f"{self.__class__.__name__}{{{', '.join(l)}}}"
 
     def make_node(self, x):
         """
@@ -3291,6 +3286,7 @@ class TrueDot(Op):
     # Simplify code by splitting into DotSS and DotSD.
 
     __props__ = ()
+
     # The grad_preserves_dense attribute doesn't change the
     # execution behavior.  To let the optimizer merge nodes with
     # different values of this attribute we shouldn't compare it
@@ -4260,3 +4256,85 @@ def grad(self, inputs, grads):
 
 
 construct_sparse_from_list = ConstructSparseFromList()
+
+
+class SparseBlockDiagonal(BaseBlockDiagonal):
+    __props__ = (
+        "n_inputs",
+        "format",
+    )
+
+    def __init__(self, n_inputs: int, format: Literal["csc", "csr"] = "csc"):
+        super().__init__(n_inputs)
+        self.format = format
+
+    def make_node(self, *matrices):
+        matrices = self._validate_and_prepare_inputs(
+            matrices, as_sparse_or_tensor_variable
+        )
+        dtype = _largest_common_dtype(matrices)
+        out_type = matrix(format=self.format, dtype=dtype)
+
+        return Apply(self, matrices, [out_type])
+
+    def perform(self, node, inputs, output_storage, params=None):
+        dtype = node.outputs[0].type.dtype
+        output_storage[0][0] = scipy.sparse.block_diag(
+            inputs, format=self.format
+        ).astype(dtype)
+
+
+def block_diag(*matrices: TensorVariable, format: Literal["csc", "csr"] = "csc"):
+    r"""
+    Construct a block diagonal matrix from a sequence of input matrices.
+
+    Given the inputs `A`, `B` and `C`, the output will have these arrays arranged on the diagonal:
+
+    [[A, 0, 0],
+     [0, B, 0],
+     [0, 0, C]]
+
+    Parameters
+    ----------
+    A, B, C ... : tensors
+        Input tensors to form the block diagonal matrix. last two dimensions of the inputs will be used, and all
+        inputs should have at least 2 dimensins.
+
+        Note that the input matrices need not be sparse themselves, and will be automatically converted to the
+        requested format if they are not.
+
+    format: str, optional
+        The format of the output sparse matrix. One of 'csr' or 'csc'. Default is 'csr'. Ignored if sparse=False.
+
+    Returns
+    -------
+    out: sparse matrix tensor
+        Symbolic sparse matrix in the specified format.
+
+    Examples
+    --------
+    Create a sparse block diagonal matrix from two sparse 2x2 matrices:
+
+    ..code-block:: python
+        import numpy as np
+        from pytensor.sparse import block_diag
+        from scipy.sparse import csr_matrix
+
+        A = csr_matrix([[1, 2], [3, 4]])
+        B = csr_matrix([[5, 6], [7, 8]])
+        result_sparse = block_diag(A, B, format='csr', name='X')
+
+        print(result_sparse)
+        >>>  SparseVariable{csr,int32}
+
+        print(result_sparse.toarray().eval())
+        >>> array([[1, 2, 0, 0],
+        >>> [3, 4, 0, 0],
+        >>> [0, 0, 5, 6],
+        >>> [0, 0, 7, 8]])
+    """
+    if len(matrices) == 1:
+        return matrices
+
+    _sparse_block_diagonal = SparseBlockDiagonal(n_inputs=len(matrices), format=format)
+    return _sparse_block_diagonal(*matrices)

pytensor/tensor/basic.py

@@ -4279,6 +4279,25 @@ def take_along_axis(arr, indices, axis=0):
     return arr[_make_along_axis_idx(arr.shape, indices, axis)]
 
 
+def ix_(*args):
+    """
+    PyTensor np.ix_ analog
+
+    See numpy.lib.index_tricks.ix_ for reference
+    """
+    out = []
+    nd = len(args)
+    for k, new in enumerate(args):
+        if new is None:
+            out.append(slice(None))
+        new = as_tensor(new)
+        if new.ndim != 1:
+            raise ValueError("Cross index must be 1 dimensional")
+        new = new.reshape((1,) * k + (new.size,) + (1,) * (nd - k - 1))
+        out.append(new)
+    return tuple(out)
+
+
 __all__ = [
     "take_along_axis",
     "expand_dims",

pytensor/tensor/slinalg.py

@@ -1,6 +1,7 @@
 import logging
 import typing
 import warnings
+from functools import reduce
 from typing import TYPE_CHECKING, Literal, Optional, Union
 
 import numpy as np
@@ -23,7 +24,6 @@
 if TYPE_CHECKING:
     from pytensor.tensor import TensorLike
 
-
 logger = logging.getLogger(__name__)
 
 
@@ -908,6 +908,107 @@ def solve_discrete_are(A, B, Q, R, enforce_Q_symmetric=False) -> TensorVariable:
     )
 
 
+def _largest_common_dtype(tensors: typing.Sequence[TensorVariable]) -> np.dtype:
+    return reduce(lambda l, r: np.promote_types(l, r), [x.dtype for x in tensors])
+
+
+class BaseBlockDiagonal(Op):
+    __props__ = ("n_inputs",)
+
+    def __init__(self, n_inputs):
+        input_sig = ",".join([f"(m{i},n{i})" for i in range(n_inputs)])
+        self.gufunc_signature = f"{input_sig}->(m,n)"
+
+        if n_inputs == 0:
+            raise ValueError("n_inputs must be greater than 0")
+        self.n_inputs = n_inputs
+
+    def grad(self, inputs, gout):
+        shapes = pt.stack([i.shape for i in inputs])
+        index_end = shapes.cumsum(0)
+        index_begin = index_end - shapes
+        slices = [
+            ptb.ix_(
+                pt.arange(index_begin[i, 0], index_end[i, 0]),
+                pt.arange(index_begin[i, 1], index_end[i, 1]),
+            )
+            for i in range(len(inputs))
+        ]
+        return [gout[0][slc] for slc in slices]
+
+    def infer_shape(self, fgraph, nodes, shapes):
+        first, second = zip(*shapes)
+        return [(pt.add(*first), pt.add(*second))]
+
+    def _validate_and_prepare_inputs(self, matrices, as_tensor_func):
+        if len(matrices) != self.n_inputs:
+            raise ValueError(
+                f"Expected {self.n_inputs} matri{'ces' if self.n_inputs > 1 else 'x'}, got {len(matrices)}"
+            )
+        matrices = list(map(as_tensor_func, matrices))
+        if any(mat.type.ndim != 2 for mat in matrices):
+            raise TypeError("All inputs must have dimension 2")
+        return matrices
+
+
+class BlockDiagonal(BaseBlockDiagonal):
+    __props__ = ("n_inputs",)
+
+    def make_node(self, *matrices):
+        matrices = self._validate_and_prepare_inputs(matrices, pt.as_tensor)
+        dtype = _largest_common_dtype(matrices)
+        out_type = pytensor.tensor.matrix(dtype=dtype)
+        return Apply(self, matrices, [out_type])
+
+    def perform(self, node, inputs, output_storage, params=None):
+        dtype = node.outputs[0].type.dtype
+        output_storage[0][0] = scipy.linalg.block_diag(*inputs).astype(dtype)
+
+
+def block_diag(*matrices: TensorVariable):
+    """
+    Construct a block diagonal matrix from a sequence of input tensors.
+
+    Given the inputs `A`, `B` and `C`, the output will have these arrays arranged on the diagonal:
+
+    [[A, 0, 0],
+     [0, B, 0],
+     [0, 0, C]]
+
+    Parameters
+    ----------
+    A, B, C ... : tensors
+        Input tensors to form the block diagonal matrix. last two dimensions of the inputs will be used, and all
+        inputs should have at least 2 dimensins.
+
+    Returns
+    -------
+    out: tensor
+        The block diagonal matrix formed from the input matrices.
+
+    Examples
+    --------
+    Create a block diagonal matrix from two 2x2 matrices:
+
+    ..code-block:: python
+
+        import numpy as np
+        from pytensor.tensor.linalg import block_diag
+
+        A = pt.as_tensor_variable(np.array([[1, 2], [3, 4]]))
+        B = pt.as_tensor_variable(np.array([[5, 6], [7, 8]]))
+
+        result = block_diagonal(A, B, name='X')
+        print(result.eval())
+        >>> Out: array([[1, 2, 0, 0],
+        >>>             [3, 4, 0, 0],
+        >>>             [0, 0, 5, 6],
+        >>>             [0, 0, 7, 8]])
+    """
+    _block_diagonal_matrix = Blockwise(BlockDiagonal(n_inputs=len(matrices)))
+    return _block_diagonal_matrix(*matrices)
+
+
 __all__ = [
     "cholesky",
     "solve",
@@ -918,4 +1019,5 @@ def solve_discrete_are(A, B, Q, R, enforce_Q_symmetric=False) -> TensorVariable:
     "solve_continuous_lyapunov",
     "solve_discrete_are",
     "solve_triangular",
+    "block_diag",
 ]