Add SparseLatticedTensor

.gitignore

@@ -1,3 +1,6 @@
+# Jupyter notebooks
+*.ipynb
+
 # uv
 uv.lock
 

src/torchjd/autogram/_engine.py

@@ -4,6 +4,8 @@
 from torch import Tensor, nn, vmap
 from torch.autograd.graph import get_gradient_edge
 
+from torchjd.sparse import make_slt
+
 from ._edge_registry import EdgeRegistry
 from ._gramian_accumulator import GramianAccumulator
 from ._gramian_computer import GramianComputer, JacobianBasedGramianComputerWithCrossTerms
@@ -173,7 +175,9 @@ def differentiation(_grad_output: Tensor) -> tuple[Tensor, ...]:
                 )
 
             output_dims = list(range(output.ndim))
-            jac_output = _make_initial_jac_output(output)
+            identity = torch.eye(output.ndim, dtype=torch.int64)
+            basis = torch.concatenate([identity, identity], dim=0)
+            jac_output = make_slt(torch.ones_like(output), basis)
 
             vmapped_diff = differentiation
             for _ in output_dims:
@@ -193,15 +197,3 @@ def differentiation(_grad_output: Tensor) -> tuple[Tensor, ...]:
                 gramian_computer.reset()
 
         return gramian
-
-
-def _make_initial_jac_output(output: Tensor) -> Tensor:
-    if output.ndim == 0:
-        return torch.ones_like(output)
-    p_index_ranges = [torch.arange(s, device=output.device) for s in output.shape]
-    p_indices_grid = torch.meshgrid(*p_index_ranges, indexing="ij")
-    v_indices_grid = p_indices_grid + p_indices_grid
-
-    res = torch.zeros(list(output.shape) * 2, device=output.device, dtype=output.dtype)
-    res[v_indices_grid] = 1.0
-    return res

src/torchjd/sparse/__init__.py

@@ -0,0 +1,3 @@
+# Need to import this to execute the code inside and thus to override the functions
+from . import _aten_function_overrides
+from ._sparse_latticed_tensor import SparseLatticedTensor, make_slt

src/torchjd/sparse/_aten_function_overrides/__init__.py

@@ -0,0 +1 @@
+from . import backward, einsum, pointwise, shape

src/torchjd/sparse/_aten_function_overrides/backward.py

@@ -0,0 +1,36 @@
+from torch import Tensor
+from torch.ops import aten  # type: ignore
+
+from torchjd.sparse._sparse_latticed_tensor import SparseLatticedTensor, impl
+
+
+@impl(aten.threshold_backward.default)
+def threshold_backward_default(
+    grad_output: SparseLatticedTensor, self: Tensor, threshold
+) -> SparseLatticedTensor:
+    new_physical = aten.threshold_backward.default(grad_output.physical, self, threshold)
+
+    return SparseLatticedTensor(new_physical, grad_output.basis)
+
+
+@impl(aten.hardtanh_backward.default)
+def hardtanh_backward_default(
+    grad_output: SparseLatticedTensor,
+    self: Tensor,
+    min_val: Tensor | int | float,
+    max_val: Tensor | int | float,
+) -> SparseLatticedTensor:
+    if isinstance(self, SparseLatticedTensor):
+        raise NotImplementedError()
+
+    new_physical = aten.hardtanh_backward.default(grad_output.physical, self, min_val, max_val)
+    return SparseLatticedTensor(new_physical, grad_output.basis)
+
+
+@impl(aten.hardswish_backward.default)
+def hardswish_backward_default(grad_output: SparseLatticedTensor, self: Tensor):
+    if isinstance(self, SparseLatticedTensor):
+        raise NotImplementedError()
+
+    new_physical = aten.hardswish_backward.default(grad_output.physical, self)
+    return SparseLatticedTensor(new_physical, grad_output.basis)

src/torchjd/sparse/_aten_function_overrides/einsum.py

@@ -0,0 +1,253 @@
+import torch
+from torch import Tensor, tensor
+from torch.ops import aten  # type: ignore
+
+from torchjd.sparse._sparse_latticed_tensor import (
+    SparseLatticedTensor,
+    impl,
+    to_most_efficient_tensor,
+    to_sparse_latticed_tensor,
+)
+
+
+def einsum(*args: tuple[SparseLatticedTensor, list[int]], output: list[int]) -> Tensor:
+    raise NotImplementedError()
+
+    # First part of the algorithm, determine how to cluster physical indices as well as the common
+    # p_shapes corresponding to matching v_dims. Second part translates to physical einsum.
+
+    # get a map from einsum index to (tensor_idx, v_dims)
+    # get a map from einsum index to merge of strides corresponding to v_dims with that index
+    # use to_target_physical_strides on each physical and v_to_ps
+    # cluster pairs of (einsum_index, new_stride) using new_v_to_ps and possibly its corresponding
+    #  p_to_vs
+    # get unique indices
+    # map output indices (there can be splits)
+    # call physical einsum
+    # build resulting sst
+
+    # OVER
+
+    # an index in the physical einsum is uniquely characterized by a virtual einsum index and a
+    # stride corresponding to the physical stride in the virtual one (note that as the virtual shape
+    # for two virtual index that match should match, then we want to match the strides and reshape
+    # accordingly).
+    # We want to cluster such indices whenever several appear in the same p_to_vs
+
+    # TODO: Handle ellipsis
+    # If we have an index v for some virtual dim whose corresponding v_to_ps is a non-trivial list
+    # [p_1, ..., p_k], then we have to create fresh sub-indices for each dimension.
+    # For this reason, an index is decomposed into sub-indices that are then independently
+    # clustered.
+    # So if an index i in args for some SparseLatticedTensor corresponds to a v_to_ps [j, k, l],
+    # We will consider three indices (i, 0), (i, 1) and (i, 2).
+    # If furthermore [k] correspond to the v_to_ps of some other tensor with index j, then
+    # (i, 1) and (j, 0) will be clustered together (and end up being mapped to the same indice in
+    # the resulting einsum).
+    # Note that this is a problem if two virtual dimensions (from possibly different
+    # SparseLatticedTensors) have the same size but not the same decomposition into physical
+    # dimension sizes. For now lets leave the responsibility to care about that in the calling
+    # functions, if we can factor code later on we will.
+
+    index_parents = dict[tuple[int, int], tuple[int, int]]()
+
+    def get_representative(index: tuple[int, int]) -> tuple[int, int]:
+        if index not in index_parents:
+            # If an index is not yet in a cluster, put it in its own.
+            index_parents[index] = index
+        current = index_parents[index]
+        if current != index:
+            # Compress path to representative
+            index_parents[index] = get_representative(current)
+        return index_parents[index]
+
+    def group_indices(indices: list[tuple[int, int]]) -> None:
+        first_representative = get_representative(indices[0])
+        for i in indices[1:]:
+            curr_representative = get_representative(i)
+            index_parents[curr_representative] = first_representative
+
+    new_indices_pair = list[list[tuple[int, int]]]()
+    physicals = list[Tensor]()
+    indices_to_n_pdims = dict[int, int]()
+    for t, indices in args:
+        assert isinstance(t, SparseLatticedTensor)
+        physicals.append(t.physical)
+        for pdims, index in zip(t.v_to_ps, indices):
+            if index in indices_to_n_pdims:
+                if indices_to_n_pdims[index] != len(pdims):
+                    raise NotImplementedError(
+                        "einsum currently does not support having a different number of physical "
+                        "dimensions corresponding to matching virtual dimensions of different "
+                        f"tensors. Found {[(t.debug_info(), indices) for t, indices in args]}, "
+                        f"output_indices={output}."
+                    )
+            else:
+                indices_to_n_pdims[index] = len(pdims)
+        p_to_vs = ...  # p_to_vs_from_v_to_ps(t.v_to_ps)
+        for indices_ in p_to_vs:
+            # elements in indices[indices_] map to the same dimension, they should be clustered
+            # together
+            group_indices([(indices[i], sub_i) for i, sub_i in indices_])
+        # record the physical dimensions, index[v] for v in vs will end-up mapping to the same
+        # final dimension as they were just clustered, so we can take the first, which exists as
+        # t is a valid SST.
+        new_indices_pair.append([(indices[vs[0][0]], vs[0][1]) for vs in p_to_vs])
+
+    current = 0
+    pair_to_int = dict[tuple[int, int], int]()
+
+    def unique_int(pair: tuple[int, int]) -> int:
+        nonlocal current
+        if pair in pair_to_int:
+            return pair_to_int[pair]
+        pair_to_int[pair] = current
+        current += 1
+        return pair_to_int[pair]
+
+    new_indices = [
+        [unique_int(get_representative(i)) for i in indices] for indices in new_indices_pair
+    ]
+    new_output = list[int]()
+    v_to_ps = list[list[int]]()
+    for i in output:
+        current_v_to_ps = []
+        for j in range(indices_to_n_pdims[i]):
+            k = unique_int(get_representative((i, j)))
+            if k in new_output:
+                current_v_to_ps.append(new_output.index(k))
+            else:
+                current_v_to_ps.append(len(new_output))
+                new_output.append(k)
+        v_to_ps.append(current_v_to_ps)
+
+    physical = torch.einsum(*[x for y in zip(physicals, new_indices) for x in y], new_output)
+    # Need to use the safe constructor, otherwise the dimensions may not be maximally grouped.
+    # Maybe there is a way to fix that though.
+    return to_most_efficient_tensor(physical, v_to_ps)
+
+
+def prepare_for_elementwise_op(
+    t1: Tensor | int | float, t2: Tensor | int | float
+) -> tuple[SparseLatticedTensor, SparseLatticedTensor]:
+    """
+    Prepares two SLTs of the same shape from two args, one of those being a SLT, and the other being
+    a SLT, Tensor, int or float.
+    """
+
+    assert isinstance(t1, SparseLatticedTensor) or isinstance(t2, SparseLatticedTensor)
+
+    if isinstance(t1, int) or isinstance(t1, float):
+        t1_ = tensor(t1, device=t2.device)
+    else:
+        t1_ = t1
+
+    if isinstance(t2, int) or isinstance(t2, float):
+        t2_ = tensor(t2, device=t1.device)
+    else:
+        t2_ = t2
+
+    t1_, t2_ = aten.broadcast_tensors.default([t1_, t2_])
+    t1_ = to_sparse_latticed_tensor(t1_)
+    t2_ = to_sparse_latticed_tensor(t2_)
+
+    return t1_, t2_
+
+
+@impl(aten.mul.Tensor)
+def mul_Tensor(t1: Tensor | int | float, t2: Tensor | int | float) -> Tensor:
+    # Element-wise multiplication with broadcasting
+    t1_, t2_ = prepare_for_elementwise_op(t1, t2)
+    all_dims = list(range(t1_.ndim))
+    return einsum((t1_, all_dims), (t2_, all_dims), output=all_dims)
+
+
+@impl(aten.div.Tensor)
+def div_Tensor(t1: Tensor | int | float, t2: Tensor | int | float) -> Tensor:
+    t1_, t2_ = prepare_for_elementwise_op(t1, t2)
+    t2_ = SparseLatticedTensor(1.0 / t2_.physical, t2_.basis)
+    all_dims = list(range(t1_.ndim))
+    return einsum((t1_, all_dims), (t2_, all_dims), output=all_dims)
+
+
+@impl(aten.mul.Scalar)
+def mul_Scalar(t: SparseLatticedTensor, scalar) -> SparseLatticedTensor:
+    # TODO: maybe it could be that scalar is a scalar SLT and t is a normal tensor. Need to check
+    #  that
+
+    assert isinstance(t, SparseLatticedTensor)
+    new_physical = aten.mul.Scalar(t.physical, scalar)
+    return SparseLatticedTensor(new_physical, t.basis)
+
+
+@impl(aten.add.Tensor)
+def add_Tensor(
+    t1: Tensor | int | float, t2: Tensor | int | float, alpha: Tensor | float = 1.0
+) -> SparseLatticedTensor:
+    t1_, t2_ = prepare_for_elementwise_op(t1, t2)
+
+    if torch.equal(t1_.basis, t2_.basis):
+        new_physical = t1_.physical + t2_.physical * alpha
+        return SparseLatticedTensor(new_physical, t1_.basis)
+    else:
+        raise NotImplementedError()
+
+
+@impl(aten.bmm.default)
+def bmm_default(mat1: Tensor, mat2: Tensor) -> Tensor:
+    assert isinstance(mat1, SparseLatticedTensor) or isinstance(mat2, SparseLatticedTensor)
+    assert (
+        mat1.ndim == 3
+        and mat2.ndim == 3
+        and mat1.shape[0] == mat2.shape[0]
+        and mat1.shape[2] == mat2.shape[1]
+    )
+
+    mat1_ = to_sparse_latticed_tensor(mat1)
+    mat2_ = to_sparse_latticed_tensor(mat2)
+
+    # TODO: Verify that the dimension `0` of mat1_ and mat2_ have the same physical dimension sizes
+    #  decompositions. If not, can reshape to common decomposition?
+    return einsum((mat1_, [0, 1, 2]), (mat2_, [0, 2, 3]), output=[0, 1, 3])
+
+
+@impl(aten.mm.default)
+def mm_default(mat1: Tensor, mat2: Tensor) -> Tensor:
+    assert isinstance(mat1, SparseLatticedTensor) or isinstance(mat2, SparseLatticedTensor)
+    assert mat1.ndim == 2 and mat2.ndim == 2 and mat1.shape[1] == mat2.shape[0]
+
+    mat1_ = to_sparse_latticed_tensor(mat1)
+    mat2_ = to_sparse_latticed_tensor(mat2)
+
+    return einsum((mat1_, [0, 1]), (mat2_, [1, 2]), output=[0, 2])
+
+
+@impl(aten.mean.default)
+def mean_default(t: SparseLatticedTensor) -> Tensor:
+    assert isinstance(t, SparseLatticedTensor)
+    return aten.sum.default(t.physical) / t.numel()
+
+
+@impl(aten.sum.default)
+def sum_default(t: SparseLatticedTensor) -> Tensor:
+    assert isinstance(t, SparseLatticedTensor)
+    return aten.sum.default(t.physical)
+
+
+@impl(aten.sum.dim_IntList)
+def sum_dim_IntList(
+    t: SparseLatticedTensor, dim: list[int], keepdim: bool = False, dtype=None
+) -> Tensor:
+    assert isinstance(t, SparseLatticedTensor)
+
+    if dtype:
+        raise NotImplementedError()
+
+    all_dims = list(range(t.ndim))
+    result = einsum((t, all_dims), output=[d for d in all_dims if d not in dim])
+
+    if keepdim:
+        for d in dim:
+            result = result.unsqueeze(d)
+
+    return result