Source code for pytblis.einsum_impl

# Contains code from opt_einsum, which is licensed under the MIT License.
# The MIT License (MIT)

# Copyright (c) 2014 Daniel Smith

# Permission is hereby granted, free of charge, to any person obtaining a copy
# of this software and associated documentation files (the "Software"), to deal
# in the Software without restriction, including without limitation the rights
# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
# copies of the Software, and to permit persons to whom the Software is
# furnished to do so, subject to the following conditions:

# The above copyright notice and this permission notice shall be included in all
# copies or substantial portions of the Software.

# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
# SOFTWARE.

from contextlib import nullcontext

import numpy as np

from .defaultorder import _default_order, get_default_array_order, use_default_array_order
from .numpy_einsumpath import einsum_path
from .wrappers import contract, transpose_add


[docs] def einsum(*operands, out=None, optimize=True, complex_real_contractions=True, **kwargs): """ einsum(subscripts, *operands, out=None, order='K', optimize='greedy') Evaluates the Einstein summation convention on the operands. Drop-in replacement for numpy.einsum, using TBLIS for tensor contractions. Parameters ---------- subscripts : str Specifies the subscripts for summation as comma separated list of subscript labels. An implicit (classical Einstein summation) calculation is performed unless the explicit indicator '->' is included as well as subscript labels of the precise output form. operands : list of array_like These are the arrays for the operation. out : ndarray, optional If provided, the calculation is done into this array. order : {'C', 'F', 'A', 'K'}, optional Controls the memory layout of the output. 'C' means it should be C contiguous. 'F' means it should be Fortran contiguous, 'A' means it should be 'F' if the inputs are all 'F', 'C' otherwise. 'K' is ignored, for now. Default is 'C'. optimize : {False, True, 'greedy', 'optimal'}, default True Controls the optimization strategy used to compute the contraction. complex_real_contractions : bool, default True If True, handle contractions between complex and real tensors by performing separate contractions for the real and imaginary parts of the complex tensor. This avoids NumPy type promotion if the complex and real tensors have the same precision (e.g., complex128 and float64). Returns ------- output : ndarray The calculation based on the Einstein summation convention. """ specified_out = out is not None if optimize not in (False, True, "greedy", "optimal"): raise ValueError("optimize must be one of False, True, 'greedy', or 'optimal'") # Check the kwargs to avoid a more cryptic error later, without having to # repeat default values here valid_einsum_kwargs = ["order"] unknown_kwargs = [k for (k, v) in kwargs.items() if k not in valid_einsum_kwargs] if unknown_kwargs: msg = f"Did not understand the following kwargs: {unknown_kwargs}" raise TypeError(msg) # calculate contraction path operands, contraction_list = einsum_path(*operands, optimize=optimize, einsum_call=True) # Handle order kwarg for output array, c_einsum allows mixed case order_given = "order" in kwargs output_order = kwargs.get("order", _default_order.get()) if output_order not in ("C", "F", "A", "K"): raise ValueError("order must be one of 'C', 'F', 'A', or 'K'") if output_order == "A": output_order = "F" if all(arr.flags.f_contiguous for arr in operands) else "C" elif output_order == "K": # ignore K. output_order = get_default_array_order() # Start contraction loop for num, contraction in enumerate(contraction_list): inds, einsum_str, _ = contraction tmp_operands = [operands.pop(x) for x in inds] # Do we need to deal with the output? handle_out = specified_out and ((num + 1) == len(contraction_list)) if handle_out: out_kwarg = out else: out_kwarg = None if ((num + 1) == len(contraction_list)) and order_given: # Set the requested output order on the final contraction. order_context = use_default_array_order(output_order) else: order_context = nullcontext() if len(tmp_operands) == 2: # two operands: use contract with order_context: new_view = contract( einsum_str, *tmp_operands, out=out_kwarg, allow_partial_trace=True, complex_real_contractions=complex_real_contractions, ) elif len(tmp_operands) == 1: # check if only a transpose subscript_a, subscript_b = einsum_str.split("->") if sorted(subscript_a) == sorted(subscript_b): # only a transpose, use numpy for this (should return view) new_view = np.einsum(einsum_str, tmp_operands[0], out=out_kwarg, **kwargs) # may involve a trace or replication, use tblis transpose_add for this else: with order_context: new_view = transpose_add(einsum_str, tmp_operands[0], out=out_kwarg) else: # 3 or more operands, fall back to numpy einsum new_view = np.einsum(einsum_str, *tmp_operands, out=out_kwarg, **kwargs) # Append new items and dereference what we can operands.append(new_view) del tmp_operands, new_view if specified_out: return out return operands[0]