File size: 3,895 Bytes
6288873 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 | """
Helpers to apply contractions.
"""
from functools import partial
import jax
import jax.numpy as jnp
from varipeps.peps import PEPS_Tensor
from .definitions import Definitions, Definition
from typing import Sequence, List, Tuple, Dict, Union, Optional
@partial(jax.jit, static_argnames=("name", "disable_identity_check"))
def apply_contraction(
name: str,
peps_tensors: Sequence[jnp.ndarray],
peps_tensor_objs: Sequence[PEPS_Tensor],
additional_tensors: Sequence[jnp.ndarray],
*,
disable_identity_check: bool = True,
) -> jnp.ndarray:
"""
Apply a contraction to a list of tensors.
For details on the contractions and their definition see
:class:`varipeps.contractions.Definitions`.
Args:
name (:obj:`str`):
Name of the contraction. Must be a class attribute of the class
:class:`varipeps.contractions.Definitions`.
peps_tensors (:term:`sequence` of :obj:`jax.numpy.ndarray`):
The PEPS tensor arrays that should be contracted.
peps_tensor_objs (:term:`sequence` of :obj:`~varipeps.peps.PEPS_Tensor`):
The PEPS tensor objects corresponding the the arrays. These arguments are
split up due to limitation of the jax library.
additional_tensors (:term:`sequence` of :obj:`jax.numpy.ndarray`):
Additional non-PEPS tensors which should be contracted (e.g. gates).
Keyword args:
disable_identity_check (:obj:`bool`):
Disable the check if the tensor is identical to the one of the
corresponding object.
Returns:
jax.numpy.ndarray:
The contracted tensor.
"""
if len(peps_tensors) != len(peps_tensor_objs):
raise ValueError(
"Number of PEPS tensors have to match number of PEPS tensor objects."
)
if (
not disable_identity_check
and not all(isinstance(t, jax.core.Tracer) for t in peps_tensors)
and not all(isinstance(to.tensor, jax.core.Tracer) for to in peps_tensor_objs)
and not all(
jnp.allclose(peps_tensors[i], peps_tensor_objs[i].tensor)
for i in range(len(peps_tensors))
)
):
raise ValueError(
"Sequence of PEPS tensors mismatch the objects sequence. Please check your code!"
)
contraction = getattr(Definitions, name)
if len(contraction["filter_peps_tensors"]) != len(peps_tensors):
raise ValueError(
f"Number of PEPS tensor ({len(peps_tensors)}) objects does not fit the expected number ({len(contraction['filter_peps_tensors'])})."
)
if len(contraction["filter_additional_tensors"]) != len(additional_tensors):
raise ValueError(
f"Number of additional tensor ({len(additional_tensors)}) objects does not fit the expected number ({len(contraction['filter_additional_tensors'])})."
)
if not isinstance(additional_tensors, list):
additional_tensors = list(additional_tensors)
tensors = []
for ti, t_filter in enumerate(contraction["filter_peps_tensors"]):
for f in t_filter:
if f == "tensor":
tensors.append(peps_tensors[ti])
elif f == "tensor_conj":
if (
hasattr(peps_tensor_objs[ti], "tensor_conj")
and peps_tensor_objs[ti].tensor_conj is not None
):
tensors.append(peps_tensor_objs[ti].tensor_conj)
else:
tensors.append(peps_tensors[ti].conj())
else:
tensors.append(getattr(peps_tensor_objs[ti], f))
tensors += additional_tensors
tensor_shapes = tuple(tuple(e.shape) for e in tensors)
return jnp.einsum(
contraction["einsum_network"],
*tensors,
optimize="optimal" if len(tensors) < 10 else "dp",
)
apply_contraction_jitted = apply_contraction
|