Source code for pennylane.devices.execution_config

# Copyright 2018-2025 Xanadu Quantum Technologies Inc.

# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at

#     http://www.apache.org/licenses/LICENSE-2.0

# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Contains the :class:`ExecutionConfig` and :class:`MCMConfig` data classes.
"""
from __future__ import annotations

from collections.abc import MutableMapping
from copy import deepcopy
from dataclasses import dataclass, field
from typing import TYPE_CHECKING, Literal

from pennylane.concurrency.executors.backends import ExecBackends, get_executor
from pennylane.math.interface_utils import Interface, get_canonical_interface_name
from pennylane.transforms.core import TransformDispatcher

if TYPE_CHECKING:
    from pennylane.concurrency.executors.base import RemoteExec


class FrozenMapping(MutableMapping):
    """
    Custom immutable mapping.
    Inherit from MutableMapping to ensure all mutable methods are implemented.
    """

    def __init__(self, *args, **kwargs):
        self._data = dict(*args, **kwargs)
        self._hash = None  # Cache the hash value

    def __getitem__(self, key):
        return self._data[key]

    def __iter__(self):
        return iter(self._data)

    def __len__(self):
        return len(self._data)

    def __setitem__(self, key, value):
        raise TypeError(
            "FrozenMapping is immutable. To update this field please use `dataclasses.replace`. "
        )

    def __delitem__(self, key):
        raise TypeError(
            "FrozenMapping is immutable. To update this field please use `dataclasses.replace`. "
        )

    def __repr__(self):
        return f"{self._data}"

    def __hash__(self):
        """Makes the object hashable, allowing it to be used in sets and as a dict key."""
        if self._hash is None:
            self._hash = hash(frozenset(self._data.items()))
        return self._hash

    def copy(self):
        """Returns a standard, mutable shallow copy of the data."""
        return self._data.copy()

    def __copy__(self):
        """Supports copy.copy() by returning a mutable dict."""
        return self.copy()

    def __deepcopy__(self, memo=None):
        """Supports copy.deepcopy() by returning a mutable dict with deep-copied contents."""
        return deepcopy(self._data, memo)


[docs] @dataclass(frozen=True) class MCMConfig: """A class to store mid-circuit measurement configurations.""" mcm_method: ( Literal["deferred", "one-shot", "tree-traversal", "single-branch-statistics"] | str | None ) = None """The mid-circuit measurement strategy to use. Use ``"deferred"`` for the deferred measurements principle and ``"one-shot"`` if using finite shots to execute the circuit for each shot separately. Any other value will be passed to the device, and the device is expected to handle mid-circuit measurements using the requested method. If not specified, the device will decide which method to use.""" postselect_mode: Literal["hw-like", "fill-shots", "pad-invalid-samples"] | str | None = None """How postselection is handled with finite-shots. If ``"hw-like"``, invalid shots will be discarded and only results for valid shots will be returned. In this case, fewer samples may be returned than the original number of shots. If ``"fill-shots"``, the returned samples will be of the same size as the original number of shots. If not specified, the device will decide which mode to use. Note that internally ``"pad-invalid-samples"`` is used internally instead of ``"hw-like"`` when using jax/catalyst.""" def __post_init__(self): """Validate the configured mid-circuit measurement options.""" if self.postselect_mode not in ("hw-like", "fill-shots", "pad-invalid-samples", None): raise ValueError(f"Invalid postselection mode '{self.postselect_mode}'.")
# pylint: disable=too-many-instance-attributes
[docs] @dataclass(frozen=True) class ExecutionConfig: """ A class to configure the execution of a quantum circuit on a device. See the Attributes section to learn more about the various configurable options. """ grad_on_execution: bool | None = None """Whether or not to compute the gradient at the same time as the execution. If ``None``, then the device or execution pipeline can decide which one is most efficient for the situation. """ use_device_gradient: bool | None = None """Whether or not to compute the gradient on the device. ``None`` indicates to use the device if possible, but to fall back to pennylane behaviour if it isn't. True indicates a request to either use the device gradient or fail. """ use_device_jacobian_product: bool | None = None """Whether or not to use the device provided vjp or jvp to compute gradients. ``None`` indicates to use the device if possible, but to fall back to the device Jacobian or PennyLane behaviour if it isn't. ``True`` indicates to either use the device Jacobian products or fail. """ gradient_method: str | TransformDispatcher | None = None """The method used to compute the gradient of the quantum circuit being executed""" gradient_keyword_arguments: dict = field(default_factory=FrozenMapping) """Arguments used to control a gradient transform""" device_options: dict = field(default_factory=FrozenMapping) """Various options for the device executing a quantum circuit""" interface: str | Interface | None = Interface.NUMPY """The machine learning framework to use""" derivative_order: int = 1 """The derivative order to compute while evaluating a gradient""" mcm_config: MCMConfig | dict = field(default_factory=MCMConfig) """Configuration options for handling mid-circuit measurements""" convert_to_numpy: bool = True """Whether or not to convert parameters to numpy before execution. If ``False`` and using the jax-jit, no pure callback will occur and the device execution itself will be jitted. """ executor_backend: RemoteExec | None = None """ Defines the class for the executor backend. """ def __post_init__(self): """ Validate the configured execution options. Note that this hook is automatically called after init via the dataclass integration. """ object.__setattr__(self, "interface", get_canonical_interface_name(self.interface)) if self.grad_on_execution not in {True, False, None}: raise ValueError( f"grad_on_execution must be True, False, or None. Got {self.grad_on_execution} instead." ) def _validate_and_freeze_dict(field_name: str): value = getattr(self, field_name) if not isinstance(value, (dict, FrozenMapping)): raise TypeError(f"Got invalid type {type(value)} for '{field_name}'") # This handles the case when `dataclasses.replace` is used and # the field is not being modified. if isinstance(value, dict): object.__setattr__(self, field_name, FrozenMapping(value)) _validate_and_freeze_dict("device_options") _validate_and_freeze_dict("gradient_keyword_arguments") if not ( isinstance(self.gradient_method, (str, TransformDispatcher)) or self.gradient_method is None ): raise ValueError( f"Differentiation method {self.gradient_method} must be a str, TransformDispatcher, or None. Got {type(self.gradient_method)} instead." ) if isinstance(self.mcm_config, dict): object.__setattr__(self, "mcm_config", MCMConfig(**self.mcm_config)) elif not isinstance(self.mcm_config, MCMConfig): raise ValueError(f"Got invalid type {type(self.mcm_config)} for 'mcm_config'") if self.executor_backend is None: object.__setattr__(self, "executor_backend", get_executor(backend=ExecBackends.MP_Pool))
# pylint: disable=missing-function-docstring, inconsistent-return-statements def __getattr__(name): if name == "DefaultExecutionConfig": # pylint: disable=import-outside-toplevel import warnings from pennylane.exceptions import PennyLaneDeprecationWarning warnings.warn( "`pennylane.devices.execution_config.DefaultExecutionConfig` is deprecated and will be removed in v0.44. " "Please use `ExecutionConfig()` instead.", PennyLaneDeprecationWarning, stacklevel=2, ) return ExecutionConfig() raise AttributeError(f"module 'pennylane.devices.execution_config' has no attribute '{name}'")