Source code for pennylane.transforms.tape_expand

# Copyright 2018-2021 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.
"""This module contains tape expansion functions and stopping criteria to
generate such functions from."""
# pylint: disable=unused-argument,invalid-unary-operand-type, unsupported-binary-operation, no-member
import contextlib
import warnings

import pennylane as qml
from pennylane.operation import (
    gen_is_multi_term_hamiltonian,
    has_gen,
    has_grad_method,
    has_nopar,
    has_unitary_gen,
    is_measurement,
    is_trainable,
    not_tape,
)


def _update_trainable_params(tape):
    params = tape.get_parameters(trainable_only=False)
    tape.trainable_params = qml.math.get_trainable_indices(params)


def create_expand_fn(depth, stop_at=None, device=None, docstring=None):
    """
    .. warning::
        Please use the :func:`qml.transforms.decompose <.transforms.decompose>` function for decomposing circuits.

    Create a function for expanding a tape to a given depth, and
    with a specific stopping criterion. This is a wrapper around
    :meth:`~.QuantumTape.expand`.

    Args:
        depth (int): Depth for the expansion
        stop_at (callable): Stopping criterion. This must be a function with signature
            ``stop_at(obj)``, where ``obj`` is a *queueable* PennyLane object such as
            :class:`~.Operation` or :class:`~.MeasurementProcess`. It must return a
            boolean, indicating if the expansion should stop at this object.
        device (pennylane.devices.LegacyDevice): Ensure that the expanded tape only uses native gates of the
            given device.
        docstring (str): docstring for the generated expansion function

    Returns:
        callable: Tape expansion function. The returned function accepts a :class:`~.QuantumTape`,
        and returns an expanded :class:`~.QuantumTape`.

    **Example**

    Let us construct an expansion function that expands a tape in order to
    decompose trainable multi-parameter gates. We allow for up to five expansion
    steps, which can be controlled with the argument ``depth``.
    The stopping criterion is easy to write as

    >>> stop_at = ~(qml.operation.has_multipar & qml.operation.is_trainable)

    Then the expansion function can be obtained via

    >>> expand_fn = qml.transforms.create_expand_fn(depth=5, stop_at=stop_at)

    We can test the newly generated function on an example tape:

    .. code-block:: python

        ops = [
            qml.RX(0.2, wires=0),
            qml.RX(qml.numpy.array(-2.4, requires_grad=True), wires=1),
            qml.Rot(1.7, 0.92, -1.1, wires=0),
            qml.Rot(*qml.numpy.array([-3.1, 0.73, 1.36], requires_grad=True), wires=1)
        ]
        tape = qml.tape.QuantumTape(ops)

    >>> new_tape = expand_fn(tape)
    >>> print(qml.drawer.tape_text(tape, decimals=1))
    0: ──RX(0.2)───Rot(1.7,0.9,-1.1)─┤
    1: ──RX(-2.4)──Rot(-3.1,0.7,1.4)─┤
    >>> print(qml.drawer.tape_text(new_tape, decimals=1))
    0: ──RX(0.2)───Rot(1.7,0.9,-1.1)───────────────────┤
    1: ──RX(-2.4)──RZ(-3.1)───────────RY(0.7)──RZ(1.4)─┤

    """
    # pylint: disable=unused-argument
    if device is not None:
        if stop_at is None:
            stop_at = device.stopping_condition
        else:
            stop_at &= device.stopping_condition

    def expand_fn(tape, depth=depth, **kwargs):
        with qml.QueuingManager.stop_recording():
            if not all(stop_at(op) for op in tape.operations):
                (tape,), _ = qml.transforms.decompose(tape, max_expansion=depth, gate_set=stop_at)
            else:
                return tape

            _update_trainable_params(tape)

        return tape

    if docstring:
        expand_fn.__doc__ = docstring

    return expand_fn


_expand_multipar_doc = """Expand out a tape so that all its parametrized
operations have a single parameter.

This is achieved by decomposing all parametrized operations that do not have
a generator, up to maximum depth ``depth``.
For a sufficient ``depth``, it should always be possible to obtain a tape containing
only single-parameter operations.

Args:
    tape (.QuantumTape): the input tape to expand
    depth (int) : the maximum expansion depth
    **kwargs: additional keyword arguments are ignored

Returns:
    .QuantumTape: the expanded tape
"""

expand_multipar = create_expand_fn(
    depth=None,
    stop_at=not_tape | is_measurement | has_nopar | (has_gen & ~gen_is_multi_term_hamiltonian),
    docstring=_expand_multipar_doc,
)

_expand_trainable_multipar_doc = """Expand out a tape so that all its trainable
operations have a single parameter.

This is achieved by decomposing all trainable operations that do not have
a generator, up to maximum depth ``depth``.
For a sufficient ``depth``, it should always be possible to obtain a tape containing
only single-parameter operations.

Args:
    tape (.QuantumTape): the input tape to expand
    depth (int) : the maximum expansion depth
    **kwargs: additional keyword arguments are ignored

Returns:
    .QuantumTape: the expanded tape
"""

expand_trainable_multipar = create_expand_fn(
    depth=None,
    stop_at=not_tape
    | is_measurement
    | has_nopar
    | (~is_trainable)
    | (has_gen & ~gen_is_multi_term_hamiltonian),
    docstring=_expand_trainable_multipar_doc,
)


def create_expand_trainable_multipar(tape, use_tape_argnum=False):
    """Creates the expand_trainable_multipar expansion transform with an option to include argnums."""

    if not use_tape_argnum:
        return expand_trainable_multipar

    # pylint: disable=protected-access
    trainable_par_info = [tape.par_info[i] for i in tape.trainable_params]
    trainable_ops = [info["op"] for info in trainable_par_info]

    @qml.BooleanFn
    def _is_trainable(obj):
        return obj in trainable_ops

    return create_expand_fn(
        depth=None,
        stop_at=not_tape
        | is_measurement
        | has_nopar
        | (~_is_trainable)
        | (has_gen & ~gen_is_multi_term_hamiltonian),
        docstring=_expand_trainable_multipar_doc,
    )


_expand_nonunitary_gen_doc = """Expand out a tape so that all its parametrized
operations have a unitary generator.

This is achieved by decomposing all parametrized operations that either do not have
a generator or have a non-unitary generator, up to maximum depth ``depth``.
For a sufficient ``depth``, it should always be possible to obtain a tape containing
only unitarily generated operations.

Args:
    tape (.QuantumTape): the input tape to expand
    depth (int) : the maximum expansion depth
    **kwargs: additional keyword arguments are ignored

Returns:
    .QuantumTape: the expanded tape
"""

expand_nonunitary_gen = create_expand_fn(
    depth=None,
    stop_at=not_tape | is_measurement | has_nopar | (has_gen & has_unitary_gen),
    docstring=_expand_nonunitary_gen_doc,
)

_expand_invalid_trainable_doc = """Expand out a tape so that it supports differentiation
of requested operations.

This is achieved by decomposing all trainable operations that have
``Operation.grad_method=None`` until all resulting operations
have a defined gradient method, up to maximum depth ``depth``. Note that this
might not be possible, in which case the gradient rule will fail to apply.

Args:
    tape (.QuantumTape): the input tape to expand
    depth (int) : the maximum expansion depth
    **kwargs: additional keyword arguments are ignored

Returns:
    .QuantumTape: the expanded tape
"""

expand_invalid_trainable = create_expand_fn(
    depth=None,
    stop_at=not_tape | is_measurement | (~is_trainable) | has_grad_method,
    docstring=_expand_invalid_trainable_doc,
)


@contextlib.contextmanager
def _custom_decomp_context(custom_decomps):
    """A context manager for applying custom decompositions of operations."""

    # Creates an individual context
    @contextlib.contextmanager
    def _custom_decomposition(obj, fn):
        # Covers the case where the user passes a string to indicate the Operator
        if isinstance(obj, str):
            obj = getattr(qml, obj)

        original_decomp_method = obj.compute_decomposition
        original_has_decomp_property = obj.has_decomposition

        try:
            # Explicitly set the new compute_decomposition method
            obj.compute_decomposition = staticmethod(fn)
            obj.has_decomposition = lambda obj: True
            yield

        finally:
            obj.compute_decomposition = staticmethod(original_decomp_method)
            obj.has_decomposition = original_has_decomp_property

    # Loop through the decomposition dictionary and create all the contexts
    try:
        with contextlib.ExitStack() as stack:
            for obj, fn in custom_decomps.items():
                # We enter a new context for each decomposition the user passes
                stack.enter_context(_custom_decomposition(obj, fn))

            stack = stack.pop_all()

        yield

    finally:
        stack.close()


def create_decomp_expand_fn(custom_decomps, dev, decomp_depth=None):
    """
    .. warning::
        Please use the :func:`qml.transforms.decompose <.transforms.decompose>` function for decomposing circuits.

    Creates a custom expansion function for a device that applies
    a set of specified custom decompositions.

    Args:
        custom_decomps (Dict[Union(str, qml.operation.Operation), Callable]): Custom
            decompositions to be applied by the device at runtime.
        dev (pennylane.devices.LegacyDevice): A quantum device.
        decomp_depth: The maximum depth of the expansion.

    Returns:
        Callable: A custom expansion function that a device can call to expand
        its tapes within a context manager that applies custom decompositions.

    **Example**

    Suppose we would like a custom expansion function that decomposes all CNOTs
    into CZs. We first define a decomposition function:

    .. code-block:: python

        def custom_cnot(wires):
            return [
                qml.Hadamard(wires=wires[1]),
                qml.CZ(wires=[wires[0], wires[1]]),
                qml.Hadamard(wires=wires[1])
            ]

    We then create the custom function (passing a device, in order to pick up any
    additional stopping criteria the expansion should have), and then register the
    result as a custom function of the device:

    >>> custom_decomps = {qml.CNOT : custom_cnot}
    >>> expand_fn = qml.transforms.create_decomp_expand_fn(custom_decomps, dev)
    >>> dev.custom_expand(expand_fn)
    """
    custom_op_names = [op if isinstance(op, str) else op.__name__ for op in custom_decomps.keys()]

    # Create a new expansion function; stop at things that do not have
    # custom decompositions, or that satisfy the regular device stopping criteria
    custom_fn = qml.transforms.create_expand_fn(
        decomp_depth,
        stop_at=qml.BooleanFn(lambda obj: obj.name not in custom_op_names),
        device=dev,
    )

    # Finally, we set the device's custom_expand_fn to a new one that
    # runs in a context where the decompositions have been replaced.
    def custom_decomp_expand(self, circuit, max_expansion=decomp_depth):
        with _custom_decomp_context(custom_decomps):
            return custom_fn(circuit, max_expansion=max_expansion)

    return custom_decomp_expand


def _create_decomp_preprocessing(custom_decomps, dev, decomp_depth=None):
    """Creates a custom preprocessing method for a device that applies
    a set of specified custom decompositions.

    Args:
        custom_decomps (Dict[Union(str, qml.operation.Operation), Callable]): Custom
            decompositions to be applied by the device at runtime.
        dev (pennylane.devices.Device): A quantum device.
        decomp_depth: The maximum depth of the expansion.

    Returns:
        Callable: A custom preprocessing method that a device can call to expand
        its tapes.

    **Example**

    Suppose we would like a custom expansion function that decomposes all CNOTs
    into CZs. We first define a decomposition function:

    .. code-block:: python

        def custom_cnot(wires):
            return [
                qml.Hadamard(wires=wires[1]),
                qml.CZ(wires=[wires[0], wires[1]]),
                qml.Hadamard(wires=wires[1])
            ]

    We then create the custom function (passing a device, in order to pick up any
    additional stopping criteria the expansion should have), and then register the
    result as a custom function of the device:

    >>> custom_decomps = {qml.CNOT : custom_cnot}
    >>> new_preprocessing = _create_decomp_preprocessing(custom_decomps, dev)
    >>> dev.preprocess = new_preprocessing
    """

    def decomposer(op):
        if isinstance(op, qml.ops.Controlled) and type(op.base) in custom_decomps:
            op.base.compute_decomposition = custom_decomps[type(op.base)]
            return op.decomposition()
        if op.name in custom_decomps:
            return custom_decomps[op.name](*op.data, wires=op.wires, **op.hyperparameters)
        if type(op) in custom_decomps:
            return custom_decomps[type(op)](*op.data, wires=op.wires, **op.hyperparameters)
        return op.decomposition()

    original_preprocess = dev.preprocess

    # pylint: disable=cell-var-from-loop
    def new_preprocess(execution_config=qml.devices.DefaultExecutionConfig):
        program, config = original_preprocess(execution_config)

        for container in program:
            if container.transform == qml.devices.preprocess.decompose.transform:
                container.kwargs["decomposer"] = decomposer
                container.kwargs["max_expansion"] = decomp_depth

                for cond in ["stopping_condition", "stopping_condition_shots"]:
                    # Devices that do not support native mid-circuit measurements
                    # will not have "stopping_condition_shots".
                    if cond in container.kwargs:
                        original_stopping_condition = container.kwargs[cond]

                        def stopping_condition(obj):
                            if obj.name in custom_decomps or type(obj) in custom_decomps:
                                return False
                            return original_stopping_condition(obj)

                        container.kwargs[cond] = stopping_condition

                break

        return program, config

    return new_preprocess


[docs]@contextlib.contextmanager
def set_decomposition(custom_decomps, dev, decomp_depth=None):
    """Context manager for setting custom decompositions.

    .. warning::
        The ``decomp_depth`` argument is deprecated and will be removed in version 0.41.

    Args:
        custom_decomps (Dict[Union(str, qml.operation.Operation), Callable]): Custom
            decompositions to be applied by the device at runtime.
        dev (pennylane.devices.LegacyDevice): A quantum device.
        decomp_depth: The maximum depth of the expansion.

    **Example**

    Suppose we would like a custom expansion function that decomposes all CNOTs
    into CZs. We first define a decomposition function:

    .. code-block:: python

        def custom_cnot(wires):
            return [
                qml.Hadamard(wires=wires[1]),
                qml.CZ(wires=[wires[0], wires[1]]),
                qml.Hadamard(wires=wires[1])
            ]

    This context manager can be used to temporarily change a devices expansion
    function to one that takes into account the custom decompositions.

    .. code-block:: python

        dev = qml.device("default.qubit", wires=2)

        @qml.qnode(dev)
        def circuit():
            qml.CNOT(wires=[0, 1])
            return qml.expval(qml.Z(0))

    >>> print(qml.draw(circuit, level=None)())
    0: ─╭●─┤  <Z>
    1: ─╰X─┤

    Now let's set up a context where the custom decomposition will be applied:

    >>> with qml.transforms.set_decomposition({qml.CNOT : custom_cnot}, dev):
    ...     print(qml.draw(circuit, wire_order=[0, 1])())
    0: ────╭●────┤  <Z>
    1: ──H─╰Z──H─┤

    """
    if decomp_depth is not None:
        warnings.warn(
            "The decomp_depth argument is deprecated and will be removed in version v0.41.",
            qml.PennyLaneDeprecationWarning,
        )

    if isinstance(dev, qml.devices.LegacyDeviceFacade):
        dev = dev.target_device

        original_custom_expand_fn = dev.custom_expand_fn

        # Create a new expansion function; stop at things that do not have
        # custom decompositions, or that satisfy the regular device stopping criteria
        new_custom_expand_fn = create_decomp_expand_fn(
            custom_decomps, dev, decomp_depth=decomp_depth
        )

        # Set the custom expand function within this context only
        try:
            dev.custom_expand(new_custom_expand_fn)
            yield

        finally:
            dev.custom_expand_fn = original_custom_expand_fn

    else:
        with warnings.catch_warnings():
            warnings.filterwarnings(
                action="ignore",
                message=r"max_expansion argument is deprecated",
                category=qml.PennyLaneDeprecationWarning,
            )
            original_preprocess = dev.preprocess
            new_preprocess = _create_decomp_preprocessing(
                custom_decomps, dev, decomp_depth=decomp_depth
            )

            try:
                dev.preprocess = new_preprocess
                yield

            finally:
                dev.preprocess = original_preprocess