Merge branch 'master' into fermi-commute

Makefile

@@ -60,12 +60,10 @@ clean-docs:
 
 test:
  $(PYTHON) $(TESTRUNNER)
- $(PYTHON) $(PLUGIN_TESTRUNNER) --device=default.qubit.autograd
 
 coverage:
  @echo "Generating coverage report..."
  $(PYTHON) $(TESTRUNNER) $(COVERAGE)
- $(PYTHON) $(PLUGIN_TESTRUNNER) --device=default.qubit.autograd $(COVERAGE) --cov-append
 
 .PHONY:format
 format:

doc/development/deprecations.rst

@@ -15,18 +15,6 @@ Pending deprecations
  - Deprecated in v0.39
  - Will be removed in v0.40
 
-* All of the legacy devices (any with the name ``default.qubit.{autograd,torch,tf,jax,legacy}``) are deprecated. Use ``default.qubit`` instead,
- as it supports backpropagation for the many backends the legacy devices support.
-
- - Deprecated in v0.38
- - Will be removed in v0.39
-
-* The logic for internally switching a device for a different backpropagation
- compatible device is now deprecated, as it was in place for the deprecated ``default.qubit.legacy``.
-
- - Deprecated in v0.38
- - Will be removed in v0.39 
-
 * The ``decomp_depth`` argument in ``qml.device`` is deprecated. 
 
  - Deprecated in v0.38
@@ -88,6 +76,18 @@ Other deprecations
 Completed deprecation cycles
 ----------------------------
 
+* All of the legacy devices (any with the name ``default.qubit.{autograd,torch,tf,jax,legacy}``) are removed. Use ``default.qubit`` instead,
+ as it supports backpropagation for the many backends the legacy devices support.
+
+ - Deprecated in v0.38
+ - Removed in v0.39
+
+* The logic for internally switching a device for a different backpropagation
+ compatible device is removed, as it was in place for removed ``default.qubit.legacy``.
+
+ - Deprecated in v0.38
+ - Removed in v0.39
+
 * `Operator.expand` is now removed. Use `qml.tape.QuantumScript(op.deocomposition())` instead.
 
  - Deprecated in v0.38

doc/releases/changelog-dev.md

@@ -62,10 +62,12 @@
 * Remove support for Python 3.9.
  [(#6223)](https://github.com/PennyLaneAI/pennylane/pull/6223)
 
-* `DefaultQubitTF`, `DefaultQubitTorch`, and `DefaultQubitJax` are removed. Please use `default.qubit` for all interfaces.
+* `DefaultQubitTF`, `DefaultQubitTorch`, `DefaultQubitJax`, and `DefaultQubitAutograd` are removed.
+ Please use `default.qubit` for all interfaces.
  [(#6207)](https://github.com/PennyLaneAI/pennylane/pull/6207)
  [(#6208)](https://github.com/PennyLaneAI/pennylane/pull/6208)
  [(#6209)](https://github.com/PennyLaneAI/pennylane/pull/6209)
+ [(#6210)](https://github.com/PennyLaneAI/pennylane/pull/6210)
 
 * `expand_fn`, `max_expansion`, `override_shots`, and `device_batch_transform` are removed from the
  signature of `qml.execute`.
@@ -84,6 +86,7 @@
 * `Operator.expand` is now removed. Use `qml.tape.QuantumScript(op.deocomposition())` instead.
  [(#6227)](https://github.com/PennyLaneAI/pennylane/pull/6227)
 
+
 <h3>Deprecations 👋</h3>
 
 * `QNode.gradient_fn` is deprecated. Please use `QNode.diff_method` and `QNode.get_gradient_fn` instead.

pennylane/devices/__init__.py

@@ -27,7 +27,6 @@
 
  default_qubit
  default_qubit_legacy
- default_qubit_autograd
  default_gaussian
  default_mixed
  default_qutrit
@@ -153,9 +152,6 @@ def execute(self, circuits, execution_config = qml.devices.DefaultExecutionConfi
 from .default_qubit import DefaultQubit
 from .legacy_facade import LegacyDeviceFacade
 
-# DefaultQubitTF and DefaultQubitAutograd not imported here since this
-# would lead to an automatic import of tensorflow and autograd, which are
-# not PennyLane core dependencies.
 # DefaultTensor is not imported here to avoid warnings
 # from quimb in case it is installed on the system.
 from .default_qubit_legacy import DefaultQubitLegacy

pennylane/devices/default_qubit_legacy.py

@@ -713,9 +713,7 @@ def capabilities(cls):
  supports_analytic_computation=True,
  supports_broadcasting=True,
  returns_state=True,
- passthru_devices={
- "autograd": "default.qubit.autograd",
- },
+ passthru_devices={},
  )
  return capabilities
 

pennylane/devices/legacy_facade.py

@@ -15,7 +15,6 @@
 Defines a LegacyDeviceFacade class for converting legacy devices to the
 new interface.
 """
-import warnings
 
 # pylint: disable=not-callable, unused-argument
 from contextlib import contextmanager
@@ -318,79 +317,6 @@ def supports_derivatives(self, execution_config=None, circuit=None) -> bool:
 
  return False
 
- # pylint: disable=protected-access
- def _create_temp_device(self, batch):
- """Create a temporary device for use in a backprop execution."""
- params = []
- for t in batch:
- params.extend(t.get_parameters(trainable_only=False))
- interface = qml.math.get_interface(*params)
- if interface == "numpy":
- return self._device
-
- mapped_interface = qml.workflow.execution.INTERFACE_MAP.get(interface, interface)
-
- backprop_interface = self._device.capabilities().get("passthru_interface", None)
- if mapped_interface == backprop_interface:
- return self._device
-
- backprop_devices = self._device.capabilities().get("passthru_devices", None)
-
- if backprop_devices is None:
- raise qml.DeviceError(f"Device {self} does not support backpropagation.")
-
- if backprop_devices[mapped_interface] == self._device.short_name:
- return self._device
-
- if self.target_device.short_name != "default.qubit.legacy":
- warnings.warn(
- "The switching of devices for backpropagation is now deprecated in v0.38 and "
- "will be removed in v0.39, as this behavior was developed purely for the "
- "deprecated default.qubit.legacy.",
- qml.PennyLaneDeprecationWarning,
- )
-
- # create new backprop device
- expand_fn = self._device.expand_fn
- batch_transform = self._device.batch_transform
- if hasattr(self._device, "_debugger"):
- debugger = self._device._debugger
- else:
- debugger = "No debugger"
- tracker = self._device.tracker
-
- with warnings.catch_warnings():
- warnings.filterwarnings(
- action="ignore",
- category=qml.PennyLaneDeprecationWarning,
- message=r"use 'default.qubit'",
- )
- # we already warned about backprop device switching
- new_device = qml.device(
- backprop_devices[mapped_interface],
- wires=self._device.wires,
- shots=self._device.shots,
- ).target_device
-
- new_device.expand_fn = expand_fn
- new_device.batch_transform = batch_transform
- if debugger != "No debugger":
- new_device._debugger = debugger
- new_device.tracker = tracker
-
- return new_device
-
- # pylint: disable=protected-access
- def _update_original_device(self, temp_device):
- """After performing an execution with a backprop device, update the state of the original device."""
- # Update for state vector simulators that have the _pre_rotated_state attribute
- if hasattr(self._device, "_pre_rotated_state"):
- self._device._pre_rotated_state = temp_device._pre_rotated_state
-
- # Update for state vector simulators that have the _state attribute
- if hasattr(self._device, "_state"):
- self._device._state = temp_device._state
-
  def _validate_backprop_method(self, tape):
  if tape.shots:
  return False
@@ -441,28 +367,18 @@ def _validate_device_method(self, _):
  return self._device.capabilities().get("provides_jacobian", False)
 
  def execute(self, circuits, execution_config=DefaultExecutionConfig):
- dev = (
- self._create_temp_device(circuits)
- if execution_config.gradient_method == "backprop"
- else self._device
- )
+ dev = self.target_device
 
  kwargs = {}
  if dev.capabilities().get("supports_mid_measure", False):
  kwargs["postselect_mode"] = execution_config.mcm_config.postselect_mode
 
  first_shot = circuits[0].shots
  if all(t.shots == first_shot for t in circuits):
- results = _set_shots(dev, first_shot)(dev.batch_execute)(circuits, **kwargs)
- else:
- results = tuple(
- _set_shots(dev, t.shots)(dev.batch_execute)((t,), **kwargs)[0] for t in circuits
- )
-
- if dev is not self._device:
- self._update_original_device(dev)
-
- return results
+ return _set_shots(dev, first_shot)(dev.batch_execute)(circuits, **kwargs)
+ return tuple(
+ _set_shots(dev, t.shots)(dev.batch_execute)((t,), **kwargs)[0] for t in circuits
+ )
 
  def execute_and_compute_derivatives(self, circuits, execution_config=DefaultExecutionConfig):
  first_shot = circuits[0].shots

pennylane/devices/tests/conftest.py

@@ -36,7 +36,6 @@
 # List of all devices that are included in PennyLane
 LIST_CORE_DEVICES = {
  "default.qubit",
- "default.qubit.autograd",
 }
 
 

pennylane/tape/qscript.py

@@ -555,7 +555,7 @@ def trainable_params(self) -> list[int]:
  .. note::
 
  For devices that support native backpropagation (such as
- ``default.qubit.tf`` and ``default.qubit.autograd``), this
+ ``default.qubit`` and ``default.mixed``), this
  property contains no relevant information when using
  backpropagation to compute gradients.
 

pennylane/workflow/execution.py

@@ -165,7 +165,7 @@ def _get_ml_boundary_execute(
  else:
  from .interfaces.jax import jax_jvp_execute as ml_boundary
 
- except ImportError as e: # pragma: no-cover
+ except ImportError as e: # pragma: no cover
  raise qml.QuantumFunctionError(
  f"{mapped_interface} not found. Please install the latest "
  f"version of {mapped_interface} to enable the '{mapped_interface}' interface."

setup.py

@@ -51,7 +51,6 @@
  "default.qubit = pennylane.devices:DefaultQubit",
  "default.qubit.legacy = pennylane.devices:DefaultQubitLegacy",
  "default.gaussian = pennylane.devices:DefaultGaussian",
- "default.qubit.autograd = pennylane.devices.default_qubit_autograd:DefaultQubitAutograd",
  "default.mixed = pennylane.devices.default_mixed:DefaultMixed",
  "null.qubit = pennylane.devices.null_qubit:NullQubit",
  "default.qutrit = pennylane.devices.default_qutrit:DefaultQutrit",