Daily rc sync to master (#5611)

Automatic sync from the release candidate to master during a feature
freeze.

---------

Co-authored-by: Mudit Pandey <[email protected]>
Co-authored-by: Astral Cai <[email protected]>
Co-authored-by: Vincent Michaud-Rioux <[email protected]>
Co-authored-by: David Wierichs <[email protected]>
Co-authored-by: lillian542 <[email protected]>
Co-authored-by: GitHub Actions Bot <>

doc/development/deprecations.rst

@@ -106,7 +106,7 @@ Completed deprecation cycles
   - Deprecated in v0.35
   - Removed in v0.36
 
-* ``MeasurementProcess.name`` and ``MeasurementProcess.data`` have been deprecated, as they contain
+* ``MeasurementProcess.name`` and ``MeasurementProcess.data`` have been removed, as they contain
   dummy values that are no longer needed.
 
   - Deprecated in v0.35

doc/releases/changelog-0.36.0.md

@@ -551,6 +551,9 @@
 
 <h3>Bug fixes 🐛</h3>
 
+* Cast the keys of the `CountsMP` measurements returned `dynamic_one_shot` to the type produced by `MeasurementValue.concretize`.
+  [(#5587)](https://github.com/PennyLaneAI/pennylane/pull/5587)
+
 * `null.qubit` now automatically supports any operation without a decomposition.
   [(#5582)](https://github.com/PennyLaneAI/pennylane/pull/5582)
 

pennylane/devices/default_qutrit_mixed.py

@@ -148,7 +148,7 @@ def f(x):
             program, execution_config = dev.preprocess()
             new_batch, post_processing_fn = program([qs])
             results = dev.execute(new_batch, execution_config=execution_config)
-            return post_processing_fn(results)
+            return post_processing_fn(results)[0]
 
     >>> f(jax.numpy.array(1.2))
     DeviceArray(0.36235774, dtype=float32)

pennylane/gradients/__init__.py

@@ -98,7 +98,7 @@
 
 .. code-block:: python
 
-    dev = qml.device("default.qubit", wires=2, shots=1000)
+    dev = qml.device("default.qubit", shots=1000)
 
     @qml.qnode(dev, interface="tf")
     def circuit(weights):
@@ -143,7 +143,7 @@ def circuit(weights):
 
 .. code-block:: python
 
-    dev = qml.device("default.qubit", wires=2)
+    dev = qml.device("default.qubit")
 
     @qml.qnode(dev)
     def circuit(weights):
@@ -157,9 +157,8 @@ def circuit(weights):
 >>> circuit(weights)
 tensor([0.9658079, 0.0341921], requires_grad=True)
 >>> qml.gradients.param_shift(circuit)(weights)
-(tensor([-0.04673668,  0.04673668], requires_grad=True),
- tensor([-0.09442394,  0.09442394], requires_grad=True),
- tensor([-0.14409127,  0.14409127], requires_grad=True))
+tensor([[-0.04673668, -0.09442394, -0.14409127],
+        [ 0.04673668,  0.09442394,  0.14409127]], requires_grad=True)
 
 Comparing this to autodifferentiation:
 
@@ -173,7 +172,7 @@ def circuit(weights):
 
 .. code-block:: python
 
-    dev = qml.device("default.qubit", wires=2)
+    dev = qml.device("default.qubit")
 
     @qml.gradients.param_shift
     @qml.qnode(dev)
@@ -185,9 +184,8 @@ def decorated_circuit(weights):
         return qml.probs(wires=1)
 
 >>> decorated_circuit(weights)
-(tensor([-0.04673668,  0.04673668], requires_grad=True),
- tensor([-0.09442394,  0.09442394], requires_grad=True),
- tensor([-0.14409127,  0.14409127], requires_grad=True))
+tensor([[-0.04673668, -0.09442394, -0.14409127],
+        [ 0.04673668,  0.09442394,  0.14409127]], requires_grad=True)
 
 .. note::
 
@@ -203,6 +201,9 @@ def decorated_circuit(weights):
     when applying the transform:
 
     >>> qml.gradients.param_shift(circuit, hybrid=False)(weights)
+    (tensor([-0.04673668,  0.04673668], requires_grad=True),
+     tensor([-0.09442394,  0.09442394], requires_grad=True),
+     tensor([-0.14409127,  0.14409127], requires_grad=True))
 
 
 Differentiating gradient transforms and higher-order derivatives
@@ -213,7 +214,7 @@ def decorated_circuit(weights):
 
 .. code-block:: python
 
-    dev = qml.device("default.qubit", wires=2)
+    dev = qml.device("default.qubit")
 
     @qml.qnode(dev)
     def circuit(weights):
@@ -227,9 +228,7 @@ def circuit(weights):
 >>> circuit(weights)
 tensor(0.9316158, requires_grad=True)
 >>> qml.gradients.param_shift(circuit)(weights)  # gradient
-(tensor(-0.09347337, requires_grad=True),
- tensor(-0.18884787, requires_grad=True),
- tensor(-0.28818254, requires_grad=True))
+tensor([-0.09347337, -0.18884787, -0.28818254], requires_grad=True)
 >>> def stacked_output(weights):
 ...     return qml.numpy.stack(qml.gradients.param_shift(circuit)(weights))
 >>> qml.jacobian(stacked_output)(weights)  # hessian
@@ -298,7 +297,7 @@ def circuit(weights):
 The output tapes can then be evaluated and post-processed to retrieve
 the gradient:
 
->>> dev = qml.device("default.qubit", wires=2)
+>>> dev = qml.device("default.qubit")
 >>> fn(qml.execute(gradient_tapes, dev, None))
 (tensor(-0.09347337, requires_grad=True),
  tensor(-0.18884787, requires_grad=True),

pennylane/gradients/classical_jacobian.py

@@ -62,8 +62,7 @@ def classical_jacobian(qnode, argnum=None, expand_fn=None, trainable_only=True):
     >>> print(cjac)
     [[1.  0.  0. ]
      [0.2 0.  0. ]
-     [0.  0.  0. ]
-     [0.  1.2 0. ]
+     [0.  2.  0. ]
      [0.  0.  1. ]]
 
     The returned Jacobian has rows corresponding to gate arguments, and columns

pennylane/gradients/finite_difference.py

@@ -254,7 +254,7 @@ def finite_diff(
     This transform can be registered directly as the quantum gradient transform
     to use during autodifferentiation:
 
-    >>> dev = qml.device("default.qubit", wires=2)
+    >>> dev = qml.device("default.qubit")
     >>> @qml.qnode(dev, interface="autograd", diff_method="finite-diff")
     ... def circuit(params):
     ...     qml.RX(params[0], wires=0)
@@ -271,7 +271,7 @@ def finite_diff(
     post-processing.
 
     >>> import jax
-    >>> dev = qml.device("default.qubit", wires=2)
+    >>> dev = qml.device("default.qubit")
     >>> @qml.qnode(dev, interface="jax", diff_method="finite-diff")
     ... def circuit(params):
     ...     qml.RX(params[0], wires=0)
@@ -281,7 +281,7 @@ def finite_diff(
     >>> params = jax.numpy.array([0.1, 0.2, 0.3])
     >>> jax.jacobian(circuit)(params)
     (Array([-0.38751727, -0.18884793, -0.3835571 ], dtype=float32),
-    Array([0.6991687 , 0.34072432, 0.6920237 ], dtype=float32))
+     Array([0.6991687 , 0.34072432, 0.6920237 ], dtype=float32))
 
 
     .. details::
@@ -300,12 +300,8 @@ def finite_diff(
         ...     return qml.expval(qml.Z(0)), qml.var(qml.Z(0))
         >>> params = np.array([0.1, 0.2, 0.3], requires_grad=True)
         >>> qml.gradients.finite_diff(circuit)(params)
-        ((tensor(-0.38751724, requires_grad=True),
-          tensor(-0.18884792, requires_grad=True),
-          tensor(-0.38355709, requires_grad=True)),
-         (tensor(0.69916868, requires_grad=True),
-          tensor(0.34072432, requires_grad=True),
-          tensor(0.69202366, requires_grad=True)))
+        (tensor([-0.38751724, -0.18884792, -0.38355708], requires_grad=True),
+         tensor([0.69916868, 0.34072432, 0.69202365], requires_grad=True))
 
         This quantum gradient transform can also be applied to low-level
         :class:`~.QuantumTape` objects. This will result in no implicit quantum
@@ -318,9 +314,9 @@ def finite_diff(
         >>> gradient_tapes, fn = qml.gradients.finite_diff(tape)
         >>> gradient_tapes
         [<QuantumTape: wires=[0], params=3>,
-         <QuantumTape: wires=[0], params=3>,
-         <QuantumTape: wires=[0], params=3>,
-         <QuantumTape: wires=[0], params=3>]
+         <QuantumScript: wires=[0], params=3>,
+         <QuantumScript: wires=[0], params=3>,
+         <QuantumScript: wires=[0], params=3>]
 
         This can be useful if the underlying circuits representing the gradient
         computation need to be analyzed.
@@ -339,32 +335,30 @@ def finite_diff(
 
         The output tapes can then be evaluated and post-processed to retrieve the gradient:
 
-        >>> dev = qml.device("default.qubit", wires=2)
+        >>> dev = qml.device("default.qubit")
         >>> fn(qml.execute(gradient_tapes, dev, None))
         ((tensor(-0.56464251, requires_grad=True),
-         tensor(-0.56464251, requires_grad=True),
-         tensor(-0.56464251, requires_grad=True)),
-        (tensor(0.93203912, requires_grad=True),
-         tensor(0.93203912, requires_grad=True),
-         tensor(0.93203912, requires_grad=True)))
+          tensor(-0.56464251, requires_grad=True),
+          tensor(-0.56464251, requires_grad=True)),
+         (tensor(0.93203912, requires_grad=True),
+          tensor(0.93203912, requires_grad=True),
+          tensor(0.93203912, requires_grad=True)))
 
         This gradient transform is compatible with devices that use shot vectors for execution.
 
         >>> shots = (10, 100, 1000)
-        >>> dev = qml.device("default.qubit", wires=2, shots=shots)
+        >>> dev = qml.device("default.qubit", shots=shots)
         >>> @qml.qnode(dev)
         ... def circuit(params):
         ...     qml.RX(params[0], wires=0)
         ...     qml.RY(params[1], wires=0)
         ...     qml.RX(params[2], wires=0)
         ...     return qml.expval(qml.Z(0)), qml.var(qml.Z(0))
         >>> params = np.array([0.1, 0.2, 0.3], requires_grad=True)
-        >>> qml.gradients.finite_diff(circuit, h=10e-2)(params)
-        (((array(-2.), array(-2.), array(0.)), (array(3.6), array(3.6), array(0.))),
-         ((array(1.), array(0.4), array(1.)),
-          (array(-1.62), array(-0.624), array(-1.62))),
-         ((array(-0.48), array(-0.34), array(-0.46)),
-          (array(0.84288), array(0.6018), array(0.80868))))
+        >>> qml.gradients.finite_diff(circuit, h=0.1)(params)
+        ((array([-2., -2.,  0.]), array([3.6, 3.6, 0. ])),
+         (array([1. , 0.2, 0.4]), array([-1.78 , -0.34 , -0.688])),
+         (array([-0.9 , -0.22, -0.48]), array([1.5498 , 0.3938 , 0.84672])))
 
         The outermost tuple contains results corresponding to each element of the shot vector.
     """

pennylane/gradients/general_shift_rules.py

@@ -311,7 +311,7 @@ def generate_shift_rule(frequencies, shifts=None, order=1):
            [ 0.5       , -3.14159265]])
 
     This corresponds to the shift rule
-    :math:`\frac{\partial^2 f}{\partial phi^2} = \frac{1}{2} \left[f(\phi) - f(\phi-\pi)\right]`.
+    :math:`\frac{\partial^2 f}{\partial \phi^2} = \frac{1}{2} \left[f(\phi) - f(\phi-\pi)\right]`.
     """
     frequencies = tuple(f for f in frequencies if f > 0)
     rule = _get_shift_rule(frequencies, shifts=shifts)
@@ -369,9 +369,12 @@ def generate_multi_shift_rule(frequencies, shifts=None, orders=None):
 
     .. math::
 
+        \begin{align*}
         \frac{\partial^2 f}{\partial x\partial y} &= \frac{1}{4}
-        \left[f(x+\pi/2, y+\pi/2) - f(x+\pi/2, y-\pi/2)\\
-        &~~~- f(x-\pi/2, y+\pi/2) + f(x-\pi/2, y-\pi/2) \right].
+        [f(x+\pi/2, y+\pi/2) - f(x+\pi/2, y-\pi/2)\\
+        &\phantom{\frac{1}{4}[}-f(x-\pi/2, y+\pi/2) + f(x-\pi/2, y-\pi/2) ].
+        \end{align*}
+
     """
     rules = []
     shifts = shifts or [None] * len(frequencies)

pennylane/gradients/hadamard_gradient.py

@@ -116,17 +116,18 @@ def hadamard_grad(
     to use during autodifferentiation:
 
     >>> import jax
-    >>> dev = qml.device("default.qubit", wires=2)
+    >>> dev = qml.device("default.qubit")
     >>> @qml.qnode(dev, interface="jax", diff_method="hadamard")
     ... def circuit(params):
     ...     qml.RX(params[0], wires=0)
     ...     qml.RY(params[1], wires=0)
     ...     qml.RX(params[2], wires=0)
-    ...     return qml.expval(qml.Z(0)), qml.var(qml.Z(0))
+    ...     return qml.expval(qml.Z(0)), qml.probs(wires=0)
     >>> params = jax.numpy.array([0.1, 0.2, 0.3])
     >>> jax.jacobian(circuit)(params)
-    (Array([-0.38751727, -0.18884793, -0.3835571 ], dtype=float32),
-    Array([0.6991687 , 0.34072432, 0.6920237 ], dtype=float32))
+    (Array([-0.3875172 , -0.18884787, -0.38355704], dtype=float64),
+     Array([[-0.1937586 , -0.09442394, -0.19177852],
+            [ 0.1937586 ,  0.09442394,  0.19177852]], dtype=float64))
 
     .. details::
         :title: Usage Details
@@ -136,7 +137,7 @@ def hadamard_grad(
         as the ``diff_method`` argument of the QNode decorator, and differentiating with your
         preferred machine learning framework.
 
-        >>> dev = qml.device("default.qubit", wires=2)
+        >>> dev = qml.device("default.qubit")
         >>> @qml.qnode(dev)
         ... def circuit(params):
         ...     qml.RX(params[0], wires=0)
@@ -145,23 +146,21 @@ def hadamard_grad(
         ...     return qml.expval(qml.Z(0))
         >>> params = np.array([0.1, 0.2, 0.3], requires_grad=True)
         >>> qml.gradients.hadamard_grad(circuit)(params)
-        (tensor([-0.3875172], requires_grad=True),
-         tensor([-0.18884787], requires_grad=True),
-         tensor([-0.38355704], requires_grad=True))
+        tensor([-0.3875172 , -0.18884787, -0.38355704], requires_grad=True)
 
         This quantum gradient transform can also be applied to low-level
         :class:`~.QuantumTape` objects. This will result in no implicit quantum
         device evaluation. Instead, the processed tapes, and post-processing
         function, which together define the gradient are directly returned:
 
-        >>> ops = [qml.RX(p, wires=0) for p in params]
+        >>> ops = [qml.RX(params[0], 0), qml.RY(params[1], 0), qml.RX(params[2], 0)]
         >>> measurements = [qml.expval(qml.Z(0))]
         >>> tape = qml.tape.QuantumTape(ops, measurements)
         >>> gradient_tapes, fn = qml.gradients.hadamard_grad(tape)
         >>> gradient_tapes
-        [<QuantumTape: wires=[0, 1], params=3>,
-         <QuantumTape: wires=[0, 1], params=3>,
-         <QuantumTape: wires=[0, 1], params=3>]
+        [<QuantumScript: wires=[0, 1], params=3>,
+         <QuantumScript: wires=[0, 1], params=3>,
+         <QuantumScript: wires=[0, 1], params=3>]
 
         This can be useful if the underlying circuits representing the gradient
         computation need to be analyzed.
@@ -180,14 +179,16 @@ def hadamard_grad(
 
         The output tapes can then be evaluated and post-processed to retrieve the gradient:
 
-        >>> dev = qml.device("default.qubit", wires=2)
+        >>> dev = qml.device("default.qubit")
         >>> fn(qml.execute(gradient_tapes, dev, None))
-        (array(-0.3875172), array(-0.18884787), array(-0.38355704))
+        (tensor(-0.3875172, requires_grad=True),
+         tensor(-0.18884787, requires_grad=True),
+         tensor(-0.38355704, requires_grad=True))
 
         This transform can be registered directly as the quantum gradient transform
         to use during autodifferentiation:
 
-        >>> dev = qml.device("default.qubit", wires=3)
+        >>> dev = qml.device("default.qubit")
         >>> @qml.qnode(dev, interface="jax", diff_method="hadamard")
         ... def circuit(params):
         ...     qml.RX(params[0], wires=0)
@@ -196,7 +197,7 @@ def hadamard_grad(
         ...     return qml.expval(qml.Z(0))
         >>> params = jax.numpy.array([0.1, 0.2, 0.3])
         >>> jax.jacobian(circuit)(params)
-        [-0.3875172  -0.18884787 -0.38355704]
+        Array([-0.3875172 , -0.18884787, -0.38355704], dtype=float64)
 
         If you use custom wires on your device, you need to pass an auxiliary wire.
 
@@ -210,24 +211,24 @@ def hadamard_grad(
         ...    return qml.expval(qml.Z("a"))
         >>> params = jax.numpy.array([0.1, 0.2, 0.3])
         >>> jax.jacobian(circuit)(params)
-        [-0.3875172  -0.18884787 -0.38355704]
+        Array([-0.3875172 , -0.18884787, -0.38355704], dtype=float64)
 
     .. note::
 
         ``hadamard_grad`` will decompose the operations that are not in the list of supported operations.
 
-        - ``pennylane.RX``
-        - ``pennylane.RY``
-        - ``pennylane.RZ``
-        - ``pennylane.Rot``
-        - ``pennylane.PhaseShift``
-        - ``pennylane.U1``
-        - ``pennylane.CRX``
-        - ``pennylane.CRY``
-        - ``pennylane.CRZ``
-        - ``pennylane.IsingXX``
-        - ``pennylane.IsingYY``
-        - ``pennylane.IsingZZ``
+        - :class:`~.pennylane.RX`
+        - :class:`~.pennylane.RY`
+        - :class:`~.pennylane.RZ`
+        - :class:`~.pennylane.Rot`
+        - :class:`~.pennylane.PhaseShift`
+        - :class:`~.pennylane.U1`
+        - :class:`~.pennylane.CRX`
+        - :class:`~.pennylane.CRY`
+        - :class:`~.pennylane.CRZ`
+        - :class:`~.pennylane.IsingXX`
+        - :class:`~.pennylane.IsingYY`
+        - :class:`~.pennylane.IsingZZ`
 
         The expansion will fail if a suitable decomposition in terms of supported operation is not found.
         The number of trainable parameters may increase due to the decomposition.