Merge branch 'master' into dm-wires

doc/development/deprecations.rst

@@ -23,6 +23,12 @@ Pending deprecations
 
   These keys are still accessible in v0.31 and will be removed in v0.32.
 
+* ``QuantumScript``'s ``name`` keyword argument and property are deprecated.
+  This also affects ``QuantumTape`` and ``OperationRecorder``.
+
+  - Deprecated in v0.31
+  - Will be removed in v0.32
+
 * The ``observables`` argument in ``QubitDevice.statistics`` is deprecated. Please use ``circuit``
   instead. Using a list of observables in ``QubitDevice.statistics`` is deprecated. Please use a
   ``QuantumTape`` instead.

doc/releases/changelog-dev.md

@@ -12,6 +12,10 @@
   given subspace.
   [(#2845)](https://github.com/PennyLaneAI/pennylane/pull/2845)
 
+* Added the `TRY` qutrit rotation operator, which allows applying a Y rotation on a
+  given subspace.
+  [(#2846)](https://github.com/PennyLaneAI/pennylane/pull/2846)
+
 <h3>Improvements 🛠</h3>
 
 * `DiagonalQubitUnitary` now decomposes into `RZ`, `IsingZZ` and `MultiRZ` gates
@@ -70,6 +74,10 @@
   the hash of any operators/measurement processes.
   [(#4087)](https://github.com/PennyLaneAI/pennylane/pull/4087)
 
+* All drawing methods changed their default value for the keyword argument `show_matrices`
+  to `True`. This allows quick insights into broadcasted tapes for example.
+  [(#3920)](https://github.com/PennyLaneAI/pennylane/pull/3920)
+
 * Support for adjoint differentiation has been added to the `DefaultQubit2` device.
   [(#4037)](https://github.com/PennyLaneAI/pennylane/pull/4037)
 
@@ -113,11 +121,23 @@
 * The new device interface in integrated with `qml.execute` for autograd, backpropagation, and no differentiation.
   [(#3903)](https://github.com/PennyLaneAI/pennylane/pull/3903)
 
+* `CZ` now inherits from the `ControlledOp` class. It now supports exponentiation to arbitrary powers with `pow`, which is no longer limited to integers. It also supports `sparse_matrix` and `decomposition` representations.
+  [(#4117)](https://github.com/PennyLaneAI/pennylane/pull/4117)
+
 * The construction of the pauli representation for the `Sum` class is now faster.
   [(#4142)](https://github.com/PennyLaneAI/pennylane/pull/4142)
 
+* Updated the `gradients` module to use the new `Shots` object internally.
+  [(#4152)](https://github.com/PennyLaneAI/pennylane/pull/4152)
+
+* The new device interface in integrated with `qml.execute` for Jax.
+  [(#4137)](https://github.com/PennyLaneAI/pennylane/pull/4137)
+
 <h3>Breaking changes 💔</h3>
 
+* All drawing methods changed their default value for the keyword argument `show_matrices` to `True`.
+  [(#3920)](https://github.com/PennyLaneAI/pennylane/pull/3920)
+
 * `DiagonalQubitUnitary` does not decompose into `QubitUnitary` any longer, but into `RZ`, `IsingZZ`
   and `MultiRZ` gates.
   [(#4035)](https://github.com/PennyLaneAI/pennylane/pull/4035)
@@ -141,6 +161,10 @@
 
 * `Operation.base_name` is deprecated. Please use `Operation.name` or `type(op).__name__` instead.
 
+* ``QuantumScript``'s ``name`` keyword argument and property are deprecated.
+  This also affects ``QuantumTape`` and ``OperationRecorder``.
+  [(#4141)](https://github.com/PennyLaneAI/pennylane/pull/4141)
+
 * `qml.grouping` module is removed. The functionality has been reorganized in the `qml.pauli` module.
 
 <h3>Documentation 📝</h3>
@@ -158,18 +182,23 @@
 * Removes a patch in `interfaces/autograd.py` that checks for the `strawberryfields.gbs` device.  That device
   is pinned to PennyLane <= v0.29.0, so that patch is no longer necessary.
 
+* `qml.pauli.are_identical_pauli_words` now treats all identities as equal. Identity terms on Hamiltonians with non-standard
+  wire orders are no longer eliminated.
+  [(#4161)](https://github.com/PennyLaneAI/pennylane/pull/4161)
+
 <h3>Contributors ✍️</h3>
 
 This release contains contributions from (in alphabetical order):
 
 Isaac De Vlugt,
 Soran Jahangiri,
 Edward Jiang,
-Korbinian Kottmann
+Korbinian Kottmann,
 Christina Lee,
 Vincent Michaud-Rioux,
 Romain Moyard,
 Mudit Pandey,
+Borja Requena,
 Matthew Silverman,
 Jay Soni,
 David Wierichs.

pennylane/_device.py

@@ -83,7 +83,7 @@ def _process_shot_sequence(shot_list):
         else:
             # Iterate through the shots, and group consecutive identical shots
             split_at_repeated = np.split(shot_list, np.diff(shot_list).nonzero()[0] + 1)
-            shot_vector = [ShotTuple(shots=i[0], copies=len(i)) for i in split_at_repeated]
+            shot_vector = [ShotTuple(shots=int(i[0]), copies=len(i)) for i in split_at_repeated]
 
     elif all(isinstance(s, (int, tuple)) for s in shot_list):
         # shot list contains tuples; assume it is already in a sparse representation

pennylane/devices/default_qutrit.py

@@ -62,6 +62,7 @@ class DefaultQutrit(QutritDevice):
         "TSWAP",
         "THadamard",
         "TRX",
+        "TRY",
     }
 
     # Identity is supported as an observable for qml.state() to work correctly. However, any

pennylane/drawer/draw.py

@@ -30,7 +30,7 @@ def draw(
     show_all_wires=False,
     decimals=2,
     max_length=100,
-    show_matrices=False,
+    show_matrices=True,
     expansion_strategy=None,
 ):
     """Create a function that draws the given qnode or quantum function.
@@ -100,15 +100,15 @@ def circuit(a, w):
     >>> print(qml.draw(circuit2)("x"))
     0: ──RX(x)─┤  <Z>
 
-    When requested with ``show_matrices=True``, matrix valued parameters are printed below the
-    circuit:
+    When requested with ``show_matrices=True`` (the default), matrix valued parameters
+    are printed below the circuit. For ``show_matrices=False``, they are not printed:
 
     >>> @qml.qnode(qml.device('default.qubit', wires=2))
     ... def circuit3():
     ...     qml.QubitUnitary(np.eye(2), wires=0)
     ...     qml.QubitUnitary(-np.eye(4), wires=(0,1))
     ...     return qml.expval(qml.Hermitian(np.eye(2), wires=1))
-    >>> print(qml.draw(circuit3, show_matrices=True)())
+    >>> print(qml.draw(circuit3)())
     0: ──U(M0)─╭U(M1)─┤
     1: ────────╰U(M1)─┤  <𝓗(M0)>
     M0 =
@@ -119,6 +119,9 @@ def circuit(a, w):
     [-0. -1. -0. -0.]
     [-0. -0. -1. -0.]
     [-0. -0. -0. -1.]]
+    >>> print(qml.draw(circuit3, show_matrices=False)())
+    0: ──U(M0)─╭U(M1)─┤
+    1: ────────╰U(M1)─┤  <𝓗(M0)>
 
     The ``max_length`` keyword warps long circuits:
 
@@ -232,7 +235,7 @@ def _draw_qnode(
     show_all_wires=False,
     decimals=2,
     max_length=100,
-    show_matrices=False,
+    show_matrices=True,
     expansion_strategy=None,
 ):
     @wraps(qnode)
@@ -261,10 +264,12 @@ def wrapper(*args, **kwargs):
                 )
                 for t in tapes[0]
             ]
-            if show_matrices:
-                mat_str = "\n"
+            if show_matrices and cache["matrices"]:
+                mat_str = ""
                 for i, mat in enumerate(cache["matrices"]):
                     mat_str += f"\nM{i} = \n{mat}"
+                if mat_str:
+                    mat_str = "\n" + mat_str
                 return "\n\n".join(res) + mat_str
             return "\n\n".join(res)
 

pennylane/drawer/tape_text.py

@@ -101,7 +101,7 @@ def tape_text(
     show_all_wires=False,
     decimals=None,
     max_length=100,
-    show_matrices=False,
+    show_matrices=True,
     cache=None,
 ):
     """Text based diagram for a Quantum Tape.
@@ -116,7 +116,7 @@ def tape_text(
             Default ``None`` will omit parameters from operation labels.
         max_length (Int) : Maximum length of a individual line.  After this length, the diagram will
             begin anew beneath the previous lines.
-        show_matrices=False (bool): show matrix valued parameters below all circuit diagrams
+        show_matrices=True (bool): show matrix valued parameters below all circuit diagrams
         cache (dict): Used to store information between recursive calls. Necessary keys are ``'tape_offset'``
             and ``'matrices'``.
 
@@ -204,7 +204,7 @@ def tape_text(
 
     Matrix valued parameters are always denoted by ``M`` followed by an integer corresponding to
     unique matrices.  The list of unique matrices can be printed at the end of the diagram by
-    selecting ``show_matrices=True``:
+    selecting ``show_matrices=True`` (the default):
 
     .. code-block:: python
 
@@ -213,7 +213,7 @@ def tape_text(
             qml.QubitUnitary(np.eye(2), wires=1)
             qml.expval(qml.Hermitian(np.eye(4), wires=(0,1)))
 
-    >>> print(qml.drawer.tape_text(tape, show_matrices=True))
+    >>> print(qml.drawer.tape_text(tape))
     0: ──U(M0)─┤ ╭<𝓗(M1)>
     1: ──U(M0)─┤ ╰<𝓗(M1)>
     M0 =
@@ -230,7 +230,7 @@ def tape_text(
     tape offset.
 
     >>> cache = {'matrices': [-np.eye(3)]}
-    >>> print(qml.drawer.tape_text(tape, show_matrices=True, cache=cache))
+    >>> print(qml.drawer.tape_text(tape, cache=cache))
     0: ──U(M1)─┤ ╭<𝓗(M2)>
     1: ──U(M1)─┤ ╰<𝓗(M2)>
     M0 =
@@ -332,6 +332,7 @@ def tape_text(
             totals = [filler.join([t, s]) for t, s in zip(totals, layer_str)]
         if ender:
             totals = [s + "─┤" for s in totals]
+            line_length += 2
 
     # Recursively handle nested tapes #
     tape_totals = "\n".join(finished_lines + totals)

pennylane/gradients/finite_difference.py

@@ -23,8 +23,7 @@
 from scipy.special import factorial
 
 import pennylane as qml
-from pennylane._device import _get_num_copies
-from pennylane.measurements import ProbabilityMP
+from pennylane.measurements import ProbabilityMP, Shots
 
 from .general_shift_rules import generate_shifted_tapes
 from .gradient_transform import (
@@ -157,21 +156,15 @@ def finite_diff_coeffs(n, approx_order, strategy):
     return coeffs_and_shifts
 
 
-def _processing_fn(results, shots=None, single_shot_batch_fn=None):
-    shot_vector = isinstance(shots, Sequence)
-
-    if not shot_vector:
-        grads_tuple = single_shot_batch_fn(results)
-    else:
-        grads_tuple = []
-        len_shot_vec = _get_num_copies(shots)
-        for idx in range(len_shot_vec):
-            res = [tape_res[idx] for tape_res in results]
-            g_tuple = single_shot_batch_fn(res)
-            grads_tuple.append(g_tuple)
-        grads_tuple = tuple(grads_tuple)
-
-    return grads_tuple
+def _processing_fn(results, shots: Shots = Shots(None), single_shot_batch_fn=None):
+    if not shots.has_partitioned_shots:
+        return single_shot_batch_fn(results)
+    grads_tuple = []
+    for idx in range(shots.num_copies):
+        res = [tape_res[idx] for tape_res in results]
+        g_tuple = single_shot_batch_fn(res)
+        grads_tuple.append(g_tuple)
+    return tuple(grads_tuple)
 
 
 @gradient_transform
@@ -345,6 +338,7 @@ def finite_diff(
             validate_params=validate_params,
             shots=shots,
         )
+    shots = Shots(shots)
     if argnum is None and not tape.trainable_params:
         return _no_trainable_grad(tape, shots)