Adding QFT gate to natively reason about Quantum Fourier Transforms (#…

…11463)

* initial commit

* release notes

* fixing synthesis plugin options

* finalize merge conflicts

* fixing default option values for qft plugins'

* additional tests for qft plugins

* renaming QftGate to QFTGate

* Also renaming Qft to QFT in synthesis method names

* appplying Jake's suggestion from code review

* inlining _basic_definition into _define

* docstring improvements

* more docstring improvements

* renaming do_swaps to reverse_qubits in the new code

* typos

* adding synth_qft_full to __init__

* round of suggestions from code review

* another round of code review suggestions

* fixes

* also adding QFTGate plugins to the docs

pyproject.toml

@@ -88,6 +88,9 @@ sk = "qiskit.transpiler.passes.synthesis.solovay_kitaev_synthesis:SolovayKitaevS
 "permutation.kms" = "qiskit.transpiler.passes.synthesis.high_level_synthesis:KMSSynthesisPermutation"
 "permutation.basic" = "qiskit.transpiler.passes.synthesis.high_level_synthesis:BasicSynthesisPermutation"
 "permutation.acg" = "qiskit.transpiler.passes.synthesis.high_level_synthesis:ACGSynthesisPermutation"
+"qft.full" = "qiskit.transpiler.passes.synthesis.high_level_synthesis:QFTSynthesisFull"
+"qft.line" = "qiskit.transpiler.passes.synthesis.high_level_synthesis:QFTSynthesisLine"
+"qft.default" = "qiskit.transpiler.passes.synthesis.high_level_synthesis:QFTSynthesisFull"
 "permutation.token_swapper" = "qiskit.transpiler.passes.synthesis.high_level_synthesis:TokenSwapperSynthesisPermutation"
 
 [project.entry-points."qiskit.transpiler.init"]

qiskit/circuit/library/__init__.py

@@ -226,6 +226,7 @@
    :template: autosummary/class_no_inherited_members.rst
 
    QFT
+   QFTGate
 
 Arithmetic Circuits
 ===================
@@ -523,7 +524,7 @@
     XOR,
     InnerProduct,
 )
-from .basis_change import QFT
+from .basis_change import QFT, QFTGate
 from .arithmetic import (
     FunctionalPauliRotations,
     LinearPauliRotations,

qiskit/circuit/library/basis_change/__init__.py

@@ -12,4 +12,4 @@
 
 """The basis change circuits."""
 
-from .qft import QFT
+from .qft import QFT, QFTGate

qiskit/circuit/library/basis_change/qft.py

@@ -10,14 +10,13 @@
 # copyright notice, and modified files need to carry a notice indicating
 # that they have been altered from the originals.
 
-"""Quantum Fourier Transform Circuit."""
+"""Define a Quantum Fourier Transform circuit (QFT) and a native gate (QFTGate)."""
 
-from typing import Optional
+from __future__ import annotations
 import warnings
 import numpy as np
 
-from qiskit.circuit import QuantumCircuit, QuantumRegister, CircuitInstruction
-
+from qiskit.circuit.quantumcircuit import QuantumCircuit, QuantumRegister, CircuitInstruction, Gate
 from ..blueprintcircuit import BlueprintCircuit
 
 
@@ -75,12 +74,12 @@ class QFT(BlueprintCircuit):
 
     def __init__(
         self,
-        num_qubits: Optional[int] = None,
+        num_qubits: int | None = None,
         approximation_degree: int = 0,
         do_swaps: bool = True,
         inverse: bool = False,
         insert_barriers: bool = False,
-        name: Optional[str] = None,
+        name: str | None = None,
     ) -> None:
         """Construct a new QFT circuit.
 
@@ -293,3 +292,38 @@ def _build(self) -> None:
 
         wrapped = circuit.to_instruction() if self.insert_barriers else circuit.to_gate()
         self.compose(wrapped, qubits=self.qubits, inplace=True)
+
+
+class QFTGate(Gate):
+    r"""Quantum Fourier Transform Gate.
+
+    The Quantum Fourier Transform (QFT) on :math:`n` qubits is the operation
+
+    .. math::
+
+        |j\rangle \mapsto \frac{1}{2^{n/2}} \sum_{k=0}^{2^n - 1} e^{2\pi ijk / 2^n} |k\rangle
+
+    """
+
+    def __init__(
+        self,
+        num_qubits: int,
+    ):
+        """
+        Args:
+            num_qubits: The number of qubits on which the QFT acts.
+        """
+        super().__init__(name="qft", num_qubits=num_qubits, params=[])
+
+    def __array__(self, dtype=complex):
+        """Return a numpy array for the QFTGate."""
+        n = self.num_qubits
+        nums = np.arange(2**n)
+        outer = np.outer(nums, nums)
+        return np.exp(2j * np.pi * outer * (0.5**n), dtype=dtype) * (0.5 ** (n / 2))
+
+    def _define(self):
+        """Provide a specific decomposition of the QFTGate into a quantum circuit."""
+        from qiskit.synthesis.qft import synth_qft_full
+
+        self.definition = synth_qft_full(num_qubits=self.num_qubits)

qiskit/qpy/binary_io/circuits.py

@@ -397,6 +397,8 @@ def _read_instruction(
             "DiagonalGate",
         }:
             gate = gate_class(params)
+        elif gate_name == "QFTGate":
+            gate = gate_class(len(qargs), *params)
         else:
             if gate_name == "Barrier":
                 params = [len(qargs)]

qiskit/synthesis/__init__.py

@@ -91,6 +91,7 @@
 ======================
 
 .. autofunction:: synth_qft_line
+.. autofunction:: synth_qft_full
 
 Unitary Synthesis
 =================
@@ -162,7 +163,7 @@
     synth_circuit_from_stabilizers,
 )
 from .discrete_basis import SolovayKitaevDecomposition, generate_basic_approximations
-from .qft import synth_qft_line
+from .qft import synth_qft_line, synth_qft_full
 from .unitary.qsd import qs_decomposition
 from .unitary import aqc
 from .one_qubit import OneQubitEulerDecomposer

qiskit/synthesis/qft/__init__.py

@@ -13,3 +13,4 @@
 """Module containing stabilizer QFT circuit synthesis."""
 
 from .qft_decompose_lnn import synth_qft_line
+from .qft_decompose_full import synth_qft_full

qiskit/synthesis/qft/qft_decompose_full.py

@@ -0,0 +1,79 @@
+# This code is part of Qiskit.
+#
+# (C) Copyright IBM 2024.
+#
+# This code is licensed under the Apache License, Version 2.0. You may
+# obtain a copy of this license in the LICENSE.txt file in the root directory
+# of this source tree or at http://www.apache.org/licenses/LICENSE-2.0.
+#
+# Any modifications or derivative works of this code must retain this
+# copyright notice, and modified files need to carry a notice indicating
+# that they have been altered from the originals.
+"""
+Circuit synthesis for a QFT circuit.
+"""
+
+from __future__ import annotations
+import numpy as np
+from qiskit.circuit.quantumcircuit import QuantumCircuit
+
+
+def synth_qft_full(
+    num_qubits: int,
+    do_swaps: bool = True,
+    approximation_degree: int = 0,
+    insert_barriers: bool = False,
+    inverse: bool = False,
+    name: str | None = None,
+) -> QuantumCircuit:
+    """Construct a circuit for the Quantum Fourier Transform using all-to-all connectivity.
+
+    .. note::
+
+        With the default value of ``do_swaps = True``, this synthesis algorithm creates a
+        circuit that faithfully implements the QFT operation. This circuit contains a sequence
+        of swap gates at the end, corresponding to reversing the order of its output qubits.
+        In some applications this reversal permutation can be avoided. Setting ``do_swaps = False``
+        creates a circuit without this reversal permutation, at the expense that this circuit
+        implements the "QFT-with-reversal" instead of QFT. Alternatively, the
+        :class:`~.ElidePermutations` transpiler pass is able to remove these swap gates.
+
+    Args:
+        num_qubits: The number of qubits on which the Quantum Fourier Transform acts.
+        do_swaps: Whether to synthesize the "QFT" or the "QFT-with-reversal" operation.
+        approximation_degree: The degree of approximation (0 for no approximation).
+            It is possible to implement the QFT approximately by ignoring
+            controlled-phase rotations with the angle beneath a threshold. This is discussed
+            in more detail in https://arxiv.org/abs/quant-ph/9601018 or
+            https://arxiv.org/abs/quant-ph/0403071.
+        insert_barriers: If ``True``, barriers are inserted for improved visualization.
+        inverse: If ``True``, the inverse Quantum Fourier Transform is constructed.
+        name: The name of the circuit.
+
+    Returns:
+        A circuit implementing the QFT operation.
+
+    """
+
+    circuit = QuantumCircuit(num_qubits, name=name)
+
+    for j in reversed(range(num_qubits)):
+        circuit.h(j)
+        num_entanglements = max(0, j - max(0, approximation_degree - (num_qubits - j - 1)))
+        for k in reversed(range(j - num_entanglements, j)):
+            # Use negative exponents so that the angle safely underflows to zero, rather than
+            # using a temporary variable that overflows to infinity in the worst case.
+            lam = np.pi * (2.0 ** (k - j))
+            circuit.cp(lam, j, k)
+
+        if insert_barriers:
+            circuit.barrier()
+
+    if do_swaps:
+        for i in range(num_qubits // 2):
+            circuit.swap(i, num_qubits - i - 1)
+
+    if inverse:
+        circuit = circuit.inverse()
+
+    return circuit

qiskit/synthesis/qft/qft_decompose_lnn.py

@@ -21,21 +21,29 @@
 def synth_qft_line(
     num_qubits: int, do_swaps: bool = True, approximation_degree: int = 0
 ) -> QuantumCircuit:
-    """Synthesis of a QFT circuit for a linear nearest neighbor connectivity.
-    Based on Fig 2.b in Fowler et al. [1].
+    """Construct a circuit for the Quantum Fourier Transform using linear
+    neighbor connectivity.
 
-    Note that this method *reverts* the order of qubits in the circuit,
-    compared to the original :class:`.QFT` code.
-    Hence, the default value of the ``do_swaps`` parameter is ``True``
-    since it produces a circuit with fewer CX gates.
+    The construction is based on Fig 2.b in Fowler et al. [1].
+
+    .. note::
+
+        With the default value of ``do_swaps = True``, this synthesis algorithm creates a
+        circuit that faithfully implements the QFT operation. When ``do_swaps = False``,
+        this synthesis algorithm creates a circuit that corresponds to "QFT-with-reversal":
+        applying the QFT and reversing the order of its output qubits.
 
     Args:
-        num_qubits: The number of qubits on which the QFT acts.
+        num_qubits: The number of qubits on which the Quantum Fourier Transform acts.
         approximation_degree: The degree of approximation (0 for no approximation).
-        do_swaps: Whether to include the final swaps in the QFT.
+            It is possible to implement the QFT approximately by ignoring
+            controlled-phase rotations with the angle beneath a threshold. This is discussed
+            in more detail in https://arxiv.org/abs/quant-ph/9601018 or
+            https://arxiv.org/abs/quant-ph/0403071.
+        do_swaps: Whether to synthesize the "QFT" or the "QFT-with-reversal" operation.
 
     Returns:
-        A circuit implementation of the QFT circuit.
+        A circuit implementing the QFT operation.
 
     References:
         1. A. G. Fowler, S. J. Devitt, and L. C. L. Hollenberg,