Add MCMT Pauli Bloqs from Cirq-FT

qualtran/bloqs/multi_control_multi_target_pauli.py

@@ -0,0 +1,117 @@
+#  Copyright 2023 Google LLC
+#
+#  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
+#
+#      https://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.
+
+from typing import Tuple
+
+import cirq
+import numpy as np
+from attrs import field, frozen
+from cirq._compat import cached_property
+from numpy.typing import NDArray
+
+from qualtran import GateWithRegisters, Register, Signature
+from qualtran.bloqs.and_bloq import And, MultiAnd
+from qualtran.cirq_interop.t_complexity_protocol import t_complexity, TComplexity
+
+
+@frozen
+class MultiTargetCNOT(GateWithRegisters):
+    """Implements single control, multi-target CNOT_{n} gate in 2*log(n) + 1 CNOT depth.
+
+    Implements CNOT_{n} = |0><0| I + |1><1| X^{n} using a circuit of depth 2*log(n) + 1
+    containing only CNOT gates. See Appendix B.1 of https://arxiv.org/abs/1812.00954 for
+    reference.
+    """
+
+    bitsize: int
+
+    @cached_property
+    def signature(self) -> Signature:
+        return Signature.build(control=1, targets=self.bitsize)
+
+    def decompose_from_registers(
+        self,
+        *,
+        context: cirq.DecompositionContext,
+        **quregs: NDArray[cirq.Qid],  # type:ignore[type-var]
+    ):
+        control, targets = quregs['control'], quregs['targets']
+
+        def cnots_for_depth_i(i: int, q: NDArray[cirq.Qid]) -> cirq.OP_TREE:
+            for c, t in zip(q[: 2**i], q[2**i : min(len(q), 2 ** (i + 1))]):
+                yield cirq.CNOT(c, t)
+
+        depth = len(targets).bit_length()
+        for i in range(depth):
+            yield cirq.Moment(cnots_for_depth_i(depth - i - 1, targets))
+        yield cirq.CNOT(*control, targets[0])
+        for i in range(depth):
+            yield cirq.Moment(cnots_for_depth_i(i, targets))
+
+    def _circuit_diagram_info_(self, _) -> cirq.CircuitDiagramInfo:
+        return cirq.CircuitDiagramInfo(wire_symbols=["@"] + ["X"] * self.bitsize)
+
+
+@frozen
+class MultiControlPauli(GateWithRegisters):
+    """Implements multi-control, single-target C^{n}P gate.
+
+    Implements $C^{n}P = (1 - |1^{n}><1^{n}|) I + |1^{n}><1^{n}| P^{n}$ using $n-1$
+    clean ancillas using a multi-controlled `AND` gate.
+
+    References:
+        [Constructing Large Controlled Nots]
+        (https://algassert.com/circuits/2015/06/05/Constructing-Large-Controlled-Nots.html)
+    """
+
+    cvs: Tuple[int, ...] = field(converter=lambda v: (v,) if isinstance(v, int) else tuple(v))
+    target_gate: cirq.Pauli = cirq.X
+
+    @cached_property
+    def signature(self) -> Signature:
+        return Signature.build(controls=len(self.cvs), target=1)
+
+    def decompose_from_registers(
+        self, *, context: cirq.DecompositionContext, **quregs: NDArray['cirq.Qid']
+    ) -> cirq.OP_TREE:
+        controls, target = quregs['controls'], quregs['target']
+        qm = context.qubit_manager
+        and_ancilla, and_target = np.array(qm.qalloc(len(self.cvs) - 2)), qm.qalloc(1)
+        ctrl, junk = controls[:, np.newaxis], and_ancilla[:, np.newaxis]
+        if len(self.cvs) == 2:
+            and_op = And(*self.cvs).on_registers(ctrl=ctrl, target=and_target)
+        else:
+            and_op = MultiAnd(self.cvs).on_registers(ctrl=ctrl, junk=junk, target=and_target)
+        yield and_op
+        yield self.target_gate.on(*target).controlled_by(*and_target)
+        yield and_op**-1
+        qm.qfree([*and_ancilla, *and_target])
+
+    def _circuit_diagram_info_(self, _) -> cirq.CircuitDiagramInfo:
+        wire_symbols = ["@" if b else "@(0)" for b in self.cvs]
+        wire_symbols += [str(self.target_gate)]
+        return cirq.CircuitDiagramInfo(wire_symbols=wire_symbols)
+
+    def _t_complexity_(self) -> TComplexity:
+        and_gate = And(*self.cvs) if len(self.cvs) == 2 else MultiAnd(self.cvs)
+        and_cost = t_complexity(and_gate)
+        controlled_pauli_cost = t_complexity(self.target_gate.controlled(1))
+        and_inv_cost = t_complexity(and_gate**-1)
+        return and_cost + controlled_pauli_cost + and_inv_cost
+
+    def _apply_unitary_(self, args: 'cirq.ApplyUnitaryArgs') -> np.ndarray:
+        return cirq.apply_unitary(self.target_gate.controlled(control_values=self.cvs), args)
+
+    def _has_unitary_(self) -> bool:
+        return True

qualtran/bloqs/multi_control_multi_target_pauli_test.py

@@ -0,0 +1,43 @@
+#  Copyright 2023 Google LLC
+#
+#  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
+#
+#      https://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.
+
+import cirq
+import numpy as np
+import pytest
+
+from qualtran.bloqs.multi_control_multi_target_pauli import MultiControlPauli, MultiTargetCNOT
+from qualtran.cirq_interop.testing import assert_decompose_is_consistent_with_t_complexity
+from qualtran.testing import assert_valid_bloq_decomposition
+
+
+@pytest.mark.parametrize("num_targets", [3, 4, 6, 8, 10])
+def test_multi_target_cnot(num_targets):
+    qubits = cirq.LineQubit.range(num_targets + 1)
+    naive_circuit = cirq.Circuit(cirq.CNOT(qubits[0], q) for q in qubits[1:])
+    op = MultiTargetCNOT(num_targets).on(*qubits)
+    cirq.testing.assert_circuits_with_terminal_measurements_are_equivalent(
+        cirq.Circuit(op), naive_circuit, atol=1e-6
+    )
+    optimal_circuit = cirq.Circuit(cirq.decompose_once(op))
+    assert len(optimal_circuit) == 2 * np.ceil(np.log2(num_targets)) + 1
+    assert_valid_bloq_decomposition(op.gate)
+
+
+@pytest.mark.parametrize("num_controls", [*range(7, 17)])
+@pytest.mark.parametrize("pauli", [cirq.X, cirq.Y, cirq.Z])
+@pytest.mark.parametrize('cv', [0, 1])
+def test_t_complexity_mcp(num_controls: int, pauli: cirq.Pauli, cv: int):
+    gate = MultiControlPauli([cv] * num_controls, target_gate=pauli)
+    assert_valid_bloq_decomposition(gate)
+    assert_decompose_is_consistent_with_t_complexity(gate)