Source code for mitiq.zne.scaling.parameter

# Copyright (C) 2020 Unitary Fund
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# GNU General Public License for more details.
# You should have received a copy of the GNU General Public License
# along with this program.  If not, see <>.

from typing import Optional, Callable, List, cast
import numpy as np

import copy

from cirq import Circuit, EigenGate, Moment
from cirq import (
from mitiq.interface import noise_scaling_converter
from mitiq import QPROGRAM

[docs]class GateTypeException(Exception): pass
def _get_base_gate(gate: EigenGate) -> EigenGate: BASE_GATES = [ZPowGate, HPowGate, XPowGate, YPowGate, CXPowGate, CZPowGate] for base_gate in BASE_GATES: if isinstance(gate, base_gate): return cast(EigenGate, base_gate) raise GateTypeException( "Must have circuit be made of rotation gates. " "Your gate {} may not be supported".format(gate) )
[docs]class CircuitMismatchException(Exception): pass
def _generate_parameter_calibration_circuit( qubits: List[Qid], depth: int, gate: EigenGate ) -> Circuit: """Generates a circuit which should be the identity. Given a rotation gate R(param), it applies R(2 * pi / depth) depth times, resulting in R(2*pi). Requires that the gate is periodic in 2*pi. Args: qubits: A list of qubits. depth: The length of the circuit to create. gate: The base gate to apply several times, must be periodic in 2*pi. Returns: A parameter calibration circuit that can be used for estimating the parameter noise of the input gate. """ num_qubits = gate().num_qubits() if num_qubits != len(qubits): raise CircuitMismatchException( "Number of qubits does not match domain size of gate." ) return Circuit( gate(exponent=2 * np.pi / depth).on(*qubits) for _ in range(depth) )
[docs]def compute_parameter_variance( executor: Callable[..., float], gate: EigenGate, qubit: Qid, depth: int = 100, ) -> float: """Given an executor and a gate, determines the effective variance in the control parameter that can be used as the ``base_variance`` argument in ``mitiq.zne.scaling.scale_parameters``. Note: Only works for one qubit gates for now. Args: executor: A function that takes in a quantum circuit and returns an expectation value. gate: The quantum gate that you wish to profile. qubit: The index of the qubit you wish to profile. depth: The number of operations you would like to use to profile your gate. Returns: The estimated variance of the control parameter. """ base_gate = _get_base_gate(gate) circuit = _generate_parameter_calibration_circuit( [qubit], depth, base_gate ) expectation = executor(circuit) error_prob = (1 - np.power(2 * expectation - 1, 1 / depth)) / 2 variance = -0.5 * np.log(1 - 2 * error_prob) return variance
[docs]@noise_scaling_converter def scale_parameters( circuit: QPROGRAM, scale_factor: float, base_variance: float, seed: Optional[int] = None, ) -> Circuit: """Applies parameter-noise scaling to the input circuit, assuming that each gate has the same base level of noise. Args: circuit: The circuit to scale as a QPROGRAM. All measurements should be in the last moment of the circuit. scale_factor: The amount to scale the base noise level by. base_variance: The base level (variance) of parameter noise, assumed to be the same for each gate of the circuit. seed: Optional seed for random number generator. Returns: The parameter noise scaled circuit. """ final_moments = [] noise = (scale_factor - 1) * base_variance rng = np.random.RandomState(seed) for moment in circuit: curr_moment = [] for op in moment.operations: # type: ignore gate = copy.deepcopy(op.gate) qubits = op.qubits if isinstance(gate, MeasurementGate): curr_moment.append(gate(*qubits)) else: assert isinstance(gate, EigenGate) base_gate = _get_base_gate(gate) param = gate.exponent * np.pi error = rng.normal(loc=0.0, scale=np.sqrt(noise)) new_param = param + error curr_moment.append( base_gate(exponent=new_param / np.pi)(*qubits) ) final_moments.append(Moment(curr_moment)) return Circuit(final_moments)