Source code for qibo.models.error_mitigation

"""Error Mitigation Methods."""

import math
from functools import cache, reduce
from inspect import signature
from operator import mul
from typing import List, Optional, Tuple, Union

import numpy as np
from numpy.typing import ArrayLike
from scipy.optimize import curve_fit

from qibo import gates
from qibo.backends import (
    Backend,
    _check_backend,
    get_backend,
)
from qibo.config import raise_error
from qibo.hamiltonians.hamiltonians import Hamiltonian, SymbolicHamiltonian
from qibo.noise import NoiseModel
from qibo.result import CircuitResult
from qibo.symbols import X, Y, Z


# all this roundabout due to circular imports
@cache
def SIMULATION_BACKEND():
    """Cached Numpy backend."""
    from qibo.backends import NumpyBackend  # pylint: disable=import-outside-toplevel

    return NumpyBackend()


@cache
def CLIFFORD_BACKEND(platform: str = "numpy"):
    """Cached Clifford backend."""
    from qibo.backends import CliffordBackend  # pylint: disable=import-outside-toplevel

    return CliffordBackend(platform)


[docs]def get_gammas(noise_levels: ArrayLike, analytical: bool = True): """Standalone function to compute the ZNE coefficients given the noise levels. Args: noise_levels (numpy.ndarray): array containing the different noise levels. Note that in the CNOT insertion paradigm this corresponds to the number of CNOT pairs to be inserted. The canonical ZNE noise levels are obtained as ``2 * c + 1``. analytical (bool, optional): if ``True``, computes the coeffients by solving the linear system. If ``False``, use the analytical solution valid for the CNOT insertion method. Default is ``True``. Returns: numpy.ndarray: the computed coefficients. """ if analytical: noise_levels = 2 * noise_levels + 1 a_matrix = np.array([noise_levels**power for power in range(len(noise_levels))]) b_vector = np.zeros(len(noise_levels)) b_vector[0] = 1 zne_coefficients = np.linalg.solve(a_matrix, b_vector) else: max_noise_level = noise_levels[-1] zne_coefficients = np.array( [ 1 / (2 ** (2 * max_noise_level) * math.factorial(item)) * (-1) ** item / (1 + 2 * item) * math.factorial(1 + 2 * max_noise_level) / ( math.factorial(max_noise_level) * math.factorial(max_noise_level - item) ) for item in noise_levels ] ) return zne_coefficients
[docs]def get_noisy_circuit( circuit: "Circuit", # type: ignore num_insertions: int, global_unitary_folding: bool = True, insertion_gate: bool = None, ): """Standalone function to generate the noisy circuit with the inverse gate pairs insertions. Args: circuit (:class:`qibo.models.circuit.Circuit`): circuit to modify. num_insertions (int): number of insertion gate pairs / global unitary folds to add. global_unitary_folding (bool): If ``True``, noise is increased by global unitary folding. If ``False``, local unitary folding is used. Defaults to ``True``. insertion_gate (str, optional): gate to be folded in the local unitary folding. If ``RX``, the gate used is :math:``RX(\\pi / 2)``. Otherwise, it is the ``CNOT`` gate. Returns: :class:`qibo.models.Circuit`: circuit with the inserted gate pairs or with global folding. """ from qibo import Circuit # pylint: disable=import-outside-toplevel if global_unitary_folding: copy_circuit = Circuit(**circuit.init_kwargs) for g in circuit.queue: if not isinstance(g, gates.M): copy_circuit.add(g) noisy_circuit = copy_circuit for _ in range(num_insertions): noisy_circuit += copy_circuit.invert() + copy_circuit for m in circuit.measurements: # pragma: no cover noisy_circuit.add(m) else: if insertion_gate is None or insertion_gate not in ( "CNOT", "RX", ): # pragma: no cover raise_error( ValueError, "Invalid insertion gate specification. Please select between 'CNOT' and 'RX'.", ) if insertion_gate == "CNOT" and circuit.nqubits < 2: # pragma: no cover raise_error( ValueError, "Provide a circuit with at least 2 qubits when using the 'CNOT' insertion gate. " + "Alternatively, try with the 'RX' insertion gate instead.", ) i_gate = gates.CNOT if insertion_gate == "CNOT" else gates.RX theta = np.pi / 2 noisy_circuit = Circuit(**circuit.init_kwargs) for gate in circuit.queue: noisy_circuit.add(gate) if isinstance(gate, i_gate): if insertion_gate == "CNOT": control = gate.control_qubits[0] target = gate.target_qubits[0] for _ in range(num_insertions): noisy_circuit.add(gates.CNOT(control, target)) noisy_circuit.add(gates.CNOT(control, target)) elif insertion_gate == "RX": qubit = gate.qubits[0] theta = gate.init_kwargs["theta"] for _ in range(num_insertions): noisy_circuit.add(gates.RX(qubit, theta=theta)) noisy_circuit.add(gates.RX(qubit, theta=-theta)) return noisy_circuit
[docs]def ZNE( circuit: "Circuit", # type: ignore observable: Union[Hamiltonian, SymbolicHamiltonian], noise_levels: ArrayLike, noise_model: NoiseModel = None, nshots: int = 10000, solve_for_gammas: bool = False, global_unitary_folding: bool = True, insertion_gate: str = "CNOT", readout: dict = None, qubit_map: List[int] = None, seed: Optional[int] = None, backend: Optional[Backend] = None, ): """Runs the Zero Noise Extrapolation (ZNE) method for error mitigation. The different noise levels are realized by the insertion of pairs of either ``CNOT`` or ``RX(pi/2)`` gates that resolve to the identiy in the noise-free case. Args: circuit (:class:`qibo.models.Circuit`): input circuit. observable (:class:`qibo.hamiltonians.Hamiltonian` or :class:`qibo.hamiltonians.SymbolicHamiltonian`): Observable to measure. noise_levels (ndarray): Sequence of noise levels. noise_model (:class:`qibo.noise.NoiseModel`, optional): Noise model applied to simulate noisy computation. nshots (int, optional): Number of shots. Defaults to :math:`10000`. If ``None``, the circuit is simulated without sampling: with statevector simulation if `noise_model` is not `None` and with density matrix simulation otherwise. Readout mitigation is not available with exact simulation. solve_for_gammas (bool, optional): If ``True``, explicitly solve the equations to obtain the ``gamma`` coefficients. Default is ``False``. global_unitary_folding (bool, optional): If ``True``, noise is increased by global unitary folding. If ``False``, local unitary folding is used. Defaults to ``True``. insertion_gate (str, optional): gate to be folded in the local unitary folding. If ``RX``, the gate used is :math:``RX(\\pi / 2)``. Otherwise, it is the :class:`qibo.gates.gates.CNOT` gate. readout (dict, optional): a dictionary that may contain the following keys: * ncircuits: int, specifies the number of random circuits to use for the randomized method of readout error mitigation. * response_matrix: ``ndarray``, used for applying a pre-computed response matrix for readout error mitigation. * ibu_iters: int, specifies the number of iterations for the iterative Bayesian unfolding method of readout error mitigation. If provided, the corresponding readout error mitigation method is used. Defaults to an empty ``dict``. qubit_map (list, optional): the qubit map. If None, a list of range of circuit's qubits is used. Defaults to ``None``. seed (int or :class:`numpy.random.Generator`, optional): Either a generator of random numbers or a fixed seed to initialize a generator. If ``None``, initializes a generator with a random seed. Default: ``None``. backend (:class:`qibo.backends.abstract.Backend`, optional): backend to be used in the execution. If ``None``, it uses the current backend. Defaults to ``None``. Returns: ArrayLike: Estimate of the expected value of ``observable`` in the noise free condition. Reference: 1. K. Temme, S. Bravyi et al, *Error mitigation for short-depth quantum circuits*. `Phys. Rev. Lett. 119, 180509 (2017) <https://doi.org/10.1103/PhysRevLett.119.180509>`_. """ backend = _check_backend(backend) backend.set_seed(seed) if readout is None: readout = {} expected_values = [] for num_insertions in noise_levels: noisy_circuit = get_noisy_circuit( circuit, num_insertions, global_unitary_folding, insertion_gate=insertion_gate, ) if nshots is None: circuit_result = _execute_circuit( noisy_circuit, qubit_map, noise_model, nshots, backend=backend ) val = observable.expectation_from_state(circuit_result.state()) else: val = get_expectation_val_with_readout_mitigation( noisy_circuit, observable, noise_model, nshots, readout, qubit_map, backend=backend, ) expected_values.append(val) expected_values = backend.cast(expected_values, dtype=type(expected_values[0])) gamma = get_gammas(noise_levels, analytical=solve_for_gammas) gamma = backend.cast(gamma, dtype=gamma.dtype) return backend.sum(gamma * expected_values)
[docs]def sample_training_circuit_cdr( circuit: "Circuit", # type: ignore replacement_gates: List[Tuple[str, dict]] = None, sigma: float = 0.5, seed: Optional[int] = None, backend: Optional[Backend] = None, ): """Samples a training circuit for CDR by susbtituting some of the non-Clifford gates. Args: circuit (:class:`qibo.models.Circuit`): circuit to sample from, decomposed in ``RX(pi/2)``, ``X``, ``CNOT`` and ``RZ`` gates. replacement_gates (list, optional): candidates for the substitution of the non-Clifford gates. The ``list`` should be composed by ``tuples`` of the form (``gates.XYZ``, ``kwargs``). For example, phase gates are used by default: ``list((RZ, {'theta':0}), (RZ, {'theta':pi/2}), (RZ, {'theta':pi}), (RZ, {'theta':3*pi/2}))``. sigma (float, optional): standard devation of the Gaussian distribution used for sampling. seed (int or :class:`numpy.random.Generator`, optional): Either a generator of random numbers or a fixed seed to initialize a generator. If ``None``, initializes a generator with a random seed. Default: ``None``. backend (:class:`qibo.backends.abstract.Backend`, optional): backend to be used in the execution. If ``None``, it uses the current backend. Defaults to ``None``. Returns: :class:`qibo.models.Circuit`: The sampled circuit. """ backend = _check_backend(backend) backend.set_seed(seed) if replacement_gates is None: replacement_gates = [(gates.RZ, {"theta": n * np.pi / 2}) for n in range(4)] gates_to_replace = [] for i, gate in enumerate(circuit.queue): if isinstance(gate, gates.RZ): if not gate.clifford: gates_to_replace.append((i, gate)) if not gates_to_replace: raise_error(ValueError, "No non-Clifford RZ gate found, no circuit sampled.") replacement, distance = [], [] for _, gate in gates_to_replace: rep_gates = np.array( [rg(*gate.init_args, **kwargs) for rg, kwargs in replacement_gates] ) replacement.append(rep_gates) gate_matrix = gate.matrix(backend) rep_gate_matrix = [rep_gate.matrix(backend) for rep_gate in rep_gates] rep_gate_matrix = backend.cast(rep_gate_matrix, dtype=rep_gate_matrix[0].dtype) matrix_norm = backend.matrix_norm( gate_matrix - rep_gate_matrix, "fro", axis=(1, 2) ) distance.append(backend.real(matrix_norm)) distance = backend.vstack(distance) prob = backend.exp(-(distance**2) / sigma**2) index = backend.random_choice( range(len(gates_to_replace)), size=min(int(len(gates_to_replace) / 2), 50), replace=False, p=backend.sum(prob, -1) / backend.sum(prob), # seed=local_state, dtype=backend.int64, ) index = [int(ind) for ind in index] gates_to_replace = np.array([gates_to_replace[ind] for ind in index]) prob = prob[index] replacement = np.array([replacement[ind] for ind in index]) replacement = [ replacement[ind][ int( backend.random_choice( range(len(p)), size=1, p=p / backend.sum(p), # seed=local_state )[0] ) ] for ind, p in enumerate(prob) ] replacement = {ind[0]: g for ind, g in zip(gates_to_replace, replacement)} sampled_circuit = circuit.__class__(**circuit.init_kwargs) for ind, gate in enumerate(circuit.queue): sampled_circuit.add(replacement.get(ind, gate)) return sampled_circuit
def _curve_fit( backend: Backend, model: callable, params: ArrayLike, xdata: ArrayLike, ydata: ArrayLike, lr: Union[float, int] = 1.0, max_iter: int = 100, tolerance_grad: float = 1e-5, ): """ Fits a model with given parameters on the data points (x,y). This is generally based on the `scipy.optimize.curve_fit` function, except for the `PyTorchBackend` which makes use of the `torch.optim.LBFGS` optimizer. Args: backend (:class:`qibo.backends.Backend`): simulation engine, this is only useful for `pytorch`. model (function): model to fit, it should be a callable ``model(x, *params)``. params (ArrayLike): initial parameters of the model. xdata (ArrayLike): x data, i.e. inputs to the model. ydata (ArrayLike): y data, i.e. targets ``y = model(x, *params)``. lr (float or int, optional): learning rate, defaults to :math:`1`. Used only in the `pytorch` case. max_iter (int, optional): maximum number of iterations, defaults to :math:`100`. Used only in the `pytorch` case. tolerance_grad (float, optional): gradient tolerance, optimization stops after reaching it, defaults to :math:`10^-5`. Used only in the `pytorch` case. Returns: ArrayLike: The optimal parameters. """ if backend.platform == "pytorch": # pragma: no cover # pytorch has some problems with the `scipy.optim.curve_fit` function # thus we use a `torch.optim` optimizer params.requires_grad = True loss = lambda pred, target: backend.mean((pred - target) ** 2) optimizer = backend.engine.optim.LBFGS( [params], lr=lr, max_iter=max_iter, tolerance_grad=tolerance_grad ) def closure(): optimizer.zero_grad() output = model(xdata, *params) loss_val = loss(output, ydata) loss_val.backward(retain_graph=True) return loss_val optimizer.step(closure) return params if backend.platform in ("cupy", "cuquantum"): # pragma: no cover # Currrently, ``cupy`` does not have compatibility with ``scipy.optimize``. xdata = backend.to_numpy(xdata) ydata = backend.to_numpy(ydata) params = backend.to_numpy(params) optimal_params = curve_fit(model, xdata, ydata, p0=params)[0] return backend.cast(optimal_params, dtype=optimal_params.dtype)
[docs]def CDR( circuit: "Circuit", # type: ignore observable: Union[Hamiltonian, SymbolicHamiltonian], noise_model: NoiseModel, nshots: int = 10000, model: callable = lambda x, a, b: a * x + b, n_training_samples: int = 100, full_output: bool = False, readout: Optional[dict] = None, qubit_map: Optional[List[int]] = None, seed=None, backend: Optional[Backend] = None, ): """Runs the Clifford Data Regression error mitigation method. Args: circuit (:class:`qibo.models.Circuit`): input circuit decomposed in the primitive gates ``X``, ``CNOT``, ``RX(pi/2)``, ``RZ(theta)``. observable (:class:`qibo.hamiltonians.Hamiltonian/:class:`qibo.hamiltonians.SymbolicHamiltonian`): observable to be measured. noise_model (:class:`qibo.noise.NoiseModel`): noise model used for simulating noisy computation. nshots (int, optional): Number of shots. Defaults to :math:`10000`. If `None`, the circuit is simulated without sampling: with statevector simulation if `noise_model` is not `None` and with density matrix simulation otherwise. Readout mitigation is not available with exact simulation. model (callable, optional): model used for fitting. This should be a callable function object ``f(x, *params)``, taking as input the predictor variable and the parameters. Default is a simple linear model ``f(x,a,b) := a*x + b``. n_training_samples (int, optional): number of training circuits to sample. Defaults to :math:`100`. full_output (bool, optional): if ``True``, this function returns additional information: ``val``, ``optimal_params``, ``train_val``. Defaults to ``False``. readout (dict, optional): a dictionary that may contain the following keys: * ncircuits: int, specifies the number of random circuits to use for the randomized method of readout error mitigation. * response_matrix: numpy.ndarray, used for applying a pre-computed response matrix for readout error mitigation. * ibu_iters: int, specifies the number of iterations for the iterative Bayesian unfolding method of readout error mitigation. If provided, the corresponding readout error mitigation method is used. Defaults to an empty ``dict``. qubit_map (list, optional): the qubit map. If None, a list of range of circuit's qubits is used. Defaults to ``None``. seed (int or :class:`numpy.random.Generator`, optional): Either a generator of random numbers or a fixed seed to initialize a generator. If ``None``, initializes a generator with a random seed. Default: ``None``. backend (:class:`qibo.backends.abstract.Backend`, optional): backend to be used in the execution. If ``None``, it uses the current backend. Defaults to ``None``. Returns: mit_val (float): Mitigated expectation value of `observable`. val (float): Noisy expectation value of `observable`. optimal_params (list): Optimal values for `params`. train_val (dict): Contains the noise-free and noisy expectation values obtained with the training circuits. Reference: 1. P. Czarnik, A. Arrasmith et al, *Error mitigation with Clifford quantum-circuit data*. `arXiv:2005.10189 [quant-ph] <https://arxiv.org/abs/2005.10189>`_. """ backend = _check_backend(backend) backend.set_seed(seed) if readout is None: readout = {} training_circuits = [ sample_training_circuit_cdr(circuit, backend=backend) for _ in range(n_training_samples) ] train_val = {"noise-free": [], "noisy": []} for circ in training_circuits: observable.backend = SIMULATION_BACKEND() val_noiseless = observable.expectation(circ, nshots) observable.backend = backend if nshots is None: circuit_result = _execute_circuit( circ, qubit_map, noise_model, nshots, backend=backend ) val_noisy = observable.expectation_from_state(circuit_result.state()) else: val_noisy = get_expectation_val_with_readout_mitigation( circ, observable, noise_model, nshots, readout, qubit_map, backend=backend, ) train_val["noise-free"].append(val_noiseless) train_val["noisy"].append(val_noisy) nparams = ( len(signature(model).parameters) - 1 ) # first arg is the input and the *params afterwards params = backend.random_sample(nparams) train_val_noisy = train_val["noisy"] train_val_noisy = backend.cast(train_val_noisy, dtype=type(train_val_noisy[0])) train_val_noiseless = train_val["noise-free"] train_val_noiseless = backend.cast( train_val_noiseless, dtype=train_val_noiseless[0].dtype ) optimal_params = _curve_fit( backend, model, params, train_val_noisy, train_val_noiseless, ) if nshots is None: circuit_result = _execute_circuit( circuit, qubit_map, noise_model, nshots, backend=backend ) val = observable.expectation_from_state(circuit_result.state()) else: val = get_expectation_val_with_readout_mitigation( circuit, observable, noise_model, nshots, readout, qubit_map, backend=backend, ) mit_val = model(val, *optimal_params) if full_output: return mit_val, val, optimal_params, train_val return mit_val
[docs]def vnCDR( circuit: "Circuit", # type: ignore observable: Union[Hamiltonian, SymbolicHamiltonian], noise_levels: ArrayLike, noise_model: NoiseModel, nshots: int = 10000, model: callable = None, n_training_samples: int = 100, insertion_gate: str = "CNOT", full_output: bool = False, readout: Optional[dict] = None, qubit_map: Optional[List[int]] = None, seed: Optional[int] = None, backend: Optional[Backend] = None, ): """Runs the variable-noise Clifford Data Regression error mitigation method. Args: circuit (:class:`qibo.models.Circuit`): input circuit decomposed in the primitive gates ``X``, ``CNOT``, ``RX(pi/2)``, ``RZ(theta)``. observable (:class:`qibo.hamiltonians.Hamiltonian` or :class:`qibo.hamiltonians.SymbolicHamiltonian`): observable to be measured. noise_levels (ArrayLike): sequence of noise levels. noise_model (:class:`qibo.noise.NoiseModel`): noise model used for simulating noisy computation. nshots (int, optional): Number of shots. Defaults to :math:`10000`. If `None`, the circuit is simulated without sampling: with statevector simulation if `noise_model` is not `None` and with density matrix simulation otherwise. Readout mitigation is not available with exact simulation. model (callable, optional): model used for fitting. This should be a callable function object ``f(x, *params)``, taking as input the predictor variable and the parameters. Default is a simple linear model ``f(x,a,b) := a*x + b``. n_training_samples (int, optional): number of training circuits to sample. insertion_gate (str, optional): gate to be used in the insertion. If ``"RX"``, the gate used is :math:``RX(\\pi / 2)``. Default is ``"CNOT"``. full_output (bool, optional): if ``True``, this function returns additional information: ``val``, ``optimal_params``, ``train_val``. Defaults to ``False``. readout (dict, optional): a dictionary that may contain the following keys: * ncircuits: int, specifies the number of random circuits to use for the randomized method of readout error mitigation. * response_matrix: numpy.ndarray, used for applying a pre-computed response matrix for readout error mitigation. * ibu_iters: int, specifies the number of iterations for the iterative Bayesian unfolding method of readout error mitigation. If provided, the corresponding readout error mitigation method is used. Defaults to empty ``dict``. qubit_map (list, optional): the qubit map. If ``None``, a list of range of circuit's qubits is used. Defaults to ``None``. seed (int or :class:`numpy.random.Generator`, optional): Either a generator of random numbers or a fixed seed to initialize a generator. If ``None``, initializes a generator with a random seed. Default: ``None``. backend (:class:`qibo.backends.abstract.Backend`, optional): backend to be used in the execution. If ``None``, it uses the current backend. Defaults to ``None``. Returns: mit_val (float): Mitigated expectation value of `observable`. val (list): Expectation value of `observable` with increased noise levels. optimal_params (list): Optimal values for `params`. train_val (dict): Contains the noise-free and noisy expectation values obtained with the training circuits. Reference: 1. A. Lowe, MH. Gordon et al, *Unified approach to data-driven quantum error mitigation*. `arXiv:2011.01157 [quant-ph] <https://arxiv.org/abs/2011.01157>`_. """ backend = _check_backend(backend) backend.set_seed(seed) if model is None: if backend.platform in ("cupy", "cuquantum"): model = lambda x, *params: np.sum(x * np.vstack(params), axis=0) else: model = lambda x, *params: backend.sum(x * backend.vstack(params), axis=0) if readout is None: readout = {} training_circuits = [ sample_training_circuit_cdr(circuit, seed=seed, backend=backend) for _ in range(n_training_samples) ] train_val = {"noise-free": [], "noisy": []} for circ in training_circuits: observable.backend = SIMULATION_BACKEND() val = observable.expectation(circ, nshots) observable.backend = backend train_val["noise-free"].append(float(val.real)) for level in noise_levels: noisy_c = get_noisy_circuit(circ, level, insertion_gate=insertion_gate) if nshots is None: circuit_result = _execute_circuit( noisy_c, qubit_map, noise_model, nshots, backend=backend ) val = float(observable.expectation_from_state(circuit_result.state())) else: val = get_expectation_val_with_readout_mitigation( noisy_c, observable, noise_model, nshots, readout, qubit_map, backend=backend, ) train_val["noisy"].append(float(val.real)) train_val_noisy = train_val["noisy"] noisy_array = backend.cast(train_val_noisy, dtype=type(train_val_noisy[0])) noisy_array = backend.reshape(noisy_array, (-1, len(noise_levels))) # params = local_state.random(len(noise_levels)) params = backend.random_sample(len(noise_levels)) train_val_noiseless = train_val["noise-free"] train_val_noiseless = backend.cast( train_val_noiseless, dtype=type(train_val_noiseless[0]) ) optimal_params = _curve_fit( backend, model, params, noisy_array.T, train_val_noiseless, lr=1, tolerance_grad=1e-7, ) val = [] for level in noise_levels: noisy_c = get_noisy_circuit(circuit, level, insertion_gate=insertion_gate) if nshots is None: circuit_result = _execute_circuit( noisy_c, qubit_map, noise_model, nshots, backend=backend ) expval = float(observable.expectation_from_state(circuit_result.state())) else: expval = get_expectation_val_with_readout_mitigation( noisy_c, observable, noise_model, nshots, readout, qubit_map, backend=backend, ) val.append(expval) mit_val = model( backend.cast(val, dtype=type(val[0])).reshape(-1, 1), *optimal_params, )[0] if full_output: return mit_val, val, optimal_params, train_val return mit_val
[docs]def iterative_bayesian_unfolding( probabilities: ArrayLike, response_matrix: ArrayLike, iterations: int = 10 ): """ Iterative Bayesian Unfolding (IBU) method for readout mitigation. Args: probabilities (ArrayLike): the input probabilities to be unfolded. response_matrix (ArrayLike): the response matrix. iterations (int, optional): the number of iterations to perform. Defaults to 10. Returns: ArrayLike: the unfolded probabilities. Reference: 1. B. Nachman, M. Urbanek et al, *Unfolding Quantum Computer Readout Noise*. `arXiv:1910.01969 [quant-ph] <https://arxiv.org/abs/1910.01969>`_. 2. S. Srinivasan, B. Pokharel et al, *Scalable Measurement Error Mitigation via Iterative Bayesian Unfolding*. `arXiv:2210.12284 [quant-ph] <https://arxiv.org/abs/2210.12284>`_. """ unfolded_probabilities = np.ones((len(probabilities), 1)) / len(probabilities) for _ in range(iterations): unfolded_probabilities = unfolded_probabilities * ( np.transpose(response_matrix)
[docs] @ (probabilities / (response_matrix @ unfolded_probabilities)) ) return unfolded_probabilities
def get_response_matrix( nqubits: int, qubit_map: Optional[List[int]] = None, noise_model: Optional[NoiseModel] = None, nshots: int = 10000, backend: Optional[Backend] = None, ): """Computes the response matrix for readout mitigation. Args: nqubits (int): Total number of qubits. qubit_map (list, optional): the qubit map. If None, a list of range of circuit's qubits is used. Defaults to ``None``. noise_model (:class:`qibo.noise.NoiseModel`, optional): noise model used for simulating noisy computation. This matrix can be used to mitigate the effect of `qibo.noise.ReadoutError`. nshots (int, optional): number of shots. Defaults to :math:`10000`. backend (:class:`qibo.backends.abstract.Backend`, optional): backend to be used in the execution. If ``None``, it uses the current backend. Defaults to ``None``. Returns: ArrayLike: the computed (`nqubits`, `nqubits`) response matrix for readout mitigation. """ from qibo import Circuit # pylint: disable=import-outside-toplevel backend = _check_backend(backend) response_matrix = np.zeros((2**nqubits, 2**nqubits)) for elem in range(2**nqubits): binary_state = format(elem, f"0{nqubits}b") circuit = Circuit(nqubits, density_matrix=True) for qubit, bit in enumerate(binary_state): if bit == "1": circuit.add(gates.X(qubit)) circuit.add(gates.M(*range(nqubits))) circuit_result = _execute_circuit( circuit, qubit_map, noise_model, nshots, backend=backend ) frequencies = circuit_result.frequencies() column = np.zeros(2**nqubits) for key, value in frequencies.items(): column[int(key, 2)] = value / nshots response_matrix[:, elem] = column return response_matrix
[docs]def apply_resp_mat_readout_mitigation( state: CircuitResult, response_matrix: ArrayLike, iterations: Optional[int] = None ): """ Applies readout error mitigation to the given state using the provided response matrix. Args: state (:class:`qibo.measurements.CircuitResult`): the input state to be updated. This state should contain the frequencies that need to be mitigated. response_matrix (numpy.ndarray): the response matrix for readout mitigation. iterations (int, optional): the number of iterations to use for the Iterative Bayesian Unfolding method. If ``None`` the 'inverse' method is used. Defaults to ``None``. Returns: :class:`qibo.measurements.CircuitResult`: the input state with the updated (mitigated) frequencies. """ frequencies = np.zeros(2 ** len(state.measurements[0].qubits)) for key, value in state.frequencies().items(): frequencies[int(key, 2)] = value frequencies = frequencies.reshape(-1, 1) if iterations is None: calibration_matrix = np.linalg.inv(response_matrix) mitigated_frequencies = calibration_matrix @ frequencies else: mitigated_probabilities = iterative_bayesian_unfolding( frequencies / np.sum(frequencies), response_matrix, iterations ) mitigated_frequencies = np.round( mitigated_probabilities * np.sum(frequencies), 0 ) mitigated_frequencies = ( mitigated_frequencies / np.sum(mitigated_frequencies) ) * np.sum(frequencies) for i, value in enumerate(mitigated_frequencies): state._frequencies[i] = float(value[0].real) # pylint: disable=W0212 return state
[docs]def apply_randomized_readout_mitigation( circuit, noise_model=None, nshots: int = 10000, ncircuits: int = 10, qubit_map=None, seed=None, backend=None, ): """Readout mitigation method that transforms the bias in an expectation value into a measurable multiplicative factor. This factor can be eliminated at the expense of increased sampling complexity for the observable. Args: circuit (:class:`qibo.models.Circuit`): input circuit. noise_model(:class:`qibo.noise.NoiseModel`, optional): noise model used for simulating noisy computation. Defaults to ``None``. nshots (int, optional): number of shots. Defaults to :math:`10000`. ncircuits (int, optional): number of randomized circuits. Each of them uses ``int(nshots / ncircuits)`` shots. Defaults to 10. qubit_map (list, optional): the qubit map. If ``None``, a list of range of circuit's qubits is used. Defaults to ``None``. seed (int or :class:`numpy.random.Generator`, optional): Either a generator of random numbers or a fixed seed to initialize a generator. If ``None``, initializes a generator with a random seed. Default: ``None``. backend (:class:`qibo.backends.abstract.Backend`, optional): backend to be used in the execution. If ``None``, it uses the current backend. Defaults to ``None``. Returns: :class:`qibo.measurements.CircuitResult`: the state of the input circuit with mitigated frequencies. References: 1. Ewout van den Berg, Zlatko K. Minev et al, *Model-free readout-error mitigation for quantum expectation values*. `arXiv:2012.09738 [quant-ph] <https://arxiv.org/abs/2012.09738>`_. """ from qibo import Circuit # pylint: disable=import-outside-toplevel from qibo.quantum_info import ( # pylint: disable=import-outside-toplevel random_pauli, ) backend = _check_backend(backend) backend.set_seed(seed) meas_qubits = circuit.measurements[0].qubits nshots_r = int(nshots / ncircuits) freq = np.zeros((ncircuits, 2), object) for k in range(ncircuits): circuit_c = circuit.copy(True) circuit_c.queue.pop() cal_circuit = Circuit(circuit.nqubits, density_matrix=True) x_gate = random_pauli( circuit.nqubits, 1, subset=["I", "X"], backend=backend ).queue error_map = {} for j, gate in enumerate(x_gate): if isinstance(gate, gates.X) and gate.qubits[0] in meas_qubits: error_map[gate.qubits[0]] = 1 circuits = [circuit_c, cal_circuit] results = [] freqs = [] for circ in circuits: circ.add(x_gate) circ.add(gates.M(*meas_qubits)) result = _execute_circuit( circ, qubit_map, noise_model, nshots_r, backend=backend ) result._samples = result.apply_bitflips(error_map) # pylint: disable=W0212 results.append(result) freqs.append(result.frequencies(binary=False)) freq[k, :] = freqs for j in range(2): results[j].nshots = nshots freq_sum = freq[0, j] for frs in freq[1::, j]: freq_sum += frs results[j]._frequencies = freq_sum # pylint: disable=W0212 return results
[docs]def get_expectation_val_with_readout_mitigation( circuit: "Circuit", # type: ignore observable: Union[Hamiltonian, SymbolicHamiltonian], noise_model: Optional[NoiseModel] = None, nshots: int = 10000, readout: Optional[dict] = None, qubit_map: Optional[List[int]] = None, seed: Optional[int] = None, backend: Optional[Backend] = None, ): """CDR: Applies readout error mitigation to the given circuit and observable. Args: circuit (:class:`qibo.models.Circuit`): input circuit. observable (:class:`qibo.hamiltonians.Hamiltonian/:class:`qibo.hamiltonians.SymbolicHamiltonian`): The observable to be measured. noise_model (:class:`qibo.models.noise.NoiseModel`, optional): the noise model to be applied. Defaults to ``None``. nshots (int, optional): the number of shots for the circuit execution. Defaults to :math:`10^{4}`. readout (dict, optional): a dictionary that may contain the following keys: * ncircuits: int, specifies the number of random circuits to use for the randomized method of readout error mitigation. * response_matrix: numpy.ndarray, used for applying a pre-computed response matrix for readout error mitigation. * ibu_iters: int, specifies the number of iterations for the iterative Bayesian unfolding method of readout error mitigation. If provided, the corresponding readout error mitigation method is used. Defaults to an empty ``dict``. qubit_map (list, optional): the qubit map. If None, a list of range of circuit's qubits is used. Defaults to ``None``. seed (int or :class:`numpy.random.Generator`, optional): Either a generator of random numbers or a fixed seed to initialize a generator. If ``None``, initializes a generator with a random seed. Default: ``None``. backend (qibo.backends.abstract.Backend, optional): the backend to be used in the execution. If ``None``, it uses the global backend. Defaults to ``None``. Returns: float: The mitigated expectation value of the observable. """ backend = _check_backend(backend) backend.set_seed(seed) if readout is None: # pragma: no cover readout = {} if len(circuit.measurements) == 0: circuit = circuit.copy() circuit._final_state = None # pylint: disable=W0212 qubits = [ factor.target_qubit for term in observable.terms for factor in term.factors if factor.__class__.__name__ != "I" ] circuit.add(gates.M(*qubits)) if "ncircuits" in readout: circuit_result, circuit_result_cal = ( # pylint: disable=W0632 apply_randomized_readout_mitigation( circuit, noise_model, nshots, readout["ncircuits"], backend=backend, ) ) else: circuit_result = _execute_circuit( circuit, qubit_map, noise_model, nshots, backend=backend ) if "response_matrix" in readout: circuit_result = apply_resp_mat_readout_mitigation( circuit_result, readout["response_matrix"], readout.get("ibu_iters", None), ) exp_val = circuit_result.expectation_from_samples(observable) if "ncircuits" in readout: return float(exp_val / circuit_result_cal.expectation_from_samples(observable)) return float(exp_val)
[docs]def sample_clifford_training_circuit( circuit: "Circuit", # type: ignore seed: Optional[int] = None, backend: Optional[Backend] = None, ): """Samples a training circuit for CDR by susbtituting all the non-Clifford gates. Args: circuit (:class:`qibo.models.Circuit`): circuit to sample from. seed (int or :class:`numpy.random.Generator`, optional): Either a generator of random numbers or a fixed seed to initialize a generator. If ``None``, initializes a generator with a random seed. Default: ``None``. backend (:class:`qibo.backends.abstract.Backend`, optional): backend to be used in the execution. If ``None``, it uses the current backend. Defaults to ``None``. Returns: :class:`qibo.models.Circuit`: the sampled circuit. """ from qibo.quantum_info import ( # pylint: disable=import-outside-toplevel random_clifford, ) backend = _check_backend(backend) backend.set_seed(seed) non_clifford_gates_indices = [ i for i, gate in enumerate(circuit.queue) if not gate.clifford and not isinstance(gate, gates.M) ] if not non_clifford_gates_indices: raise_error(ValueError, "No non-Clifford gate found, no circuit sampled.") sampled_circuit = circuit.__class__(**circuit.init_kwargs) for i, gate in enumerate(circuit.queue): if i in non_clifford_gates_indices: clifford_matrix = random_clifford( len(gate.qubits), return_circuit=True, backend=backend, ) clifford_matrix = clifford_matrix.unitary(backend) gate = gates.Unitary( clifford_matrix, *gate.qubits, ) gate.clifford = True sampled_circuit.add(gate) return sampled_circuit
def error_sensitive_circuit( circuit: "Circuit", # type: ignore observable: Union[Hamiltonian, SymbolicHamiltonian], seed: Optional[int] = None, backend: Optional[Backend] = None, ): """ Generates a Clifford circuit that preserves the same circuit frame as the input circuit, and stabilizes the specified Pauli observable. Args: circuit (:class:`qibo.models.Circuit`): input circuit. observable (:class:`qibo.hamiltonians.Hamiltonian` or :class:`qibo.hamiltonians.SymbolicHamiltonian`): Pauli observable to be measured. seed (int or :class:`numpy.random.Generator`, optional): Either a generator of random numbers or a fixed seed to initialize a generator. If ``None``, initializes a generator with a random seed. Default: ``None``. backend (:class:`qibo.backends.CliffordBackend`, optional): Clifford backend to be used in the execution. if ``None``, it uses the `CliffordBackend` with the platform of the current backend as engine. Defaults to ``None``. Returns: :class:`qibo.models.Circuit`: the error sensitive circuit. :class:`qibo.models.Circuit`: the sampled Clifford circuit. list: the list of adjustment gates. Reference: 1. Dayue Qin, Yanzhu Chen et al, *Error statistics and scalability of quantum error mitigation formulas*. `npj Quantum Inf 9, 35 (2023) <https://doi.org/10.1038/s41534-023-00707-7>`_. """ backend_temp = _check_backend(backend) backend = ( CLIFFORD_BACKEND() if backend is None else backend_temp ) # pragma: no cover backend.set_seed(seed) sampled_circuit = sample_clifford_training_circuit(circuit, backend=backend) result = backend.execute_circuit(sampled_circuit.invert(), nshots=1) symplectic_matrix = result.symplectic_matrix[:-1, :-1] terms = observable.terms[0].factors pauli_symplectic = backend.zeros((2 * circuit.nqubits, 1), dtype=backend.uint8) for term in terms: term = str(term) index = int(term[1]) if term[0] in ["X", "Y"]: # pragma: no cover pauli_symplectic[index] = 1 if term[0] in ["Z", "Y"]: # pragma: no cover pauli_symplectic[index + circuit.nqubits] = 1 # U --> symplectic_matrix and O --> observable_sym, # U*O*Ut --> symplectic_matrix.T @ observable_sym # To get Ut*O*U we need the symplectic_matrix of the inverse circuit Ut pauli_symplectic = (symplectic_matrix.T @ pauli_symplectic) % 2 x_component = pauli_symplectic[: circuit.nqubits] == 1 z_component = pauli_symplectic[circuit.nqubits :] == 1 y_obs = reduce( mul, [ Y(qubit, backend=backend) for qubit in (x_component * z_component).nonzero()[0] ], 1, ) x_obs = reduce( mul, [ X(qubit, backend=backend) for qubit in (x_component * ~z_component).nonzero()[0] ], 1, ) z_obs = reduce( mul, [ Z(qubit, backend=backend) for qubit in (z_component * ~x_component).nonzero()[0] ], 1, ) observable = y_obs * x_obs * z_obs observable = SymbolicHamiltonian( observable, nqubits=circuit.nqubits, backend=backend ) terms = observable.terms[0].factors adjustment_gates = [] for term in terms: term = str(term) qubit = int(term[1]) if term[0] in ["X", "Y"]: adjustment_gate = gates.M(qubit, basis=term[0]).basis[0] else: adjustment_gate = gates.I(qubit) adjustment_gates.append(adjustment_gate) sensitive_circuit = sampled_circuit.__class__(**sampled_circuit.init_kwargs) sensitive_circuit.add(adjustment_gates) for gate in sampled_circuit.queue: sensitive_circuit.add(gate) return sensitive_circuit, sampled_circuit, adjustment_gates
[docs]def ICS( circuit: "Circuit", # type: ignore observable: Union[Hamiltonian, SymbolicHamiltonian], readout: Optional[dict] = None, qubit_map: Optional[List[int]] = None, noise_model: Optional[NoiseModel] = None, nshots: int = int(1e4), n_training_samples: int = 10, full_output: bool = False, seed: Optional[int] = None, backend: Optional[Backend] = None, ): """ Computes the Important Clifford Sampling method. Args: circuit (:class:`qibo.models.Circuit`): input circuit. observable (:class:`qibo.hamiltonians.Hamiltonian` or :class:`qibo.hamiltonians.SymbolicHamiltonian`): the observable to be measured. readout (dict, optional): a dictionary that may contain the following keys: * ncircuits: int, specifies the number of random circuits to use for the randomized method of readout error mitigation. * response_matrix: numpy.ndarray, used for applying a pre-computed response matrix for readout error mitigation. * ibu_iters: int, specifies the number of iterations for the iterative Bayesian unfolding method of readout error mitigation. If provided, the corresponding readout error mitigation method is used. Defaults to an empty ``dict``. qubit_map (list, optional): the qubit map. If ``None``, a list of range of circuit's qubits is used. Defaults to ``None``. noise_model (:class:`qibo.models.noise.NoiseModel`, optional): the noise model to be applied. Defaults to ``None``. nshots (int, optional): Number of shots. Defaults to :math:`10000`. If `None`, the circuit is simulated without sampling: with statevector simulation if `noise_model` is not `None` and with density matrix simulation otherwise. Readout mitigation is not available with exact simulation. n_training_samples (int, optional): the number of training samples. Defaults to 10. full_output (bool, optional): if ``True``, this function returns additional information: ``val``, ``optimal_params``, ``train_val``. Defaults to ``False``. seed (int or :class:`numpy.random.Generator`, optional): Either a generator of random numbers or a fixed seed to initialize a generator. If ``None``, initializes a generator with a random seed. Default: ``None``. backend (qibo.backends.abstract.Backend, optional): the backend to be used in the execution. If ``None``, it uses the global backend. Defaults to ``None``. Returns: mitigated_expectation (float): the mitigated expectated value. mitigated_expectation_std (float): the standard deviation of the mitigated expectated value. dep_param (float): the depolarizing parameter. dep_param_std (float): the standard deviation of the depolarizing parameter. lambda_list (list): the list of the depolarizing parameters. data (dict): the data dictionary containing the noise-free and noisy expectation values obtained with the training circuits. Reference: 1. Dayue Qin, Yanzhu Chen et al, *Error statistics and scalability of quantum error mitigation formulas*, `npj Quantum Inf 9, 35 (2023) <https://doi.org/10.1038/s41534-023-00707-7>`_. """ backend = _check_backend(backend) backend.set_seed(seed) platform = backend.name if backend.platform is None else backend.platform clifford_backend = CLIFFORD_BACKEND(platform) if readout is None: readout = {} if qubit_map is None: qubit_map = list(range(circuit.nqubits)) training_circuits = [ error_sensitive_circuit(circuit, observable, backend=clifford_backend)[0] for _ in range(n_training_samples) ] data = {"noise-free": [], "noisy": []} lambda_list = [] for training_circuit in training_circuits: training_circuit_copy = training_circuit.copy(deep=True) observable.backend = clifford_backend expectation = observable.expectation(training_circuit_copy, nshots) observable.backend = backend if nshots is None: circuit_result = _execute_circuit( training_circuit, qubit_map, noise_model, nshots, backend=backend ) noisy_expectation = observable.expectation_from_state( circuit_result.state() ) else: noisy_expectation = get_expectation_val_with_readout_mitigation( training_circuit, observable, noise_model, nshots, readout, qubit_map, backend=backend, ) data["noise-free"].append(expectation) data["noisy"].append(noisy_expectation) lambda_list.append(1 - noisy_expectation / expectation) lambda_list = backend.cast(lambda_list, dtype=lambda_list[0].dtype) dep_param = backend.mean(lambda_list) dep_param_std = backend.std(lambda_list) if nshots is None: circuit_result = _execute_circuit( circuit, qubit_map, noise_model, nshots, backend=backend ) noisy_expectation = float(observable.expectation(circuit_result.state())) else: noisy_expectation = get_expectation_val_with_readout_mitigation( circuit, observable, noise_model, nshots, readout, qubit_map, backend=backend, ) one_dep_squared = (1 - dep_param) ** 2 dep_std_squared = dep_param_std**2 mitigated_expectation = ( (1 - dep_param) * noisy_expectation / (one_dep_squared + dep_std_squared) ) mitigated_expectation_std = ( dep_param_std * abs(noisy_expectation) * abs((1 - dep_param) ** 2 - dep_std_squared) / (one_dep_squared + dep_std_squared) ** 2 ) if full_output: return ( mitigated_expectation, mitigated_expectation_std, dep_param, dep_param_std, lambda_list, data, ) return mitigated_expectation
def _execute_circuit(circuit, qubit_map, noise_model=None, nshots=10000, backend=None): """ Helper function to execute the given circuit with the specified parameters. Args: circuit (:class:`qibo.models.Circuit`): input circuit. qubit_map (list): the qubit map. If ``None``, a list of range of circuit's qubits is used. Defaults to ``None``. noise_model (:class:`qibo.models.noise.NoiseModel`, optional): The noise model to be applied. Defaults to ``None``. nshots (int): the number of shots for the circuit execution. Defaults to :math:`10^{4}`. backend (qibo.backends.abstract.Backend, optional): the backend to be used in the execution. If ``None``, it uses the global backend. Defaults to ``None``. Returns: :class:`qibo.result.CircuitResult`: The result of the circuit execution. """ if backend is None: # pragma: no cover backend = get_backend() elif backend.name == "qibolab": # pragma: no cover circuit.wire_names = qubit_map elif noise_model is not None: circuit = noise_model.apply(circuit) if nshots is None: circuit.density_matrix = True circuit_result = backend.execute_circuit(circuit, nshots=nshots) return circuit_result