# Copyright (C) Unitary Fund
#
# This source code is licensed under the GPL license (v3) found in the
# LICENSE file in the root directory of this source tree.
from dataclasses import asdict, dataclass
from enum import Enum, auto
from functools import partial
from typing import Any, Callable, Dict, List, cast
import cirq
import networkx as nx
import numpy as np
from mitiq import QPROGRAM, Executor
from mitiq.benchmarks import (
generate_ghz_circuit,
generate_mirror_circuit,
generate_rb_circuits,
generate_rotated_rb_circuits,
generate_w_circuit,
)
from mitiq.interface import convert_from_mitiq
from mitiq.pec import execute_with_pec
from mitiq.pec.representations import (
represent_operation_with_local_biased_noise,
represent_operation_with_local_depolarizing_noise,
)
from mitiq.raw import execute
from mitiq.zne import execute_with_zne
from mitiq.zne.inference import LinearFactory, RichardsonFactory
from mitiq.zne.scaling import fold_gates_at_random, fold_global
[docs]
class MitigationTechnique(Enum):
"""Simple enum type for handling validation, and providing helper functions
when accessing mitigation techniques."""
ZNE = auto()
PEC = auto()
RAW = auto()
@property
def mitigation_function(self) -> Callable[..., float]:
if self is MitigationTechnique.ZNE:
return execute_with_zne
elif self is MitigationTechnique.PEC:
return cast(Callable[..., float], execute_with_pec)
elif self is MitigationTechnique.RAW:
return execute
calibration_supported_techniques = {
"ZNE": MitigationTechnique.ZNE,
"PEC": MitigationTechnique.PEC,
}
[docs]
@dataclass
class BenchmarkProblem:
"""A dataclass containing information for instances of problems that will
be run during the calibrations process.
Args:
id: A unique numerical id.
circuit: The circuit to be run.
type: The type of the circuit (often the name of the algorithm)
ideal_distribution: The ideal probability distribution after applying
``circuit``.
"""
id: int
circuit: cirq.Circuit
type: str
ideal_distribution: Dict[str, float]
def most_likely_bitstring(self) -> str:
distribution = self.ideal_distribution
return max(distribution, key=distribution.__getitem__)
def largest_probability(self) -> float:
return max(self.ideal_distribution.values())
[docs]
def converted_circuit(self, circuit_type: str) -> QPROGRAM:
"""Adds measurements to all qubits and convert
to the input frontend type.
Args:
circuit_type: The circuit type as a string.
For supported circuit types see mitiq.SUPPORTED_PROGRAM_TYPES.
Returns:
The converted circuit with final measurements.
"""
circuit = self.circuit.copy()
circuit.append(cirq.measure(circuit.all_qubits()))
return convert_from_mitiq(circuit, circuit_type)
@property
def num_qubits(self) -> int:
return len(self.circuit.all_qubits())
@property
def circuit_depth(self) -> int:
return len(self.circuit)
@property
def two_qubit_gate_count(self) -> int:
return sum(len(op.qubits) > 1 for op in self.circuit.all_operations())
[docs]
def to_dict(self) -> Dict[str, Any]:
"""Produces a summary of the ``BenchmarkProblem``, to be used in
recording the results when running calibration experiments.
Returns:
Dictionary summarizing important attributes of the problem's
circuit.
"""
base = asdict(self)
# remove circuit; it can be regenerated if needed
del base["circuit"]
del base["id"]
base["num_qubits"] = self.num_qubits
base["circuit_depth"] = self.circuit_depth
base["two_qubit_gate_count"] = self.two_qubit_gate_count
return base
def __repr__(self) -> str:
return str(self.to_dict())
def __str__(self) -> str:
result = ""
for key, value in self.to_dict().items():
if key == "ideal_distribution":
continue
title: str = key.replace("_", " ").capitalize()
result += f"{title}: {value}\n"
return result.rstrip()
[docs]
@dataclass
class Strategy:
"""A dataclass which describes precisely an error mitigation approach by
specifying a technique and the associated options.
Args:
id: A unique numerical id.
technique: One of Mitiq's support error mitigation strategies,
specified as a :class:`MitigationTechnique`.
technique_params: A dictionary of options to pass to the mitigation
method specified in `technique`.
"""
id: int
technique: MitigationTechnique
technique_params: Dict[str, Any]
@property
def mitigation_function(self) -> Callable[..., float]:
if self.technique is MitigationTechnique.PEC:
self.technique_params.setdefault("noise_bias", 0)
def partial_pec(circuit: cirq.Circuit, execute: Executor) -> float:
rep_function = self.technique_params["representation_function"]
operations = []
for op in circuit.all_operations():
if len(op.qubits) >= 2 and op not in operations:
operations.append(cirq.Circuit(op))
num_samples = self.technique_params["num_samples"]
if (
self.technique_params["representation_function"]
== represent_operation_with_local_biased_noise
):
reps = [
rep_function(
op,
self.technique_params["noise_level"],
self.technique_params["noise_bias"],
)
for op in operations
]
else:
reps = [
rep_function(
op,
self.technique_params["noise_level"],
)
for op in operations
]
return self.technique.mitigation_function(
circuit,
execute,
representations=reps,
num_samples=num_samples,
)
return partial_pec
elif self.technique is MitigationTechnique.ZNE:
return partial(
self.technique.mitigation_function, **self.technique_params
)
else:
raise ValueError(
"""Specified technique is not supported by calibration.
See {} for supported techniques.""",
calibration_supported_techniques,
)
[docs]
def to_dict(self) -> Dict[str, Any]:
"""A summary of the strategies parameters, without the technique added.
Returns:
A dictionary describing the strategies parameters."""
summary = {"technique": self.technique.name}
if self.technique is MitigationTechnique.ZNE:
inference_func = self.technique_params["factory"]
summary["factory"] = inference_func.__class__.__name__
summary["scale_factors"] = inference_func._scale_factors
summary["scale_method"] = self.technique_params[
"scale_noise"
].__name__
elif self.technique is MitigationTechnique.PEC:
summary["representation_function"] = self.technique_params[
"representation_function"
].__name__
summary["noise_level"] = self.technique_params["noise_level"]
summary["noise_bias"] = self.technique_params.setdefault(
"noise_bias", 0
)
summary["is_qubit_dependent"] = self.technique_params[
"is_qubit_dependent"
]
summary["num_samples"] = self.technique_params["num_samples"]
return summary
def to_pretty_dict(self) -> Dict[str, str]:
summary = self.to_dict()
if self.technique is MitigationTechnique.ZNE:
summary["scale_factors"] = str(summary["scale_factors"])[1:-1]
summary["factory"] = summary["factory"][:-7]
elif self.technique is MitigationTechnique.PEC:
summary["noise_bias"] = summary.get("noise_bias", "N/A")
summary["representation_function"] = summary[
"representation_function"
][25:]
return summary
def __repr__(self) -> str:
return str(self.to_dict())
def __str__(self) -> str:
result = ""
for key, value in self.to_pretty_dict().items():
title: str = key.replace("_", " ").capitalize()
result += f"{title}: {value}\n"
return result.rstrip()
def num_circuits_required(self) -> int:
summary = self.to_dict()
if self.technique is MitigationTechnique.ZNE:
return len(summary["scale_factors"])
elif self.technique is MitigationTechnique.PEC:
return summary["num_samples"]
elif self.technique is MitigationTechnique.RAW:
return 1
return None
[docs]
class Settings:
"""A class to store the configuration settings of a :class:`.Calibrator`.
Args:
benchmarks: A list where each element is a dictionary of parameters for
generating circuits to be used in calibration experiments. The
dictionary keys include ``circuit_type``, ``num_qubits``,
``circuit_depth``, and in the case of mirror circuits, a random
seed ``circuit_seed``. An example of input to ``benchmarks`` is::
[
{
"circuit_type": "rb",
"num_qubits": 2,
"circuit_depth": 7,
},
{
"circuit_type": "mirror",
"num_qubits": 2,
"circuit_depth": 7,
"circuit_seed": 1,
}
]
strategies: A specification of the methods/parameters to be used in
calibration experiments.
"""
def __init__(
self,
benchmarks: List[Dict[str, Any]],
strategies: List[Dict[str, Any]],
):
self.techniques = [
MitigationTechnique[technique["technique"].upper()]
for technique in strategies
]
self.technique_params = strategies
self.benchmarks = benchmarks
self.strategy_dict: Dict[int, Strategy] = {}
self.problem_dict: Dict[int, BenchmarkProblem] = {}
def get_strategy(self, strategy_id: int) -> Strategy:
return self.strategy_dict[strategy_id]
def get_problem(self, problem_id: int) -> BenchmarkProblem:
return self.problem_dict[problem_id]
[docs]
def make_problems(self) -> List[BenchmarkProblem]:
"""Generate the benchmark problems for the calibration experiment.
Returns:
A list of :class:`BenchmarkProblem` objects"""
circuits = []
for i, benchmark in enumerate(self.benchmarks):
circuit_type = benchmark["circuit_type"]
num_qubits = benchmark["num_qubits"]
# Set default to return correct type
depth = benchmark.get("circuit_depth", -1)
if circuit_type == "ghz":
circuit = generate_ghz_circuit(num_qubits)
ideal = {"0" * num_qubits: 0.5, "1" * num_qubits: 0.5}
elif circuit_type == "w":
circuit = generate_w_circuit(num_qubits)
ideal = {}
for i in range(num_qubits):
bitstring = "0" * i + "1" + "0" * (num_qubits - i - 1)
ideal[bitstring] = 1 / num_qubits
elif circuit_type == "rb":
circuit = generate_rb_circuits(num_qubits, depth)[0]
ideal = {"0" * num_qubits: 1.0}
elif circuit_type == "rotated_rb":
theta = benchmark["theta"]
if num_qubits == 1:
circuit = generate_rotated_rb_circuits(num_qubits, depth)[
0
]
p = (2 / 3) * np.sin(theta / 2) ** 2
ideal = {"0": p, "1": 1 - p}
else:
raise NotImplementedError(
"""rotated rb circuits with >1 qubits
not yet supported in calibration"""
)
elif circuit_type == "mirror":
seed = benchmark.get("circuit_seed", None)
circuit, bitstring_list = generate_mirror_circuit(
nlayers=depth,
two_qubit_gate_prob=1.0,
connectivity_graph=nx.complete_graph(num_qubits),
seed=seed,
)
ideal_bitstring = "".join(map(str, bitstring_list))
ideal = {ideal_bitstring: 1.0}
elif circuit_type == "qv":
raise NotImplementedError(
"quantum volume circuits not yet supported in calibration"
)
else:
raise ValueError(
"invalid value passed for `circuit_types`. Must be "
"one of `ghz`, `rb`, `mirror`, `w`, or `qv`, "
f"but got {circuit_type}."
)
circuit = cast(cirq.Circuit, circuit)
problem = BenchmarkProblem(
id=i,
circuit=circuit,
type=circuit_type,
ideal_distribution=ideal,
)
circuits.append(problem)
self.problem_dict[problem.id] = problem
return circuits
[docs]
def make_strategies(self) -> List[Strategy]:
"""Generates a list of :class:`Strategy` objects using the specified
configurations.
Returns:
A list of :class:`Strategy` objects."""
funcs = []
for i, (technique, params) in enumerate(
zip(self.techniques, self.technique_params)
):
params_copy = params.copy()
del params_copy["technique"]
strategy = Strategy(
id=i, technique=technique, technique_params=params_copy
)
funcs.append(strategy)
self.strategy_dict[strategy.id] = strategy
return funcs
ZNE_SETTINGS = Settings(
benchmarks=[
{
"circuit_type": "ghz",
"num_qubits": 2,
},
{
"circuit_type": "w",
"num_qubits": 2,
},
{
"circuit_type": "rb",
"num_qubits": 2,
"circuit_depth": 7,
},
{
"circuit_type": "mirror",
"num_qubits": 2,
"circuit_depth": 7,
"circuit_seed": 1,
},
],
strategies=[
{
"technique": "zne",
"scale_noise": fold_global,
"factory": RichardsonFactory([1.0, 2.0, 3.0]),
},
{
"technique": "zne",
"scale_noise": fold_global,
"factory": RichardsonFactory([1.0, 3.0, 5.0]),
},
{
"technique": "zne",
"scale_noise": fold_global,
"factory": LinearFactory([1.0, 2.0, 3.0]),
},
{
"technique": "zne",
"scale_noise": fold_global,
"factory": LinearFactory([1.0, 3.0, 5.0]),
},
{
"technique": "zne",
"scale_noise": fold_gates_at_random,
"factory": RichardsonFactory([1.0, 2.0, 3.0]),
},
{
"technique": "zne",
"scale_noise": fold_gates_at_random,
"factory": RichardsonFactory([1.0, 3.0, 5.0]),
},
{
"technique": "zne",
"scale_noise": fold_gates_at_random,
"factory": LinearFactory([1.0, 2.0, 3.0]),
},
{
"technique": "zne",
"scale_noise": fold_gates_at_random,
"factory": LinearFactory([1.0, 3.0, 5.0]),
},
],
)
PEC_SETTINGS = Settings(
benchmarks=[
{
"circuit_type": "ghz",
"num_qubits": 2,
},
{
"circuit_type": "w",
"num_qubits": 2,
},
{
"circuit_type": "rb",
"num_qubits": 2,
"circuit_depth": 7,
},
{
"circuit_type": "mirror",
"num_qubits": 2,
"circuit_depth": 7,
"circuit_seed": 1,
},
],
strategies=[
{
"technique": "pec",
"representation_function": (
represent_operation_with_local_depolarizing_noise
),
"is_qubit_dependent": False,
"noise_level": 0.001,
"num_samples": 200,
"force_run_all": False,
},
{
"technique": "pec",
"representation_function": (
represent_operation_with_local_depolarizing_noise
),
"is_qubit_dependent": False,
"noise_level": 0.01,
"num_samples": 200,
"force_run_all": False,
},
],
)
DefaultStrategy = Strategy(0, MitigationTechnique.RAW, {})
