qiskit.ignis.verification.quantum_volume.fitters のソースコード
# -*- coding: utf-8 -*-
# This code is part of Qiskit.
#
# (C) Copyright IBM 2019.
#
# 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.
"""
Functions used for the analysis of quantum volume results.
Based on Cross et al. "Validating quantum computers using
randomized model circuits", arXiv:1811.12926
"""
import math
import warnings
import numpy as np
from qiskit import QiskitError
from qiskit.visualization import plot_histogram
from ...utils import build_counts_dict_from_list
try:
from matplotlib import get_backend
from matplotlib import pyplot as plt
from matplotlib.patches import Rectangle
HAS_MATPLOTLIB = True
except ImportError:
HAS_MATPLOTLIB = False
[ドキュメント]class QVFitter:
"""Class for fitters for quantum volume."""
[ドキュメント] def __init__(self, backend_result=None, statevector_result=None,
qubit_lists=None):
"""
Args:
backend_result (list): list of results (qiskit.Result).
statevector_result (list): the ideal statevectors of each circuit
qubit_lists (list): list of qubit lists (what was passed to the
circuit generation)
"""
self._qubit_lists = qubit_lists
self._depths = [len(qubit_list) for qubit_list in qubit_lists]
self._ntrials = 0
self._result_list = []
self._heavy_output_counts = {}
self._circ_shots = {}
self._circ_counts = {}
self._all_output_prob_ideal = {}
self._heavy_output_prob_ideal = {}
self._heavy_output_prob_exp = {}
self._ydata = []
self._heavy_outputs = {}
self.add_statevectors(statevector_result)
self.add_data(backend_result)
@property
def depths(self):
"""Return depth list."""
return self._depths
@property
def qubit_lists(self):
"""Return depth list."""
return self._qubit_lists
@property
def results(self):
"""Return all the results."""
return self._result_list
@property
def heavy_outputs(self):
"""Return the ideal heavy outputs dictionary."""
return self._heavy_outputs
@property
def heavy_output_counts(self):
"""Return the number of heavy output counts as measured."""
return self._heavy_output_counts
@property
def heavy_output_prob_ideal(self):
"""Return the heavy output probability ideally."""
return self._heavy_output_prob_ideal
@property
def ydata(self):
"""Return the average and std of the output probability."""
return self._ydata
[ドキュメント] def add_statevectors(self, new_statevector_result):
"""
Add the ideal results and convert to the heavy outputs.
Assume the result is from 'statevector_simulator'
Args:
new_statevector_result (list): ideal results
Raises:
QiskitError: If the result has already been added for the circuit
"""
if new_statevector_result is None:
return
if not isinstance(new_statevector_result, list):
new_statevector_result = [new_statevector_result]
for result in new_statevector_result:
for qvcirc in result.results:
circ_name = qvcirc.header.name
# get the depth/width from the circuit name
# qv_depth_%d_trial_%d
depth = int(circ_name.split('_')[2])
if circ_name in self._heavy_outputs:
raise QiskitError("Already added the ideal result "
"for circuit %s" % circ_name)
# convert the result into probability dictionary
qstate = result.get_statevector(circ_name)
pvector = np.multiply(qstate, qstate.conjugate())
format_spec = "{0:0%db}" % depth
self._all_output_prob_ideal[circ_name] = \
{format_spec.format(b): float(np.real(pvector[b])) for b in range(2**depth)}
median_prob = self._median_probabilities([self._all_output_prob_ideal[circ_name]])
self._heavy_outputs[circ_name] = \
self._heavy_strings(self._all_output_prob_ideal[circ_name], median_prob[0])
# calculate the heavy output probability
self._heavy_output_prob_ideal[circ_name] = \
self._subset_probability(
self._heavy_outputs[circ_name],
self._all_output_prob_ideal[circ_name])
[ドキュメント] def add_data(self, new_backend_result, rerun_fit=True):
"""
Add a new result. Re calculate fit
Args:
new_backend_result (list): list of qv results
rerun_fit (bool): re calculate the means and fit the result
Raises:
QiskitError: If the ideal distribution isn't loaded yet
Additional information:
Assumes that 'result' was executed is
the output of circuits generated by qv_circuits,
"""
if new_backend_result is None:
return
if not isinstance(new_backend_result, list):
new_backend_result = [new_backend_result]
for result in new_backend_result:
self._result_list.append(result)
# update the number of trials *if* new ones
# added.
for qvcirc in result.results:
ntrials_circ = int(qvcirc.header.name.split('_')[-1])
if (ntrials_circ+1) > self._ntrials:
self._ntrials = ntrials_circ+1
if qvcirc.header.name not in self._heavy_output_prob_ideal:
raise QiskitError('Ideal distribution '
'must be loaded first')
if rerun_fit:
self.calc_data()
self.calc_statistics()
[ドキュメント] def calc_data(self):
"""
Make a count dictionary for each unique circuit from all the results.
Calculate the heavy output probability.
Additional information:
Assumes that 'result' was executed is
the output of circuits generated by qv_circuits,
"""
for trialidx in range(self._ntrials):
for _, depth in enumerate(self._depths):
circ_name = 'qv_depth_%d_trial_%d' % (depth, trialidx)
# get the counts form ALL executed circuits
count_list = []
for result in self._result_list:
try:
count_list.append(result.get_counts(circ_name))
except (QiskitError, KeyError):
pass
self._circ_counts[circ_name] = \
build_counts_dict_from_list(count_list)
self._circ_shots[circ_name] = \
sum(self._circ_counts[circ_name].values())
# calculate the experimental heavy output counts
self._heavy_output_counts[circ_name] = \
self._subset_probability(
self._heavy_outputs[circ_name],
self._circ_counts[circ_name])
# calculate the experimental heavy output probability
self._heavy_output_prob_exp[circ_name] = \
self._heavy_output_counts[circ_name]/self._circ_shots[circ_name]
# calculate the experimental heavy output probability
self._heavy_output_prob_exp[circ_name] = \
self._heavy_output_counts[circ_name] / self._circ_shots[circ_name]
[ドキュメント] def calc_statistics(self):
"""
Convert the heavy outputs in the different trials into mean and error
for plotting.
Here we assume the error is due to a binomial distribution.
Error (standard deviation) for binomial distribution is sqrt(np(1-p)),
where n is the number of trials (self._ntrials) and
p is the success probability (self._ydata[0][depthidx]/self._ntrials).
"""
self._ydata = np.zeros([4, len(self._depths)], dtype=float)
exp_vals = np.zeros(self._ntrials, dtype=float)
ideal_vals = np.zeros(self._ntrials, dtype=float)
for depthidx, depth in enumerate(self._depths):
exp_shots = 0
for trialidx in range(self._ntrials):
cname = 'qv_depth_%d_trial_%d' % (depth, trialidx)
exp_vals[trialidx] = self._heavy_output_counts[cname]
exp_shots += self._circ_shots[cname]
ideal_vals[trialidx] = self._heavy_output_prob_ideal[cname]
# Calculate mean and error for experimental data
self._ydata[0][depthidx] = np.sum(exp_vals)/np.sum(exp_shots)
self._ydata[1][depthidx] = (self._ydata[0][depthidx] *
(1.0-self._ydata[0][depthidx])
/ self._ntrials)**0.5
# Calculate mean and error for ideal data
self._ydata[2][depthidx] = np.mean(ideal_vals)
self._ydata[3][depthidx] = (self._ydata[2][depthidx] *
(1.0-self._ydata[2][depthidx])
/ self._ntrials)**0.5
[ドキュメント] def plot_qv_data(self, ax=None, show_plt=True, figsize=(7, 5), set_title=True, title=None):
"""Plot the qv data as a function of depth
Args:
ax (Axes or None): plot axis (if passed in).
show_plt (bool): display the plot.
figsize (tuple): Figure size in inches.
set_title (bool): set figure title.
title (String or None): text for setting figure title
Raises:
ImportError: If matplotlib is not installed.
Returns:
matplotlib.Figure:
A figure of Quantum Volume data (heavy
output probability) with two-sigma error
bar as a function of circuit depth.
"""
if not HAS_MATPLOTLIB:
raise ImportError('The function plot_rb_data needs matplotlib. '
'Run "pip install matplotlib" before.')
if ax is None:
fig, ax = plt.subplots(figsize=figsize)
else:
fig = None
xdata = range(len(self._depths))
# Plot the experimental data with error bars
ax.errorbar(xdata, self._ydata[0],
yerr=self._ydata[1]*2,
color='r', marker='o',
markersize=6, capsize=5,
elinewidth=2, label='Exp (2$\\sigma$ error)')
# Plot the ideal data with error bars
ax.errorbar(xdata, self._ydata[2],
yerr=self._ydata[3]*2,
color='b', marker='v',
markersize=6, capsize=5,
elinewidth=2, label='Ideal (2$\\sigma$ error)')
# Plot the threshold
ax.axhline(2/3, color='k', linestyle='dashed', linewidth=1, label='Threshold')
ax.set_xticks(xdata)
ax.set_xticklabels(self._qubit_lists, rotation=45)
ax.set_xlabel('Qubit Subset', fontsize=14)
ax.set_ylabel('Heavy Output Probability', fontsize=14)
ax.grid(True)
ax.legend()
if set_title:
if title is None:
title = (
f'Quantum Volume for up to {len(self._qubit_lists[-1])} Qubits '
f'and {self._ntrials} Trials')
ax.set_title(title)
if fig:
if get_backend() in ['module://ipykernel.pylab.backend_inline',
'nbAgg']:
plt.close(fig)
if show_plt:
plt.show()
return fig
[ドキュメント] def plot_qv_trial(self, depth, trial_index, figsize=(7, 5), ax=None):
"""Plot individual trial.
Args:
depth(int): circuit depth
trial_index(int): trial index
figsize (tuple): Figure size in inches.
ax (Axes or None): plot axis (if passed in).
Returns:
matplotlib.Figure:
A figure for histogram of ideal and experiment probabilities.
"""
circ_name = f"qv_depth_{depth}_trial_{trial_index}"
ideal_data = self._all_output_prob_ideal[circ_name]
exp_data = self._circ_counts[circ_name]
if ax is None:
fig, ax = plt.subplots(figsize=figsize)
else:
fig = None
# plot experimental histogram
plot_histogram(exp_data, legend=['Exp'], ax=ax)
# plot idea histogram overlap with experimental values
plot_histogram(ideal_data, legend=['Ideal'], bar_labels=False, ax=ax, color='r')
# get ideal histograms and change to unfilled
bars = [r for r in ax.get_children() if isinstance(r, Rectangle)]
for i in range(int(len(bars)/2), len(bars)-1):
bars[i].fill = False
# set non-black edge color to increase bar labels legibility
bars[i].set_edgecolor('saddlebrown')
# show experimental heavy output probability to the legend
ax.plot([], [], ' ', label=f'HOP~{self._heavy_output_prob_exp[circ_name]:.3f}')
# plot median probability
median_prob = self._median_probabilities([self._all_output_prob_ideal[circ_name]])
ax.axhline(median_prob, color='r', linestyle='dashed', linewidth=1, label='Median')
ax.legend()
ax.set_title(f'Quantum Volume {2**depth}, Trial #{trial_index}', fontsize=14)
# Only close mpl figures in jupyter with inline backends
if get_backend() in ['module://ipykernel.pylab.backend_inline',
'nbAgg']:
plt.close(fig)
return fig
[ドキュメント] def plot_hop_accumulative(self, depth, ax=None, figsize=(7, 5)):
"""Plot individual and accumulative heavy output probability (HOP)
as a function of number of trials.
Args:
depth (int): depth of QV circuits
ax (Axes or None): plot axis (if passed in).
figsize (tuple): figure size in inches.
Raises:
ImportError: If matplotlib is not installed.
Returns:
matplotlib.Figure:
A figure of individual and accumulative HOP as a function of number of trials,
with 2-sigma confidence interval and 2/3 threshold.
"""
if not HAS_MATPLOTLIB:
raise ImportError('The function plot_hop_accumulative needs matplotlib. '
'Run "pip install matplotlib" before.')
if ax is None:
fig, ax = plt.subplots(figsize=figsize)
else:
fig = None
trial_list = np.arange(self._ntrials) # x data
hop_list = [] # y data
for trial_index in range(self._ntrials):
circ_name = f'qv_depth_{depth}_trial_{trial_index}'
hop_list.append(self._heavy_output_prob_exp[circ_name])
hop_accumulative = np.cumsum(hop_list) / np.arange(1, self._ntrials+1)
two_sigma = 2 * (hop_accumulative * (1 - hop_accumulative) /
np.arange(1, self._ntrials+1))**0.5
# plot two-sigma shaded area
ax.errorbar(trial_list, hop_accumulative, fmt="none", yerr=two_sigma, ecolor='lightgray',
elinewidth=20, capsize=0, alpha=0.5, label='2$\\sigma$')
# plot accumulative HOP
ax.plot(trial_list, hop_accumulative, color='r', label='Cumulative HOP')
# plot inidivual HOP as scatter
ax.scatter(trial_list, hop_list, s=3, zorder=3, label='Individual HOP')
# plot 2/3 success threshold
ax.axhline(2/3, color='k', linestyle='dashed', linewidth=1, label='Threshold')
ax.set_xlim(0, self._ntrials)
ax.set_ylim(hop_accumulative[-1]-4*two_sigma[-1], hop_accumulative[-1]+4*two_sigma[-1])
ax.set_xlabel('Number of Trials', fontsize=14)
ax.set_ylabel('Heavy Output Probability', fontsize=14)
ax.set_title(f'Quantum Volume {2**depth} Trials', fontsize=14)
# re-arrange legend order
handles, labels = ax.get_legend_handles_labels()
handles = [handles[1], handles[2], handles[0], handles[3]]
labels = [labels[1], labels[2], labels[0], labels[3]]
ax.legend(handles, labels)
# Only close mpl figures in jupyter with inline backends
if fig:
if get_backend() in ['module://ipykernel.pylab.backend_inline',
'nbAgg']:
plt.close(fig)
return fig
[ドキュメント] def qv_success(self):
"""Return whether each depth was successful (> 2/3 with confidence
level > 0.977 corresponding to z_value = 2) and the confidence level.
Returns:
list: List of list of 2 elements for each depth:
- success True/False
- confidence level
"""
success_list = []
confidence_level_threshold = self.calc_confidence_level(z_value=2)
for depth_ind, _ in enumerate(self._depths):
success_list.append([False, 0.0])
hmean = self._ydata[0][depth_ind]
sigma = self._ydata[1][depth_ind]
z_value = self.calc_z_value(hmean, sigma)
confidence_level = self.calc_confidence_level(z_value)
success_list[-1][1] = confidence_level
if (hmean > 2/3 and confidence_level > confidence_level_threshold):
success_list[-1][0] = True
return success_list
[ドキュメント] def calc_z_value(self, mean, sigma):
"""Calculate z value using mean and sigma.
Args:
mean (float): mean
sigma (float): standard deviation
Returns:
float: z_value in standard normal distibution.
"""
if sigma == 0:
# assign a small value for sigma if sigma = 0
sigma = 1e-10
warnings.warn('Standard deviation sigma should not be zero.')
z_value = (mean - 2/3) / sigma
return z_value
[ドキュメント] def calc_confidence_level(self, z_value):
"""Calculate confidence level using z value.
Accumulative probability for standard normal distribution
in [-z, +infinity] is 1/2 (1 + erf(z/sqrt(2))),
where z = (X - mu)/sigma = (hmean - 2/3)/sigma
Args:
z_value (float): z value in in standard normal distibution.
Returns:
float: confidence level in decimal (not percentage).
"""
confidence_level = 0.5 * (1 + math.erf(z_value/2**0.5))
return confidence_level
[ドキュメント] def quantum_volume(self):
"""Return the volume for each depth.
Returns:
list: List of quantum volumes
"""
qv_list = 2**np.array(self._depths)
return qv_list
def _heavy_strings(self, ideal_distribution, ideal_median):
"""Return the set of heavy output strings.
Args:
ideal_distribution (dict): dict of ideal output distribution
where keys are bit strings (as strings) and values are
probabilities of observing those strings
ideal_median (float): median probability across all outputs
Returns:
list: list the set of heavy output strings, i.e. those strings
whose ideal probability of occurrence exceeds the median.
"""
return list(filter(lambda x: ideal_distribution[x] > ideal_median,
list(ideal_distribution.keys())))
def _median_probabilities(self, distributions):
"""Return a list of median probabilities.
Args:
distributions (list): list of dicts mapping binary strings
(as strings) to probabilities.
Returns:
list: a list of median probabilities.
"""
medians = []
for dist in distributions:
values = np.array(list(dist.values()))
medians.append(float(np.real(np.median(values))))
return medians
def _subset_probability(self, strings, distribution):
"""Return the probability of a subset of outcomes.
Args:
strings (list): list of bit strings (as strings)
distribution (dict): dict where keys are bit strings (as strings)
and values are probabilities of observing those strings
Returns:
float: the probability of the subset of strings, i.e. the sum
of the probabilities of each string as given by the
distribution.
"""
return sum([distribution.get(value, 0) for value in strings])