def mitigated_results(circuit, backend, results):
# Import the required methods
from qiskit.providers.aer.noise import NoiseModel
from qiskit.ignis.mitigation.measurement import (complete_meas_cal,
CompleteMeasFitter)
# Get noise model for backend
noise_model = NoiseModel.from_backend(backend)
# Create the measurement fitter
qr = QuantumRegister(circuit.num_qubits)
meas_calibs, state_labels = complete_meas_cal(qr=qr, circlabel='mcal')
job = execute(meas_calibs,
backend=Aer.get_backend('qasm_simulator'),
shots=8192,
noise_model=noise_model)
cal_results = job.result()
meas_fitter = CompleteMeasFitter(cal_results,
state_labels,
circlabel='mcal')
#print(meas_fitter.cal_matrix)
# Plot the calibration matrix
print("Calibration matrix")
meas_fitter.plot_calibration()
# Get the filter object
meas_filter = meas_fitter.filter
# Results with mitigation
mitigated_results = meas_filter.apply(results)
mitigated_counts = mitigated_results.get_counts(0)
print("Mitigated", backend, "results:\n", mitigated_counts)
return (mitigated_counts)
def test_meas_cal_on_circuit(self):
"""Test an execution on a circuit."""
print("Testing measurement calibration on a circuit")
# Generate the calibration circuits
meas_calibs, state_labels, ghz = meas_calib_circ_creation()
# Run the calibration circuits
backend = Aer.get_backend('qasm_simulator')
job = qiskit.execute(meas_calibs,
backend=backend,
shots=self.shots,
seed_simulator=SEED,
seed_transpiler=SEED)
cal_results = job.result()
# Make a calibration matrix
meas_cal = CompleteMeasFitter(cal_results, state_labels)
# Calculate the fidelity
fidelity = meas_cal.readout_fidelity()
job = qiskit.execute([ghz],
backend=backend,
shots=self.shots,
seed_simulator=SEED,
seed_transpiler=SEED)
results = job.result()
# Predicted equally distributed results
predicted_results = {'000': 0.5, '111': 0.5}
meas_filter = meas_cal.filter
# Calculate the results after mitigation
output_results_pseudo_inverse = meas_filter.apply(
results, method='pseudo_inverse').get_counts(0)
output_results_least_square = meas_filter.apply(
results, method='least_squares').get_counts(0)
# Compare with expected fidelity and expected results
self.assertAlmostEqual(fidelity, 1.0)
self.assertAlmostEqual(output_results_pseudo_inverse['000'] /
self.shots,
predicted_results['000'],
places=1)
self.assertAlmostEqual(output_results_least_square['000'] / self.shots,
predicted_results['000'],
places=1)
self.assertAlmostEqual(output_results_pseudo_inverse['111'] /
self.shots,
predicted_results['111'],
places=1)
self.assertAlmostEqual(output_results_least_square['111'] / self.shots,
predicted_results['111'],
places=1)
def test_meas_fitter_with_noise(self):
"""
Test the MeasurementFitter with noise
"""
print("Testing MeasurementFitter with noise")
# pre-generated results with noise
# load from pickled file
fo = open(
os.path.join(os.path.dirname(__file__), 'test_meas_results.pkl'),
'rb')
tests = pickle.load(fo)
fo.close()
# Set the state labels
state_labels = ['000', '001', '010', '011', '100', '101', '110', '111']
meas_cal = CompleteMeasFitter(None, state_labels, circlabel='test')
for tst_index, _ in enumerate(tests):
# Set the calibration matrix
meas_cal.cal_matrix = tests[tst_index]['cal_matrix']
# Calculate the fidelity
fidelity = meas_cal.readout_fidelity()
meas_filter = MeasurementFilter(tests[tst_index]['cal_matrix'],
state_labels)
# Calculate the results after mitigation
output_results_pseudo_inverse = meas_filter.apply(
tests[tst_index]['results'], method='pseudo_inverse')
output_results_least_square = meas_filter.apply(
tests[tst_index]['results'], method='least_squares')
# Compare with expected fidelity and expected results
self.assertAlmostEqual(fidelity,
tests[tst_index]['fidelity'],
places=0)
self.assertAlmostEqual(
output_results_pseudo_inverse['000'],
tests[tst_index]['results_pseudo_inverse']['000'],
places=0)
self.assertAlmostEqual(
output_results_least_square['000'],
tests[tst_index]['results_least_square']['000'],
places=0)
self.assertAlmostEqual(
output_results_pseudo_inverse['111'],
tests[tst_index]['results_pseudo_inverse']['111'],
places=0)
self.assertAlmostEqual(
output_results_least_square['111'],
tests[tst_index]['results_least_square']['111'],
places=0)
def mitigated_results(backend,circuit,results,results_sim):
from qiskit import Aer, execute
from qiskit import QuantumRegister
# Import the required methods
from qiskit.providers.aer.noise import NoiseModel
from qiskit.ignis.mitigation.measurement import (complete_meas_cal,CompleteMeasFitter)
from qiskit.tools.visualization import plot_histogram
import numpy
numpy.set_printoptions(formatter={'float': lambda x: "{0:0.2f}".format(x)})
# Get noise model for backend
noise_model = NoiseModel.from_backend(backend)
# Create the measurement fitter
qr = QuantumRegister(circuit.num_qubits)
meas_calibs, state_labels = complete_meas_cal(qr=qr, circlabel='mcal')
job = execute(meas_calibs, backend=Aer.get_backend('qasm_simulator'), shots=8192, noise_model=noise_model)
cal_results = job.result()
meas_fitter = CompleteMeasFitter(cal_results, state_labels, circlabel='mcal')
print(meas_fitter.cal_matrix)
# Get the filter object
meas_filter = meas_fitter.filter
# Results with mitigation
mitigated_results = meas_filter.apply(results)
mitigated_counts = mitigated_results.get_counts(0)
title = "Mitigated Grover on "+str(ibmqbackend.name())
display(plot_histogram([results_sim.get_counts(),results.get_counts(), mitigated_counts], title=title, legend=['sim','noisy', 'mitigated']))
return(mitigated_counts)
def generate_mitigator(num_qubits, noisy, shots=1024):
meas_calibs, state_labels = complete_meas_cal(range(num_qubits))
experiments = transpile(meas_calibs, backend=noisy, optimization_level=3)
job = noisy.run(assemble(experiments, shots=shots))
cal_results = job.result()
meas_fitter = CompleteMeasFitter(cal_results, state_labels)
return job, meas_fitter.filter
def _qiskit_generate_samples_MEM(self: QiskitDevice) -> np.array:
"""
This function is used to overrite the default QiskitDevice.generate_samples method.
It adds a simple mechanism to perform measurement error mitigation on the results of
the computation.
"""
# branch out depending on the type of backend
if self.backend_name in self._state_backends:
# software simulator: need to sample from probabilities
return super().generate_samples()
# hardware or hardware simulator
# ***here we actully perform the measurement error mitigation***
meas_calibs, state_labels = complete_meas_cal(qr=self._reg, circlabel="mcal")
mitigation_run_args = self.run_args.copy()
mitigation_run_args["shots"] = NUMBER_OF_SHOTS_FOR_QISKIT_ERROR_MITIGATION
meas_job = qiskit.execute(meas_calibs, backend=self.backend, **mitigation_run_args)
# TODO: Adding a cache here is a free performance boost
meas_fitter = CompleteMeasFitter(meas_job.result(), state_labels, circlabel="mcal")
mitigated_results = meas_fitter.filter.apply(self._current_job.result())
current_job_shots = sum(self._current_job.result().get_counts().values())
summed_counts = sum(mitigated_results.get_counts().values())
probs = [value / summed_counts for value in mitigated_results.get_counts().values()]
samples = np.random.choice(
np.array(list(mitigated_results.get_counts().keys())),
current_job_shots,
p=probs,
replace=True,
)
# reverse qubit order to match PennyLane convention
return np.vstack([np.array([int(i) for i in s[::-1]]) for s in samples])
def matricaZaGreske(brojQubita, noise_model):
qr = QuantumRegister(brojQubita)
meas_calibs, state_labels = complete_meas_cal(qr=qr, circlabel='mcal')
backend = Aer.get_backend('qasm_simulator')
job = execute(meas_calibs, backend=backend, shots=1000, noise_model=noise_model)
cal_results = job.result()
meas_fitter = CompleteMeasFitter(cal_results, state_labels, circlabel='mcal')
return(meas_fitter.cal_matrix)
def generate_meas_calibration(results_file_path: str, runs: int):
"""
run the measurement calibration circuits, calculates the fitter matrix in few methods
and saves the results
The simulation results files will contain a list of dictionaries with the keys:
cal_matrix - the matrix used to calculate the ideal measurement
fidelity - the calculated fidelity of using this matrix
results - results of a bell state circuit with noise
results_pseudo_inverse - the result of using the psedo-inverse method on the bell state
results_least_square - the result of using the least-squares method on the bell state
Args:
results_file_path: path of the json file of the results file
runs: the number of different runs to save
"""
results = []
for run in range(runs):
cal_results, state_labels, circuit_results = \
meas_calibration_circ_execution(3, 1000, SEED + run)
meas_cal = CompleteMeasFitter(cal_results,
state_labels,
circlabel='test')
meas_filter = MeasurementFilter(meas_cal.cal_matrix, state_labels)
# Calculate the results after mitigation
results_pseudo_inverse = meas_filter.apply(circuit_results,
method='pseudo_inverse')
results_least_square = meas_filter.apply(circuit_results,
method='least_squares')
results.append({
"cal_matrix":
convert_ndarray_to_list_in_data(meas_cal.cal_matrix),
"fidelity":
meas_cal.readout_fidelity(),
"results":
circuit_results,
"results_pseudo_inverse":
results_pseudo_inverse,
"results_least_square":
results_least_square
})
with open(results_file_path, "w") as results_file:
json.dump(results, results_file)
def generate_error_mitigation_filter(q_list, noise_model):
backend = get_sim_backend_from_name("qasm_simulator")
qr = QuantumRegister(5)
meas_cals, state_labels = complete_meas_cal(qubit_list=q_list, qr=qr)
calibration_job = execute(meas_cals,
backend=backend,
shots=8192,
noise_model=noise_model)
cal_results = calibration_job.result()
meas_fitter = CompleteMeasFitter(cal_results, state_labels)
em_filter = meas_fitter.filter
return em_filter
def measurement_error_mitigator(systemsize,
sim,
quantum_com_choice_results,
shots=8192):
if sim == "noisy_qasm":
backend, coupling_map, noise_model = quantum_com_choice_results[sim]
qr = QuantumRegister(systemsize)
meas_calibs, state_labels = complete_meas_cal(qr=qr, circlabel='mcal')
print('Calibrating POVM Matrix')
job = execute(meas_calibs,
backend=backend,
shots=shots,
noise_model=noise_model,
coupling_map=coupling_map)
cal_results = job.result()
meas_fitter = CompleteMeasFitter(cal_results,
state_labels,
circlabel='mcal')
meas_filter = meas_fitter.filter
print("Provider backend: ", backend)
return meas_filter
elif sim == "real":
backend = quantum_com_choice_results[sim]
qr = QuantumRegister(systemsize)
meas_calibs, state_labels = complete_meas_cal(qr=qr, circlabel='mcal')
print('Calibrating POVM Matrix')
job = execute(meas_calibs, backend=backend, shots=shots)
job_monitor(job, interval=2)
cal_results = job.result()
meas_fitter = CompleteMeasFitter(cal_results,
state_labels,
circlabel='mcal')
meas_filter = meas_fitter.filter
print("Provider backend: ", backend)
return meas_filter
def apply(self, qc_list, info_list, shots):
meas_calibs, state_labels, measure_list = info_list[0], info_list[
1], info_list[2]
managed_results = IBMQ.qc_experiment(qc_list + meas_calibs,
self.backend, shots).run()
results = managed_results.combine_results()
meas_fitter = CompleteMeasFitter(results,
state_labels,
qubit_list=measure_list)
meas_filter = meas_fitter.filter
measurement_error_mitigated_results = meas_filter.apply(results)
return [results, measurement_error_mitigated_results]
def get_correction_matrix(qpu, max_age, access_token, used_qubits):
"""Return a correction matrix for the given QPU whit the maximum age in minutes"""
app.logger.info('Getting calibration matrix for QPU ' + qpu + ' with max age of ' + str(max_age) + ' minutes')
# Check for existing calibration matrix
existing_matrix = calibration_matrixes.get(qpu)
if existing_matrix is not None:
age = datetime.now() - existing_matrix['Date']
app.logger.info('Calibration matrix for this QPU exists with age ' + str(age))
if age.total_seconds() < max_age * 60:
app.logger.info('Returning existing calibration matrix!')
return existing_matrix['Calibration_Matrix']
if access_token is None:
app.logger.error('Unable to create new correction matrix without access key...')
return None
# Load account to enable execution of calibration circuits
IBMQ.save_account(access_token, overwrite=True)
IBMQ.load_account()
provider = IBMQ.get_provider(group='open')
backend = provider.get_backend(qpu)
# Generate a calibration circuit for each state
qr = qiskit.QuantumRegister(used_qubits)
meas_calibs, state_labels = complete_meas_cal(qr=qr, circlabel='mcal')
# Execute each calibration circuit and store results
app.logger.info('Executing ' + str(len(meas_calibs)) + ' circuits to create calibration matrix...')
cal_results = []
for circuit in meas_calibs:
app.logger.info('Executing circuit ' + circuit.name)
cal_results.append(execute_job(circuit, backend))
# Generate calibration matrix out of measurement results
meas_fitter = CompleteMeasFitter(cal_results, state_labels, circlabel='mcal')
app.logger.info('Calibration matrix:')
app.logger.info(meas_fitter.cal_matrix)
# Store calibration matrix for later reuse
calibration_matrixes[qpu] = {"Date": datetime.now(), "Calibration_Matrix": meas_fitter.filter}
return meas_fitter.filter
def initialize_meas_calibration(self, N_qubits, layout):
""" Set up the confusion matrix needed for measurement error mitigation.
This is basically just boilerplate code from the Ignis Github
https://github.com/Qiskit/qiskit-ignis
"""
if layout is None:
cal_q = QuantumRegister(N_qubits)
meas_cals, state_labels = complete_meas_cal(qr=cal_q)
else:
meas_cals, state_labels = complete_meas_cal(qubit_list=layout)
# Run the calibration circuits with the device noise model
backend = Aer.get_backend('qasm_simulator')
job = execute(meas_cals,
backend=backend,
shots=10000,
noise_model=self.noise_model)
cal_results = job.result()
return CompleteMeasFitter(cal_results, state_labels).filter
def _run_analysis(self, experiment_data, parameter_guess=None, plot=True, ax=None):
state_labels = []
for datum in experiment_data.data():
state_label = datum['metadata']['state_label']
if state_label in state_labels:
break
state_labels.append(state_label)
meas_fitter = CompleteMeasFitter(None, state_labels, circlabel='mcal')
nstates = len(state_labels)
for job_id in experiment_data.job_ids:
full_result = experiment_data.backend.retrieve_job(job_id).result()
# full_result might contain repeated experiments
for iset in range(len(full_result.results) // nstates):
try:
date = full_result.date
except:
date = None
try:
status = full_result.status
except:
status = None
try:
header = full_result.header
except:
header = None
result = Result(full_result.backend_name, full_result.backend_version, \
full_result.qobj_id, full_result.job_id, \
full_result.success, full_result.results[iset * nstates:(iset + 1) * nstates], \
date=date, status=status, header=header, **full_result._metadata)
meas_fitter.add_data(result)
results = [
AnalysisResultData('error_matrix', meas_fitter.cal_matrix, extra=state_labels)
]
plots = []
if plot:
figure, ax = plt.subplots(1, 1)
meas_fitter.plot_calibration(ax=ax)
plots.append(figure)
return results, plots
def get_measurement_fitter(l, backend, device, shots):
qb = QuantumRegister(l)
qubit_list = list(range(l))
meas_calibs, state_labels = complete_meas_cal(qubit_list=qubit_list,
qr=qb,
circlabel='mcal')
# Execute calibration circuits
job = execute(meas_calibs,
backend=backend,
shots=shots,
noise_model=device.noise_model,
coupling_map=device.coupling_map,
basis_gates=device.basis_gates)
cal_results = job.result()
meas_fitter = CompleteMeasFitter(cal_results,
state_labels,
circlabel='mcal')
#meas_filter = meas_fitter.filter
#mitigated_results = meas_filter.apply(results)
#mitigated_counts = mitigated_results.get_counts(0)
return meas_fitter
def apply_readout_correction(self, counts, qubit_list=None, **kwargs):
if self.readout_correction_filter is None:
for key in counts.keys():
num_qubits = len(key)
break
if qubit_list is None or qubit_list == {}:
qubit_list = [i for i in range(num_qubits)]
qr = QuantumRegister(num_qubits)
meas_cals, state_labels = complete_meas_cal(qubit_list=qubit_list, qr=qr)
# Execute the calibration circuits
job = self.execute_with_retries(meas_cals, self.n_samples)
cal_results = job.result()
# Make a calibration matrix
meas_fitter = CompleteMeasFitter(cal_results, state_labels)
# Create a measurement filter from the calibration matrix
self.readout_correction_filter = meas_fitter.filter
mitigated_counts = self.readout_correction_filter.apply(counts)
return mitigated_counts
def test_meas_cal_on_circuit(self):
"""Test an execution on a circuit."""
print("Testing measurement calibration on a circuit")
# Choose 3 qubits
q1 = 1
q2 = 2
q3 = 3
# Generate the quantum register according to the pattern
qr = qiskit.QuantumRegister(5)
# Generate the calibration circuits
meas_calibs, state_labels = \
complete_meas_cal(qubit_list=[1, 2, 3],
qr=qr)
# Run the calibration circuits
backend = Aer.get_backend('qasm_simulator')
job = qiskit.execute(meas_calibs, backend=backend, shots=self.shots)
cal_results = job.result()
# Make a calibration matrix
meas_cal = CompleteMeasFitter(cal_results, state_labels)
# Calculate the fidelity
fidelity = meas_cal.readout_fidelity()
# Make a 3Q GHZ state
cr = ClassicalRegister(3)
ghz = QuantumCircuit(qr, cr)
ghz.h(qr[q1])
ghz.cx(qr[q1], qr[q2])
ghz.cx(qr[q2], qr[q3])
ghz.measure(qr[q1], cr[0])
ghz.measure(qr[q2], cr[1])
ghz.measure(qr[q3], cr[2])
job = qiskit.execute([ghz], backend=backend, shots=self.shots)
results = job.result()
# Predicted equally distributed results
predicted_results = {'000': 0.5, '111': 0.5}
meas_filter = meas_cal.filter
# Calculate the results after mitigation
output_results_pseudo_inverse = meas_filter.apply(
results, method='pseudo_inverse').get_counts(0)
output_results_least_square = meas_filter.apply(
results, method='least_squares').get_counts(0)
# Compare with expected fidelity and expected results
self.assertAlmostEqual(fidelity, 1.0)
self.assertAlmostEqual(output_results_pseudo_inverse['000'] /
self.shots,
predicted_results['000'],
places=1)
self.assertAlmostEqual(output_results_least_square['000'] / self.shots,
predicted_results['000'],
places=1)
self.assertAlmostEqual(output_results_pseudo_inverse['111'] /
self.shots,
predicted_results['111'],
places=1)
self.assertAlmostEqual(output_results_least_square['111'] / self.shots,
predicted_results['111'],
places=1)
# If no mitigation is applied, add ', memory=True', and .get_memory() to job.result()
job = execute(qc, backend, shots=4096, optimization_level=0)
results = job.result()
# Mitigation Measurement, will run twice
cal_circuits, state_labels = complete_meas_cal(
qr=qc.qregs[0], circlabel='measurement_calibration')
calculatedJob = execute(cal_circuits,
backend,
shots=4096,
optimization_level=0)
calculatedResults = calculatedJob.result()
meas_fitter = CompleteMeasFitter(calculatedResults, state_labels)
meas_filter = meas_fitter.filter
mitigated_result = meas_filter.apply(results)
device_counts = results.get_counts(qc)
mitigated_counts = mitigated_result.get_counts(qc)
output = []
# Reminder, if mitigated results are disabled, add ".get_memory()" to job.result(), and uncomment:
# for x in range(0, 4096):
# convert = int(mitigated_counts_2[x], 2)
# output.append(convert)
# And comment out:
## main part!
from qiskit.ignis.mitigation.measurement import (complete_meas_cal,
CompleteMeasFitter)
cal_circuits, state_labels = complete_meas_cal(
qr=circuit.qregs[0], circlabel="measerrormitigationcal"
) #circuit.qregs[0] is the quantum register to show all qubits we have
#cal_circuits is the calibrated circuit of all qubits possible outcomes, state labels are 000 - 111
cal_job = execute(cal_circuits, qcom, shots=1024, optimization_level=0)
print(cal_job.job_id())
job_monitor(cal_job)
result_cal = cal_job.result()
plot_histogram(result_cal.get_counts())
#Build the measurement fitter which extracts the parameters from calibration results
meas_fitter = CompleteMeasFitter(
result_cal, state_labels) #fit the calibrated results with actucal labels
meas_fitter.plot_calibration()
#Make the calibration filter to mitigate the results
#This filter can be used for other 3 qubuits circuit
meas_filter = meas_fitter.filter
result_mitigated = meas_filter.apply(
result_q
) #Generate mitigated result by applying filter to the quantum computer result
#get result
counts_q = result_q.get_counts()
counts_mitigated = result_mitigated.get_counts()
plot_histogram([counts_q, counts_mitigated],
legend=["noisy result", 'mitigated result'])
def twospin_df(J, Bx, dB, Bz_max, dt_steps, dt_steps_bool, gamma, skip,
provider, backend, pq1, pq2):
count_list = []
calibrated_count_list = []
tarray, Bzarray, dt_or_steps, _ = adiabaticramp(J, Bx, dB, Bz_max,
dt_steps, dt_steps_bool,
gamma)
if dt_steps_bool == 'dt':
dt = dt_steps
else:
dt = dt_or_steps
thetaarray, _ = theta(Bzarray, dt)
# Calibration Matrix
meas_calibs, state_labels = complete_meas_cal(qubit_list=[0, 1, 2, 3],
qr=QuantumRegister(4),
circlabel='4spin')
cal_results = execute(meas_calibs,
backend=Aer.get_backend('qasm_simulator'),
shots=1000).result()
meas_calibs, state_labels = complete_meas_cal(qubit_list=[0, 1, 2, 3],
qr=QuantumRegister(4),
circlabel='4spin')
meas_fitter = CompleteMeasFitter(cal_results,
state_labels,
circlabel='4spin')
#meas_fitter.plot_calibration()
qmap = Layout()
i = 0
while (i < len(thetaarray)):
print('Bz = %f' % (Bzarray[i]))
qr = QuantumRegister(4, 'qr')
cr = ClassicalRegister(4, 'cr')
circ = QuantumCircuit(qr, cr)
#qmap.from_dict(input_dict={qr[0]: pq1, qr[1]: pq2})
circ.initialize([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1],
[0, 1, 2, 3])
fourspin_instruction(circ, J, Bx, thetaarray[:i + 1], dt)
circ.measure([0, 1, 2, 3], [0, 1, 2, 3])
result = execute(circ, Aer.get_backend('qasm_simulator'),
shots=5000).result()
counts = result.get_counts()
counts['Time'] = tarray[i]
counts['Bz'] = Bzarray[i]
count_list.append(counts)
with open("count_list.txt", "w") as output:
output.write(str(count_list))
mitigated_counts = meas_fitter.filter.apply(result).get_counts()
mitigated_counts['Time'] = tarray[i]
mitigated_counts['Bz'] = Bzarray[i]
calibrated_count_list.append(mitigated_counts)
with open("calibrated_count_list.txt", "w") as output:
output.write(str(calibrated_count_list))
i = i + 1 + skip
# Creating dataframe
df = pd.DataFrame(count_list)
time_col = df.pop('Time')
df.insert(0, 'Time', time_col)
df['Exact'] = exactm(J, Bx, Bzarray, dt)
df['Interaction'] = interactionm(J, Bx, thetaarray, dt)
calibrated_df = pd.DataFrame(calibrated_count_list)
time_col = df.pop('Time')
df.insert(0, 'Time', time_col)
if '0000' not in df:
df['0000'] = 0
if '0001' not in df:
df['0001'] = 0
if '0010' not in df:
df['0010'] = 0
if '0011' not in df:
df['0011'] = 0
if '0100' not in df:
df['0000'] = 0
if '0101' not in df:
df['0001'] = 0
if '0110' not in df:
df['0010'] = 0
if '0111' not in df:
df['0011'] = 0
if '1000' not in df:
df['0000'] = 0
if '1001' not in df:
df['0001'] = 0
if '1010' not in df:
df['0010'] = 0
if '1011' not in df:
df['0011'] = 0
if '1100' not in df:
df['0000'] = 0
if '1101' not in df:
df['0001'] = 0
if '1110' not in df:
df['0010'] = 0
if '1111' not in df:
df['0011'] = 0
df = df.fillna(0)
if '0000' not in calibrated_df:
calibrated_df['00'] = 0
if '0001' not in calibrated_df:
calibrated_df['01'] = 0
if '0010' not in calibrated_df:
calibrated_df['10'] = 0
if '0011' not in calibrated_df:
calibrated_df['11'] = 0
if '0100' not in calibrated_df:
calibrated_df['00'] = 0
if '0101' not in calibrated_df:
calibrated_df['01'] = 0
if '0110' not in calibrated_df:
calibrated_df['10'] = 0
if '0111' not in calibrated_df:
calibrated_df['11'] = 0
if '1000' not in calibrated_df:
calibrated_df['00'] = 0
if '1001' not in calibrated_df:
calibrated_df['01'] = 0
if '1010' not in calibrated_df:
calibrated_df['10'] = 0
if '1011' not in calibrated_df:
calibrated_df['11'] = 0
if '1100' not in calibrated_df:
calibrated_df['00'] = 0
if '1101' not in calibrated_df:
calibrated_df['01'] = 0
if '1110' not in calibrated_df:
calibrated_df['10'] = 0
if '1111' not in calibrated_df:
calibrated_df['11'] = 0
calibrated_df = calibrated_df.fillna(0)
# Calculating mz
total = df['0000'] + df['0001'] + df['0010'] + df['0011'] + df[
'0100'] + df['0101'] + df['0110'] + df['0111'] + df['1000'] + df[
'1001'] + df['1010'] + df['1011'] + df['1100'] + df['1101'] + df[
'1110'] + df['1111']
df['mz'] = -(2 * df['0000'] + df['0001'] + df['0010'] + df['0100'] +
df['1000'] - df['0111'] - df['1011'] - df['1101'] -
df['1110'] - 2 * df['1111']) / total
calibrated_df['mz'] = -(2 * df['0000'] + df['0001'] + df['0010'] +
df['0100'] + df['1000'] - df['0111'] - df['1011'] -
df['1101'] - df['1110'] - 2 * df['1111']) / total
# Creating Files
if dt_steps_bool == 'dt':
df.to_csv(
'4spin J={:.3f} Bx={:.3f} dB={:.3f} BzMax={:.3f} dt={:.3f} gamma={:.3f}.csv'
.format(J, Bx, dB, Bz_max, dt_steps, gamma))
calibrated_df.to_csv(
'Calibrated 4spin J={:.3f} Bx={:.3f} dB={:.3f} BzMax={:.3f} dt={:.3f} gamma={:.3f}.csv'
.format(J, Bx, dB, Bz_max, dt_steps, gamma))
else:
df.to_csv(
'4spin J={:.3f} Bx={:.3f} dB={:.3f} BzMax={:.3f} BzSteps={:.3f} gamma={:.3f}.csv'
.format(J, Bx, dB, Bz_max, dt_steps, gamma))
calibrated_df.to_csv(
'Calibrated 4spin J={:.3f} Bx={:.3f} dB={:.3f} BzMax={:.3f} BzSteps={:.3f} gamma={:.3f}.csv'
.format(J, Bx, dB, Bz_max, dt_steps, gamma))
return df, calibrated_df, dt_or_steps, thetaarray
def all_methods_data(interested_qubits,
backend,
itr,
QDT_correlated,
shots_per_point=1024,
file_address=''):
"""Collect data for our method, Qiskit method, QDT, and standard Bayesian.
Args:
interested_qubits:
an array of ints. REMEBER TO FOLLOW THE ORDER [LAST QUBIT, ..., FIRST QUBIT]
backend:
backend from provider.get_backend().
itr:
number of iterations of job submission in our method only.
QDT_correlated:
True if want qubits corrlected in QDT method.
file_address:
file address, string ends with '/' if not empty
Returns:
None
"""
with open(file_address + 'interested_qubits.csv', mode='w',
newline='') as sgm:
param_writer = csv.writer(sgm,
delimiter=',',
quotechar='"',
quoting=csv.QUOTE_MINIMAL)
param_writer.writerow(interested_qubits)
# Record data for filters (ourmethod)
print('Our method')
collect_filter_data(backend,
itr=itr,
shots=8192,
if_monitor_job=True,
if_write=True,
file_address=file_address)
# Qiskit Method
print('Qiskit Method')
qr = QuantumRegister(len(interested_qubits))
meas_calibs, state_labels = complete_meas_cal(qr=qr, circlabel='mcal')
for i in range(len(meas_calibs)):
meas_calibs[i] = transpile(meas_calibs[i],
backend,
initial_layout=interested_qubits[::-1])
job = execute(meas_calibs,
backend=backend,
shots=8192,
optimization_level=0)
job_monitor(job)
cal_results = job.result()
meas_fitter = CompleteMeasFitter(cal_results,
state_labels,
circlabel='mcal')
meas_filter = meas_fitter.filter
cal_matrix = meas_fitter.cal_matrix
with open(file_address + 'cal_matrix.csv', mode='w', newline='') as sgm:
param_writer = csv.writer(sgm,
delimiter=',',
quotechar='"',
quoting=csv.QUOTE_MINIMAL)
for row in cal_matrix:
param_writer.writerow(row)
with open(file_address + 'state_labels.csv', mode='w', newline='') as sgm:
param_writer = csv.writer(sgm,
delimiter=',',
quotechar='"',
quoting=csv.QUOTE_MINIMAL)
param_writer.writerow(state_labels)
# QDT
print('QDT, correlation = ', QDT_correlated)
if QDT_correlated:
qdt_qubit_indices = interested_qubits
qdt_probe_kets = povmtools.pauli_probe_eigenkets
qdt_calibration_circuits = detector_tomography_circuits(
qdt_qubit_indices, qdt_probe_kets)
print('Number of Circuits needed is ', len(qdt_calibration_circuits))
# We then execute them on backend prepared earlier.
shots_number = 8192
# Perform a noisy simulation
job = execute(qdt_calibration_circuits,
backend=backend,
shots=shots_number,
optimization_level=0)
job_monitor(job)
result = job.result()
dtf = DetectorTomographyFitter()
calibration_setup = QDTCalibrationSetup.from_qiskit_results(
[result], qdt_probe_kets)
ml_povm_estimator = dtf.get_maximum_likelihood_povm_estimator(
calibration_setup)
mitigator = QDTErrorMitigator()
mitigator.prepare_mitigator(ml_povm_estimator)
trans_mat = mitigator.transition_matrix
with open(file_address + 'trans_matrix.csv', mode='w',
newline='') as sgm:
param_writer = csv.writer(sgm,
delimiter=',',
quotechar='"',
quoting=csv.QUOTE_MINIMAL)
for row in trans_mat:
param_writer.writerow(row)
else:
for q in interested_qubits:
qdt_qubit_indices = [q]
qdt_probe_kets = povmtools.pauli_probe_eigenkets
qdt_calibration_circuits = detector_tomography_circuits(
qdt_qubit_indices, qdt_probe_kets)
print('Number of Circuits needed is ',
len(qdt_calibration_circuits))
# We then execute them on backend prepared earlier.
shots_number = 8192
# Perform a noisy simulation
job = execute(qdt_calibration_circuits,
backend=backend,
shots=shots_number,
optimization_level=0)
job_monitor(job)
result = job.result()
# Create Mitigator
dtf = DetectorTomographyFitter()
calibration_setup = QDTCalibrationSetup.from_qiskit_results(
[result], qdt_probe_kets)
ml_povm_estimator = dtf.get_maximum_likelihood_povm_estimator(
calibration_setup)
mitigator = QDTErrorMitigator()
mitigator.prepare_mitigator(ml_povm_estimator)
trans_mat = mitigator.transition_matrix
with open(file_address + 'trans_matrix' + str(q) + '.csv',
mode='w',
newline='') as sgm:
param_writer = csv.writer(sgm,
delimiter=',',
quotechar='"',
quoting=csv.QUOTE_MINIMAL)
for row in trans_mat:
param_writer.writerow(row)
# Data for Standard Bayesian
# Read data to for measurement error while input is |0>
print('Write data for standard Bayesian')
prop_dict = backend.properties().to_dict()
with open(file_address + 'given_params.csv', mode='w', newline='') as sgm:
param_writer = csv.writer(sgm,
delimiter=',',
quotechar='"',
quoting=csv.QUOTE_MINIMAL)
for q in interested_qubits:
p0m0 = 1 - prop_dict['qubits'][q][5]['value']
if p0m0 == 1 or p0m0 < 0.7:
p0m0 = 0.9
p1m1 = 1 - prop_dict['qubits'][q][4]['value']
if p1m1 == 1 or p1m1 < 0.7:
p1m1 = 0.9
param_writer.writerow([p0m0, p1m1])
with open(file_address + 'Filter_data.csv', mode='r') as measfile:
reader = csv.reader(measfile)
cali01 = np.asarray([row for row in reader][0])
Data = cali01
for q in interested_qubits:
y = getData0(Data, itr * int(8192 / shots_per_point), q)
with open(file_address + 'Qubit{}.csv'.format(q), mode='w',
newline='') as sgr:
read_writer = csv.writer(sgr,
delimiter=',',
quotechar='"',
quoting=csv.QUOTE_MINIMAL)
read_writer.writerow(['x', 'y'])
for i in range(len(y)):
read_writer.writerow([0.5, y[i]])
print('\nOptimization level must be an integer between 0 and 3, inclusive. Higher levels generate more optimized circuits, at the expense of longer transpilation time.\n')
exit(1)
else:
opt_level = 1
U_f = get_U_f(func_in, n)
list_y = []
circuit = simon_program(U_f, n)
circuit.measure(range(n), range(n))
# Calibration Matrix
meas_calibs, state_labels = complete_meas_cal(qubit_list=range(n), circlabel='mcal')
cal_job = execute(meas_calibs, backend=backend, shots=8192)
cal_results = cal_job.result()
meas_fitter = CompleteMeasFitter(cal_results, state_labels, circlabel='mcal')
start = time.time()
circuit = transpile(circuit, backend, optimization_level=opt_level)
transpiletime = time.time() - start
if trials*n > 2048:
trial_list = [8192]*((trials*n)//2048) + [(4*trials*n)%8192]
else:
trial_list = [4*trials*n]
job = []
for t in trial_list:
job.append(execute(circuit, backend, optimization_level=0, shots=t))
result = []
def test_ideal_meas_cal(self):
"""Test ideal execution, without noise."""
for nq in self.nq_list:
print("Testing %d qubit measurement calibration" % nq)
for pattern_type in range(1, 2**nq):
# Generate the quantum register according to the pattern
qubits, weight = self.choose_calibration(nq, pattern_type)
# Generate the calibration circuits
meas_calibs, state_labels = \
complete_meas_cal(qubit_list=qubits,
circlabel='test')
# Perform an ideal execution on the generated circuits
backend = Aer.get_backend('qasm_simulator')
job = qiskit.execute(meas_calibs,
backend=backend,
shots=self.shots)
cal_results = job.result()
# Make a calibration matrix
meas_cal = CompleteMeasFitter(cal_results,
state_labels,
circlabel='test')
# Assert that the calibration matrix is equal to identity
IdentityMatrix = np.identity(2**weight)
self.assertListEqual(
meas_cal.cal_matrix.tolist(), IdentityMatrix.tolist(),
'Error: the calibration matrix is \
not equal to identity')
# Assert that the readout fidelity is equal to 1
self.assertEqual(
meas_cal.readout_fidelity(), 1.0,
'Error: the average fidelity \
is not equal to 1')
# Generate ideal (equally distributed) results
results_dict, results_list = \
self.generate_ideal_results(state_labels, weight)
# Output the filter
meas_filter = meas_cal.filter
# Apply the calibration matrix to results
# in list and dict forms using different methods
results_dict_1 = meas_filter.apply(results_dict,
method='least_squares')
results_dict_0 = meas_filter.apply(results_dict,
method='pseudo_inverse')
results_list_1 = meas_filter.apply(results_list,
method='least_squares')
results_list_0 = meas_filter.apply(results_list,
method='pseudo_inverse')
# Assert that the results are equally distributed
self.assertListEqual(results_list, results_list_0.tolist())
self.assertListEqual(results_list,
np.round(results_list_1).tolist())
self.assertDictEqual(results_dict, results_dict_0)
round_results = {}
for key, val in results_dict_1.items():
round_results[key] = np.round(val)
self.assertDictEqual(results_dict, round_results)
# Generates measurement calibration circuits for running measurement error
# mitigation.
q_register = QuantumRegister(5)
calibration_circuits, state_labels = complete_meas_cal(qubit_list=[2, 3, 4],
qr=q_register)
# Executes calibration circuits.
qasm_backend = Aer.get_backend("qasm_simulator") # simulates circuit
job = qiskit.execute(calibration_circuits,
backend=qasm_backend,
shots=1000,
noise_model=aer_noise_model)
calibration_results = job.result()
# Makes calibration matrix.
measure_fitter = CompleteMeasFitter(calibration_results, state_labels)
# Makes a 3-qubit GHZ state (a particular entangled state, at least 3 qubits).
# 1. applies Hadamard, putting qubit into superposition
# 2. applies CX (CNOT), putting qubits into Bell state
# 3. applies CX (CNOT), putting qubits into GHZ state
c_register = ClassicalRegister(3)
ghz = QuantumCircuit(q_register, c_register)
ghz.h(q_register[2]) # ghz.h(qubit); Hadamard gate
ghz.cx(q_register[2],
q_register[3]) # ghz.cx(control_qubit, target_qubit); CNOT
ghz.cx(q_register[3], q_register[4])
ghz.measure(q_register[2], c_register[0]) # ghz.measure(qubit, cbit):
# qubit: quantum register,
# cbit: classical register,
# measures qubit into classical bit
flip_drive = True
backend_sim = qiskit.providers.aer.PulseSimulator()
# experiments10 = cr_drive_experiments(drive_idx, target_idx, flip_drive)
# compute frequencies from the Hamiltonian
# qubit_lo_freq10 = two_qubit_model.hamiltonian.get_qubit_lo_from_drift()
#assemble qobj 1
rabifit_result = RabiFitter(result1, amps1, qubits, fit_p0 = [0.5,0.5,0.6,1.5])
if l == trainidx[0] and i == 0:
#if flag != 1:
# Generate the calibration circuits
qr = qiskit.QuantumRegister(2)
qubit_list = [0,1]
meas_calibs, state_labels = complete_meas_cal(qubit_list=qubit_list, qr=qr, circlabel='mcal')
job = qiskit.execute(meas_calibs, backend=backend, shots=10, noise_model=noise_model)
cal_results = job.result()
meas_fitter = CompleteMeasFitter(cal_results, state_labels)
# meas_fitter.cal_matrix = np.kron(meas_fitter.cal_matrices[1],meas_fitter.cal_matricesc[0])
meas_fitter.cal_matrix
plt.figure(figsize=(15, 10))
for qubit in qubits:
ax = plt.subplot(2, 2, qubit + 1)
#Xvmin, Xvmax = ax.xaxis.get_data_interval()
#Yvmin, Yvmax = ax.yaxis.get_data_interval()
#print(Xvmin, Xvmax,Yvmin, Yvmax)
Xvmin = multiplier * np.floor(0.1 / multiplier)
Xvmax = multiplier * np.ceil(0.5 / multiplier)
ax.set_xlim([Xvmin, Xvmax])
rabifit_result.plot(qubit, ax=ax)
print('Q1 Pi Amp: %f'%rabifit_result.pi_amplitude(qubit))
# analysis on multi-shots averaged result
# Results without mitigation
result2 = job2.result() plot_histogram(result2.get_counts(circuit3)) # #Mitigation grover 2 from qiskit.ignis.mitigation.measurement import (complete_meas_cal, CompleteMeasFitter) cal_circuits, state_labels = complete_meas_cal(qr = circuit2.qregs[0], circlabel = 'measerrormitigationcal') cal_circuits[2].draw(output='mpl') # cal_job = execute(cal_circuits, backend = qcomp, shots = 1024, optimization_level = 0) print(cal_job.job_id()) job_monitor(cal_job) cal_results = cal_job.result() # plot_histogram(cal_results.get_counts(cal_circuits[3])) # meas_fitter = CompleteMeasFitter(cal_results, state_labels) meas_fitter.plot_calibration() # meas_filter = meas_fitter.filter mitigated_result = meas_filter.apply(result) device_counts = result.get_counts(circuit2) mitigated_counts = mitigated_result.get_counts(circuit2) plot_histogram([device_counts, mitigated_counts], legend=['device, noisy', 'device, mitigated']) # #Mitigation grover 3 cal_circuits2, state_labels2 = complete_meas_cal(qr = circuit3.qregs[0], circlabel = 'measerrormitigationcal') cal_circuits2[2].draw(output='mpl') # cal_job2 = execute(cal_circuits2, backend = qcomp, shots = 1024, optimization_level = 0) print(cal_job2.job_id()) job_monitor(cal_job2)