def benchmarking_pi2_multi(device,
qubit_ids,
*params,
interleaver=None,
two_qubit_gate=None,
max_pulses=None,
pause_length=0,
random_sequence_num=1,
seq_lengths_num=400,
two_qubit_num=0,
random_gate_num=1):
channel_amplitudes_ = {}
pi2_pulses = {}
pi_pulses = {}
generators = {}
if max_pulses is None:
max_pulses = []
for qubit_id in qubit_ids:
coherence_measurement = Ramsey.get_Ramsey_coherence_measurement(
device, qubit_id)
T2 = float(coherence_measurement.metadata['T'])
pi2_pulses[qubit_id] = excitation_pulse.get_excitation_pulse(
device, qubit_id, np.pi / 2.)
pi2_pulse_length = float(pi2_pulses[qubit_id].metadata['length'])
pi_pulses[qubit_id] = excitation_pulse.get_excitation_pulse(
device, qubit_id, np.pi)
pi_pulse_length = float(pi_pulses[qubit_id].metadata['length'])
max_pulses.append(T2 / pi2_pulse_length)
if two_qubit_gate is not None:
max_pulses = np.asarray(max_pulses) / 3.
max_pulses = min(max_pulses)
for qubit_id in qubit_ids:
pi2_pulses[qubit_id] = excitation_pulse.get_excitation_pulse(
device, qubit_id, np.pi / 2.)
pi_pulses[qubit_id] = excitation_pulse.get_excitation_pulse(
device, qubit_id, np.pi)
channel_amplitudes_[qubit_id] = channel_amplitudes.channel_amplitudes(
device.exdir_db.select_measurement_by_id(
pi2_pulses[qubit_id].references['channel_amplitudes']))
seq_lengths = np.asarray(
np.round(np.linspace(0, max_pulses, seq_lengths_num)), int)
def get_pulse_seq_z(z_phase, length, qubit_id):
fast_control = False
z_pulse = [(c, device.pg.virtual_z, z_phase * 360 / 2 / np.pi,
fast_control)
for c, a in channel_amplitudes_[qubit_id].items()]
sequence_z = [device.pg.pmulti(device, length, *tuple(z_pulse))]
return sequence_z
def tensor_product(unitary, qubit_id):
U = [[1]]
for i in qubit_ids:
U = np.kron(U, np.identity(2) if i != qubit_id else unitary)
return U
#TODO
qubit_readout_pulse, readout_device = calibrated_readout.get_calibrated_measurer(
device, qubit_ids)
generators = {}
for qubit_id in qubit_ids:
HZ = {
'X': {
'pulses': pi_pulses[qubit_id].get_pulse_sequence(0),
'unitary': tensor_product(np.asarray([[0, 1], [1, 0]]),
qubit_id),
'price': 1.0
},
'X/2': {
'pulses':
pi2_pulses[qubit_id].get_pulse_sequence(0),
'unitary':
np.sqrt(0.5) *
tensor_product(np.asarray([[1, -1j], [-1j, 1]]), qubit_id),
'price':
1.0
},
'-X/2': {
'pulses':
pi2_pulses[qubit_id].get_pulse_sequence(np.pi),
'unitary':
np.sqrt(0.5) *
tensor_product(np.asarray([[1, 1j], [1j, 1]]), qubit_id),
'price':
1.0
},
'Z': {
'pulses': get_pulse_seq_z(np.pi, pi2_pulse_length, qubit_id),
'unitary': tensor_product([[1, 0], [0, -1]], qubit_id),
'price': 0.1
},
'Z/2': {
'pulses': get_pulse_seq_z(np.pi / 2, pi2_pulse_length,
qubit_id),
'unitary': tensor_product([[1, 0], [0, 1j]], qubit_id),
'price': 0.1
},
'-Z/2': {
'pulses': get_pulse_seq_z(-np.pi / 2., pi2_pulse_length,
qubit_id),
'unitary': tensor_product([[1, 0], [0, -1j]], qubit_id),
'price': 0.1
},
'I': {
'pulses': get_pulse_seq_z(0, pi2_pulse_length, qubit_id),
'unitary': tensor_product([[1, 0], [0, 1]], qubit_id),
'price': 0.1
}
}
generators[qubit_id] = HZ
if len(qubit_ids) == 2:
#TODO
HZ_group = clifford.two_qubit_clifford(
*tuple([g for g in generators.values()]),
plus_op_parallel=device.pg.parallel,
two_qubit_gate=two_qubit_gate)
elif len(qubit_ids) == 1:
HZ_group = clifford.generate_group(generators[qubit_ids[0]])
else:
raise ValueError('More than two qubits are unsupported')
print('group length:', len(HZ_group))
# TODO qubit sequencer
exitation_channel = [
i
for i in device.get_qubit_excitation_channel_list(qubit_ids[0]).keys()
][0]
ex_channel = device.awg_channels[exitation_channel]
if ex_channel.is_iq():
control_seq_id = ex_channel.parent.sequencer_id
else:
control_seq_id = ex_channel.channel // 2
ex_sequencers = []
for seq_id in device.pre_pulses.seq_in_use:
if seq_id != control_seq_id:
ex_seq = zi_scripts.SIMPLESequence(sequencer_id=seq_id,
awg=device.modem.awg,
awg_amp=1,
use_modulation=True,
pre_pulses=[])
else:
ex_seq = zi_scripts.SIMPLESequence(sequencer_id=seq_id,
awg=device.modem.awg,
awg_amp=1,
use_modulation=True,
pre_pulses=[],
control=True)
control_sequence = ex_seq
device.pre_pulses.set_seq_offsets(ex_seq)
device.pre_pulses.set_seq_prepulses(ex_seq)
ex_seq.start()
ex_sequencers.append(ex_seq)
seeds = np.random.randint(100000,
size=(random_sequence_num, len(qubit_ids),
len(seq_lengths)))
references = {'seeds': seeds}
pi2_bench = interleaved_benchmarking.interleaved_benchmarking(
readout_device,
ex_sequencers,
seeds,
seq_lengths,
interleavers=HZ_group,
random_sequence_num=random_sequence_num,
two_qubit_num=two_qubit_num,
random_gate_num=random_gate_num)
#TODO prepare_seq
prepare_seq = pi2_bench.create_hdawg_generator()
sequence_control.set_preparation_sequence(device, ex_sequencers,
prepare_seq)
#TODO readout sequence
#ro_seq = [device.pg.pmulti(pause_length)]+device.trigger_readout_seq+qubit_readout_pulse.get_pulse_sequence()
readout_sequencer = sequence_control.define_readout_control_seq(
device, qubit_readout_pulse)
readout_sequencer.start()
pi2_bench.random_sequence_num = random_sequence_num
seeds_ids = np.arange(seeds.shape[0])
references = references.update({('pi2_pulse', qubit_id):
pi2_pulses[qubit_id].id
for qubit_id in qubit_ids})
references = references.update({('pi_pulse', qubit_id):
pi_pulses[qubit_id].id
for qubit_id in qubit_ids})
# TODO
# pi2_bench.prepare_random_interleaving_sequences()
### search db for previous version of the interleaver measurement
found = False
try:
clifford_bench = device.exdir_db.select_measurement(
measurement_type='clifford_bench',
metadata={'qubit_ids': ','.join(qubit_ids)},
references_that=references)
found = True
except IndexError:
pass
if random_sequence_num > 1:
params = tuple([(seeds_ids, pi2_bench.set_interleaved_sequence,
'Random seeds id', '')] + [p for p in params])
if (not found) or (interleaver is None):
measurement_name = [m for m in pi2_bench.get_points().keys()][0]
fitter_arguments = (measurement_name, exp.exp_fitter(), 0,
np.arange(len(params)).tolist())
clifford_bench = device.sweeper.sweep_fit_dataset_1d_onfly(
pi2_bench,
(seq_lengths, pi2_bench.set_sequence_length, 'Gate number', ''),
*params,
fitter_arguments=fitter_arguments,
measurement_type='clifford_bench',
metadata={'qubit_ids': ','.join(qubit_ids)},
shuffle=True,
references=references)
## interleaver measurement is found, bench "interleaver" gate
references['Clifford-bench'] = clifford_bench.id
if interleaver is not None:
if 'references' in interleaver:
references.update(interleaver['references'])
pi2_bench.set_target_pulse(interleaver)
measurement_name = [m for m in pi2_bench.get_points().keys()][0]
fitter_arguments = (measurement_name, exp.exp_fitter(), 0,
np.arange(len(params)).tolist())
interleaved_bench = device.sweeper.sweep_fit_dataset_1d_onfly(
pi2_bench,
(seq_lengths, pi2_bench.set_sequence_length, 'Gate number', ''),
*params,
fitter_arguments=fitter_arguments,
measurement_type='interleaved_bench',
metadata={'qubit_ids': ','.join(qubit_ids)},
shuffle=True,
references=references)
return interleaved_bench
return clifford_bench
def benchmarking_pi2_multi(device,
qubit_ids,
*params,
interleaver=None,
two_qubit_gate=None,
max_pulses=None,
pause_length=0,
random_sequence_num=1,
seq_lengths_num=400):
channel_amplitudes_ = {}
pi2_pulses = {}
generators = {}
if max_pulses is None:
max_pulses = []
for qubit_id in qubit_ids:
coherence_measurement = Ramsey.get_Ramsey_coherence_measurement(
device, qubit_id)
T2 = float(coherence_measurement.metadata['T'])
pi2_pulses[qubit_id] = excitation_pulse.get_excitation_pulse(
device, qubit_id, np.pi / 2.)
pi2_pulse_length = float(pi2_pulses[qubit_id].metadata['length'])
max_pulses.append(T2 / pi2_pulse_length)
if two_qubit_gate is not None:
max_pulses = np.asarray(max_pulses) / 3.
max_pulses = min(max_pulses)
for qubit_id in qubit_ids:
pi2_pulses[qubit_id] = excitation_pulse.get_excitation_pulse(
device, qubit_id, np.pi / 2.)
channel_amplitudes_[qubit_id] = channel_amplitudes.channel_amplitudes(
device.exdir_db.select_measurement_by_id(
pi2_pulses[qubit_id].references['channel_amplitudes']))
seq_lengths = np.asarray(
np.round(np.linspace(0, max_pulses, seq_lengths_num)), int)
def get_pulse_seq_z(z_phase, qubit_id):
pg = device.pg
z_pulse = [(c, vz, z_phase)
for c, a in channel_amplitudes_[qubit_id].items()]
sequence_z = [pg.pmulti(0, *tuple(z_pulse))]
return sequence_z
def tensor_product(unitary, qubit_id):
U = [[1]]
for i in qubit_ids:
U = np.kron(U, np.identity(2) if i != qubit_id else unitary)
return U
qubit_readout_pulse, readout_device = calibrated_readout.get_calibrated_measurer(
device, qubit_ids)
generators = {}
for qubit_id in qubit_ids:
HZ = {
'H_' + qubit_id: {
'pulses':
get_pulse_seq_z(np.pi / 2, qubit_id) +
pi2_pulses[qubit_id].get_pulse_sequence(np.pi) +
get_pulse_seq_z(np.pi / 2, qubit_id),
'unitary':
np.sqrt(0.5) * tensor_product([[1, 1], [1, -1]], qubit_id),
'price':
1.0
},
'Z_' + qubit_id: {
'pulses': get_pulse_seq_z(np.pi, qubit_id),
'unitary': tensor_product([[1, 0], [0, -1]], qubit_id),
'price': 0.1
},
'Z/2_' + qubit_id: {
'pulses': get_pulse_seq_z(np.pi / 2, qubit_id),
'unitary': tensor_product([[1, 0], [0, 1j]], qubit_id),
'price': 0.1
},
'-Z/2_' + qubit_id: {
'pulses': get_pulse_seq_z(-np.pi / 2., qubit_id),
'unitary': tensor_product([[1, 0], [0, -1j]], qubit_id),
'price': 0.1
},
'I_' + qubit_id: {
'pulses': [],
'unitary': tensor_product([[1, 0], [0, 1]], qubit_id),
'price': 0.1
}
}
generators[qubit_id] = HZ
if len(qubit_ids) == 2:
HZ_group = clifford.two_qubit_clifford(
*tuple([g for g in generators.values()]),
plus_op_parallel=device.pg.parallel,
cphase=two_qubit_gate)
elif len(qubit_ids) == 1:
HZ_group = clifford.generate_group(generators[qubit_ids[0]])
else:
raise ValueError('More than two qubits are unsupported')
print('group length:', len(HZ_group))
ro_seq = [
device.pg.pmulti(pause_length)
] + device.trigger_readout_seq + qubit_readout_pulse.get_pulse_sequence()
pi2_bench = interleaved_benchmarking.interleaved_benchmarking(
readout_device,
set_seq=lambda x: device.pg.set_seq(device.pre_pulses + x + ro_seq),
interleavers=HZ_group)
pi2_bench.random_sequence_num = random_sequence_num
random_sequence_ids = np.arange(random_sequence_num)
references = {('pi2_pulse', qubit_id): pi2_pulses[qubit_id].id
for qubit_id in qubit_ids}
pi2_bench.prepare_random_interleaving_sequences()
### search db for previous version of the interleaver measurement
found = False
try:
clifford_bench = device.exdir_db.select_measurement(
measurement_type='clifford_bench',
metadata={'qubit_ids': ','.join(qubit_ids)},
references_that=references)
found = True
except IndexError:
pass
if random_sequence_num > 1:
params = tuple([(random_sequence_ids,
pi2_bench.set_interleaved_sequence,
'Random sequence id', '')] + [p for p in params])
if (not found) or (interleaver is None):
measurement_name = [m for m in pi2_bench.get_points().keys()][0]
fitter_arguments = (measurement_name, exp.exp_fitter(), 0,
np.arange(len(params)).tolist())
clifford_bench = device.sweeper.sweep_fit_dataset_1d_onfly(
pi2_bench,
(seq_lengths, pi2_bench.set_sequence_length_and_regenerate,
'Gate number', ''),
*params,
fitter_arguments=fitter_arguments,
measurement_type='clifford_bench',
metadata={'qubit_ids': ','.join(qubit_ids)},
shuffle=True,
references=references)
## interleaver measurement is found, bench "interleaver" gate
references['Clifford-bench'] = clifford_bench.id
if interleaver is not None:
if 'references' in interleaver:
references.update(interleaver['references'])
pi2_bench.set_target_pulse(interleaver)
measurement_name = [m for m in pi2_bench.get_points().keys()][0]
fitter_arguments = (measurement_name, exp.exp_fitter(), 0,
np.arange(len(params)).tolist())
interleaved_bench = device.sweeper.sweep_fit_dataset_1d_onfly(
pi2_bench,
(seq_lengths, pi2_bench.set_sequence_length_and_regenerate,
'Gate number', ''),
*params,
fitter_arguments=fitter_arguments,
measurement_type='interleaved_bench',
metadata={'qubit_ids': ','.join(qubit_ids)},
shuffle=True,
references=references)
return interleaved_bench
return clifford_bench
def benchmarking_pi2(device,
qubit_id,
*params,
pause_length=0,
random_sequence_num=1,
seq_lengths_num=400):
coherence_measurement = Ramsey.get_Ramsey_coherence_measurement(
device, qubit_id)
T2 = float(coherence_measurement.metadata['T'])
pi2_pulse = excitation_pulse.get_excitation_pulse(device, qubit_id,
np.pi / 2.)
pi2_pulse_length = float(pi2_pulse.metadata['length'])
pi_pulse = excitation_pulse.get_excitation_pulse(device, qubit_id, np.pi)
pi_pulse_length = float(pi_pulse.metadata['length'])
channel_amplitudes_ = channel_amplitudes.channel_amplitudes(
device.exdir_db.select_measurement_by_id(
pi2_pulse.references['channel_amplitudes']))
max_pulses = T2 / pi2_pulse_length
seq_lengths = np.asarray(
np.round(np.linspace(0, max_pulses, seq_lengths_num)), int)
def get_pulse_seq_z(z_phase, length):
fast_control = False
z_pulse = [(c, device.pg.virtual_z, z_phase * 360 / 2 / np.pi,
fast_control) for c, a in channel_amplitudes_.items()]
sequence_z = [device.pg.pmulti(device, length, *tuple(z_pulse))]
return sequence_z
#TODO
qubit_readout_pulse, readout_device = calibrated_readout.get_calibrated_measurer(
device, [qubit_id])
#HZ = {'X': {'pulses': pi_pulse.get_pulse_sequence(0), 'unitary': np.asarray([[0, 1], [1, 0]]), 'price': 1.0},
# 'X/2': {'pulses': pi2_pulse.get_pulse_sequence(0), 'unitary': np.sqrt(0.5) * np.asarray([[1, -1j], [-1j, 1]]), 'price': 1.0},
# '-X/2': {'pulses': pi2_pulse.get_pulse_sequence(np.pi), 'unitary': np.sqrt(0.5) * np.asarray([[1, 1j], [1j, 1]]), 'price': 1.0},
# 'Z': {'pulses': get_pulse_seq_z(np.pi), 'unitary': np.asarray([[1, 0], [0, -1]]), 'price':0.1},
# 'Z/2': {'pulses': get_pulse_seq_z(np.pi / 2), 'unitary': np.asarray([[1, 0], [0, 1j]]), 'price':0.1},
# '-Z/2': {'pulses': get_pulse_seq_z(-np.pi / 2.), 'unitary': np.asarray([[1, 0], [0, -1j]]), 'price':0.1},
# 'I': {'pulses': get_pulse_seq_z(0, qubit_id), 'unitary': np.asarray([[1, 0], [0, 1]]), 'price':0.1}
# }
#X_metadata = pi_pulse.metadata()
#X2_metadata = pi2_pulse.metadata()
#X_2_metadata = pi2_pulse.metadata()
#X_2_metadata['amplitude'] = float(a) * float(self.metadata['amplitude']) * np.exp(1j * phase)
HZ = {
'X': {
'pulses': pi_pulse.get_pulse_sequence(0),
'unitary': np.asarray([[0, 1], [1, 0]]),
'price': 1.0
},
'X/2': {
'pulses': pi2_pulse.get_pulse_sequence(0),
'unitary': np.sqrt(0.5) * np.asarray([[1, -1j], [-1j, 1]]),
'price': 1.0
},
'-X/2': {
'pulses': pi2_pulse.get_pulse_sequence(np.pi),
'unitary': np.sqrt(0.5) * np.asarray([[1, 1j], [1j, 1]]),
'price': 1.0
},
'Z': {
'pulses': get_pulse_seq_z(np.pi, pi2_pulse_length),
'unitary': np.asarray([[1, 0], [0, -1]]),
'price': 0.1
},
'Z/2': {
'pulses': get_pulse_seq_z(np.pi / 2, pi2_pulse_length),
'unitary': np.asarray([[1, 0], [0, 1j]]),
'price': 0.1
},
'-Z/2': {
'pulses': get_pulse_seq_z(-np.pi / 2., pi2_pulse_length),
'unitary': np.asarray([[1, 0], [0, -1j]]),
'price': 0.1
},
'I': {
'pulses': get_pulse_seq_z(0, pi2_pulse_length),
'unitary': np.asarray([[1, 0], [0, 1]]),
'price': 0.1
}
}
HZ_group = clifford.generate_group(HZ)
#TODO qubit sequencer
exitation_channel = [
i for i in device.get_qubit_excitation_channel_list(qubit_id).keys()
][0]
ex_channel = device.awg_channels[exitation_channel]
if ex_channel.is_iq():
control_seq_id = ex_channel.parent.sequencer_id
else:
control_seq_id = ex_channel.channel // 2
ex_sequencers = []
for seq_id in device.pre_pulses.seq_in_use:
if seq_id != control_seq_id:
ex_seq = zi_scripts.SIMPLESequence(sequencer_id=seq_id,
awg=device.modem.awg,
awg_amp=1,
use_modulation=True,
pre_pulses=[])
else:
ex_seq = zi_scripts.SIMPLESequence(sequencer_id=seq_id,
awg=device.modem.awg,
awg_amp=1,
use_modulation=True,
pre_pulses=[],
control=True)
control_sequence = ex_seq
device.pre_pulses.set_seq_offsets(ex_seq)
device.pre_pulses.set_seq_prepulses(ex_seq)
ex_seq.start()
ex_sequencers.append(ex_seq)
pi2_bench = interleaved_benchmarking.interleaved_benchmarking(
readout_device, ex_sequencers, HZ_group)
#pi2_bench.interleavers = HZ_group
#TODO prepare_seq
prepare_seq = pi2_bench.create_hdawg_generator()
sequence_control.set_preparation_sequence(device, ex_sequencers,
prepare_seq)
# TODO Readout sequencer
#ro_seq = [device.pg.pmulti(pause_length)]+device.trigger_readout_seq+qubit_readout_pulse.get_pulse_sequence()
readout_sequencer = sequence_control.define_readout_control_seq(
device, qubit_readout_pulse)
readout_sequencer.start()
#pi2_bench = interleaved_benchmarking.interleaved_benchmarking(readout_device,
# set_seq = lambda x: device.pg.set_seq(device.pre_pulses+x+ro_seq))
#pi2_bench.interleavers = HZ_group
pi2_bench.random_sequence_num = random_sequence_num
random_sequence_ids = np.arange(random_sequence_num)
pi2_bench.prepare_random_interleaving_sequences()
clifford_bench = device.sweeper.sweep(
pi2_bench, (seq_lengths, pi2_bench.set_sequence_length_and_regenerate,
'Gate number', ''),
*params, (random_sequence_ids, pi2_bench.set_interleaved_sequence,
'Random sequence id', ''),
shuffle=True,
measurement_type='pi2_bench',
metadata={'qubit_id': qubit_id},
references={
'pi2_pulse': pi2_pulse.id,
'pi_pulse': pi_pulse.id
})
return clifford_bench
def benchmarking_pi2(device,
qubit_id,
*params,
pause_length=0,
random_sequence_num=1,
seq_lengths_num=400):
coherence_measurement = Ramsey.get_Ramsey_coherence_measurement(
device, qubit_id)
T2 = float(coherence_measurement.metadata['T'])
pi2_pulse = excitation_pulse.get_excitation_pulse(device, qubit_id,
np.pi / 2.)
pi2_pulse_length = float(pi2_pulse.metadata['length'])
channel_amplitudes_ = channel_amplitudes.channel_amplitudes(
device.exdir_db.select_measurement_by_id(
pi2_pulse.references['channel_amplitudes']))
max_pulses = T2 / pi2_pulse_length
seq_lengths = np.asarray(
np.round(np.linspace(0, max_pulses, seq_lengths_num)), int)
def get_pulse_seq_z(z_phase):
pg = device.pg
z_pulse = [(c, vz, z_phase) for c, a in channel_amplitudes_.items()]
sequence_z = [pg.pmulti(0, *tuple(z_pulse))]
return sequence_z
qubit_readout_pulse, readout_device = calibrated_readout.get_calibrated_measurer(
device, [qubit_id])
HZ = {
'H': {
'pulses':
get_pulse_seq_z(np.pi / 2) + pi2_pulse.get_pulse_sequence(np.pi) +
get_pulse_seq_z(np.pi / 2),
'unitary':
np.sqrt(0.5) * np.asarray([[1, 1], [1, -1]]),
'price':
1.0
},
'Z': {
'pulses': get_pulse_seq_z(np.pi),
'unitary': np.asarray([[1, 0], [0, -1]]),
'price': 0.1
},
'Z/2': {
'pulses': get_pulse_seq_z(np.pi / 2),
'unitary': np.asarray([[1, 0], [0, 1j]]),
'price': 0.1
},
'-Z/2': {
'pulses': get_pulse_seq_z(-np.pi / 2.),
'unitary': np.asarray([[1, 0], [0, -1j]]),
'price': 0.1
},
'I': {
'pulses': [],
'unitary': np.asarray([[1, 0], [0, 1]]),
'price': 0.1
}
}
HZ_group = clifford.generate_group(HZ)
ro_seq = [
device.pg.pmulti(pause_length)
] + device.trigger_readout_seq + qubit_readout_pulse.get_pulse_sequence()
pi2_bench = interleaved_benchmarking.interleaved_benchmarking(
readout_device,
set_seq=lambda x: device.pg.set_seq(device.pre_pulses + x + ro_seq))
pi2_bench.interleavers = HZ_group
pi2_bench.random_sequence_num = random_sequence_num
random_sequence_ids = np.arange(random_sequence_num)
pi2_bench.prepare_random_interleaving_sequences()
clifford_bench = device.sweeper.sweep(
pi2_bench, (seq_lengths, pi2_bench.set_sequence_length_and_regenerate,
'Gate number', ''),
*params, (random_sequence_ids, pi2_bench.set_interleaved_sequence,
'Random sequence id', ''),
shuffle=True,
measurement_type='pi2_bench',
metadata={'qubit_id': qubit_id},
references={'pi2_pulse': pi2_pulse.id})
return clifford_bench