def test_multiple_does_not_change(self):
sdm = SparseDM(3)
sdm.hadamard(0)
sdm.hadamard(1)
sdm.ensure_dense(0)
sdm.ensure_dense(1)
sdm.ensure_dense(2)
before = sdm.full_dm.to_array()
assert before.shape == (8, 8)
sdm.peak_multiple_measurements([0, 1, 2])
assert np.allclose(before, sdm.full_dm.to_array())
def test_multiple_measurement_hadamard_on_classical(self):
sdm = SparseDM(2)
sdm.hadamard(0)
meas = sdm.peak_multiple_measurements([0, 1])
assert len(meas) == 2
assert meas == [({0: 0, 1: 0}, 0.5), ({0: 1, 1: 0}, 0.5)]
def test_multiple_measurement_classical_prob(self):
sdm = SparseDM(3)
sdm.hadamard(0)
sdm.hadamard(1)
sdm.classical_probability = 0.7
r = sdm.peak_multiple_measurements([0, 1, 2])
total_prob = sum(prob for outcome, prob in r)
assert total_prob == sdm.trace()
def test_multiple_measurement_gs(self):
sdm = SparseDM(3)
sdm.ensure_dense(0)
sdm.ensure_dense(1)
sdm.ensure_dense(2)
meas = sdm.peak_multiple_measurements([0, 1, 2])
assert len(meas) == 8
for state, p in meas:
if state[0] == 0 and state[1] == 0 and state[2] == 0:
assert np.allclose(p, 1)
else:
assert np.allclose(p, 0)
def test_multiple_measurement_hadamard_order2_regression(self):
sdm = SparseDM(3)
sdm.hadamard(0)
sdm.hadamard(1)
sdm.ensure_dense(0)
sdm.ensure_dense(1)
sdm.ensure_dense(2)
meas = sdm.peak_multiple_measurements([0, 1, 2])
assert len(meas) == 8
for state, p in meas:
print(meas)
if state[2] == 0:
assert np.allclose(p, 0.25)
else:
assert np.allclose(p, 0)
def test_multiple_measurement_hadamard_order1(self):
sdm = SparseDM(3)
sdm.hadamard(0)
sdm.hadamard(2)
sdm.ensure_dense(0)
sdm.ensure_dense(1)
sdm.ensure_dense(2)
meas = sdm.peak_multiple_measurements([0, 1, 2])
assert len(meas) == 8
for state, p in meas:
for x in state.values():
assert x in [0, 1]
if state[1] == 0:
assert np.allclose(p, 0.25)
else:
assert np.allclose(p, 0)
class Controller:
def __init__(self,
filename=None,
setup=None,
random_state=None,
seed=None,
qubits=None,
circuits=None,
circuit_lists=None,
adjust_gates=None,
measurement_gates=None,
angle_convert_matrices=None,
mbits=None):
"""
qubits: list of qubits in the experiment
mbits: the set of bits to receive measurements
circuits: dictionary of circuits to apply to a state
to perform an operation.
circuit_lists: corresponding circuit lists (as output by qsoverlay
builder for saving purposes)
msmt_circuits: dictionary of circuits to measure a state.
adjust_gates: a list of adjustable gates in a circuit
that the user might pass parameters to when they run
the circuit.
measurement_gates: a set of Measurement type operators
to extract extra details about the measurements made.
"""
self.circuits = circuits or {}
self.circuit_lists = circuit_lists or {}
if 'record' in self.circuits:
raise ValueError('record is a protected keyword')
if any([type(c) == int for c in self.circuits]):
raise ValueError('Circuits must not use integers for labels')
self.mbits = mbits or []
self.qubits = qubits or []
self.adjust_gates = adjust_gates or {}
self.angle_convert_matrices = angle_convert_matrices or {}
self.measurement_gates = measurement_gates or {}
self.state = None
if filename is not None:
self.load(filename, setup, random_state, seed)
self.make_state()
def load(self, filename, setup, random_state=None, seed=None):
with open(filename, 'r') as infile:
data = json.load(infile)
if type(setup) == str:
setup = Setup(filename=setup, state=random_state, seed=seed)
self.mbits = data['mbits']
self.qubits = data['qubits']
self.angle_convert_matrices = {
key: np.array(val)
for key, val in data['angle_convert_matrices'].items()
}
b = Builder(setup)
for name, cl in data['circuit_lists'].items():
b.new_circuit()
adjust_gates = b.add_circuit_list(cl)
b.finalize()
self.circuits[name] = b.circuit
self.circuit_lists[name] = cl
self.adjust_gates[name] = [
ag for ag in adjust_gates
if type(ag) is not Measurement]
self.measurement_gates[name] = [
ag for ag in adjust_gates
if type(ag) is Measurement]
if name in data['angle_convert_matrices']:
self.angle_convert_matrices = \
data['angle_convert_matrices'][name]
def save(self, filename):
data = {
'mbits': self.mbits,
'qubits': self.qubits,
'circuit_lists': self.circuit_lists,
'angle_convert_matrices': {
key: val.tolist()
for key, val in self.angle_convert_matrices.items()}
}
with open(filename, 'w') as outfile:
json.dump(data, outfile)
def make_state(self, dense_qubits=None):
self.state = SparseDM(self.qubits + self.mbits)
if dense_qubits is not None:
for qubit in dense_qubits:
self.state.ensure_dense(qubit)
def apply_circuit(self, circuit):
"""
Applies a circuit to the state.
Each entry is either:
a) a string corresponding to an entry in self.circuits
b) a tuple or list with the first entry a string corresponding
to an entry in self.circuits and the remaining entries
angles to input for self.adjust_gates.
c) a tuple or list with the first entry the reserved keyword
'record', and the remaining entries a list of mbits to
store the current value of.
d) a pair (n, circuit), where circuit is one of the above,
and n is the number of times to repeat the circuit.
"""
# Record is a reserved keyword to copy the output
# from a set of classical bits to return to the user.
if type(circuit) is list or type(circuit) is tuple:
if circuit[0] == 'record':
output = []
for mbit in circuit[1:]:
output.append(self.state.classical[mbit])
return output
op_name = circuit[0]
if type(op_name) == int:
return_data = []
for _ in range(op_name):
return_data.append(self.apply_circuit(circuit[1]))
return return_data
else:
if op_name in self.angle_convert_matrices:
angles = self.angle_convert_matrices[op_name] @ circuit[1:]
else:
angles = circuit[1:]
for gate, param in zip(
self.adjust_gates[op_name], angles):
if not isinstance(param, tuple):
param = (param,)
gate.adjust(*param)
self.circuits[op_name].apply_to(self.state,
apply_all_pending=False)
else:
op_name = circuit
self.circuits[op_name].apply_to(self.state,
apply_all_pending=False)
if op_name in self.measurement_gates:
return_data = [{
'probabilities': m.probabilities[-1],
'projects': m.projects[-1],
'measurements': m.measurements[-1]}
for m in self.measurement_gates[op_name]]
return return_data
return None
def __lt__(self, circuit):
self.apply_circuit(circuit)
def apply_circuit_list(self, circuit_list):
"""
Apply a set of operations to a state.
circuit_list: a list of circuits to apply.
"""
output_list = []
for circuit in circuit_list:
# Record is a reserved keyword to copy the output
# from a set of classical bits to return to the user.
output = self.apply_circuit(circuit)
if output is not None:
output_list.append(output)
return output_list
def simulate_tomo(self, circuit,
tomo_circuits,
measurement_model,
num_measurements,
output_format, data_type):
"""
Takes the current system state, copies it,
applies multiple tomography circuits, and runs them through
models for the measurement to return thresholded voltages.
"""
data = []
for tomo_circuit in tomo_circuits:
self.make_state()
self < circuit
self < tomo_circuit
self.state.renormalize()
rho_dist = self.state.peak_multiple_measurements(
measurement_model.qubits)
data.append(measurement_model.sample(
rho_dist, num_measurements, data_type=data_type,
output_format=output_format))
return data
def get_expectation_values(self, msmts, num_repetitions=None):
"""
Measures a set of Pauli strings on the current state.
If num_repetitions is None this is performed perfectly.
Otherwise, we treat the calculated value as a bernoulli
random variable we are attempting to approximate, and
sample an answer.
input: msmts: list of measurement dictionaries, containing
'X', 'Y', or 'Z' for each non-trivial qubit label.
"""
results = []
self.state.apply_all_pending()
self.state.renormalize()
dm = self.state.full_dm.to_array()
for msmt in msmts:
mult = 1
# Make Pauli list
pauli_list = [1] * len(self.state.idx_in_full_dm)
for qubit in msmt:
if qubit not in self.state.idx_in_full_dm:
if msmt[qubit] == 'Z':
mult *= (-1) ** self.state.classical[qubit]
elif msmt[qubit] in ['X', 'Y']:
mult = 0
else:
raise ValueError('qubit measurements must be X, Y, Z')
else:
pauli_list[self.state.idx_in_full_dm[qubit]] = msmt[qubit]
# Make measurement operator
op = pauli_dic[pauli_list[0]]
for label in pauli_list[1:]:
op = np.kron(pauli_dic[label], op)
result = mult * np.trace(op @ dm)
assert np.imag(result) < 1e-9
result = float(np.real(result))
if num_repetitions is not None:
bernoulli_rv = (1 - result) / 2
if -1e-6 < bernoulli_rv <= 0:
noisy_result = 0
elif 1 <= bernoulli_rv < 1 + 1e-6:
noisy_result = 1
else:
try:
noisy_result = np.random.beta(bernoulli_rv *
num_repetitions,
(1 - bernoulli_rv) *
num_repetitions)
except Exception:
raise ValueError(
'My bernoulli random variable is weird: {}'
.format(bernoulli_rv))
results.append(1 - 2 * noisy_result)
else:
results.append(result)
return np.array(results)
def get_prob_all_zero(self, qubits):
"""
Returns the probability that all qubits in qubits
will return a 0 measurement (regardless of any other qubit
measurement).
"""
self.state.apply_all_pending()
self.state.renormalize()
indices = [self.state.idx_in_full_dm[q] for q in qubits]
diagonal = self.state.full_dm.get_diag()
return sum([x for j, x in enumerate(diagonal) if
all([(j // 2 ** i) % 2 == 0 for i in indices])])
def test_multiple_measurement_only_classical(self):
sdm = SparseDM(2)
meas = sdm.peak_multiple_measurements([0])
assert meas == [({0: 0}, 1)]