from collections.abc import Iterable
import warnings
import numbers
import numpy as np
from qutip.qobj import Qobj
import qutip.control.pulseoptim as cpo
from qutip.operators import identity
from qutip.tensor import tensor
from qutip.mesolve import mesolve
from qutip.qip.circuit import QubitCircuit
from qutip.qip.device.processor import Processor
from qutip.qip.operations.gates import gate_sequence_product
__all__ = ['OptPulseProcessor']
[docs]class OptPulseProcessor(Processor):
"""
A processor, which takes the Hamiltonian available
as dynamic generators, calls the
`qutip.control.optimize_pulse_unitary` function
to find an optimized pulse sequence for the desired quantum circuit.
The processor can simulate the evolution under the given
control pulses using :func:`qutip.mesolve`.
(For attributes documentation, please
refer to the parent class :class:`.Processor`)
Parameters
----------
N: int
The number of component systems.
drift: `:class:`qutip.Qobj`
The drift Hamiltonian. The size must match the whole quantum system.
t1: list or float
Characterize the decoherence of amplitude damping for
each qubit. A list of size `N` or a float for all qubits.
t2: list of float
Characterize the decoherence of dephasing for
each qubit. A list of size `N` or a float for all qubits.
dims: list
The dimension of each component system.
Default value is a
qubit system of ``dim=[2,2,2,...,2]``
"""
def __init__(self, N, drift=None, t1=None, t2=None, dims=None):
super(OptPulseProcessor, self).__init__(
N, t1=t1, t2=t2, dims=dims)
if drift is not None:
self.add_drift(drift, list(range(N)))
self.spline_kind = "step_func"
[docs] def load_circuit(self, qc, min_fid_err=np.inf, merge_gates=True,
setting_args=None, verbose=False, **kwargs):
"""
Find the pulses realizing a given
:class:`.QubitCircuit` using
`qutip.control.optimize_pulse_unitary`. Further parameter for
for `qutip.control.optimize_pulse_unitary` needs to be given as
keyword arguments. By default, it first merge all the gates
into one unitary and then find the control pulses for it.
It can be turned off and one can set different parameters
for different gates. See examples for details.
Examples
--------
Same parameter for all the gates:
.. code-block:: python
qc = QubitCircuit(N=1)
qc.add_gate("SNOT", 0)
num_tslots = 10
evo_time = 10
processor = OptPulseProcessor(N=1, drift=sigmaz(),
ctrls=[sigmax()])
# num_tslots and evo_time are two keyword arguments
tlist, coeffs = processor.load_circuit(
qc, num_tslots=num_tslots, evo_time=evo_time)
Different parameters for different gates:
.. code-block:: python
qc = QubitCircuit(N=2)
qc.add_gate("SNOT", 0)
qc.add_gate("SWAP", targets=[0, 1])
qc.add_gate('CNOT', controls=1, targets=[0])
processor = OptPulseProcessor(N=2, drift=tensor([sigmaz()]*2))
processor.add_control(sigmax(), cyclic_permutation=True)
processor.add_control(sigmay(), cyclic_permutation=True)
processor.add_control(tensor([sigmay(), sigmay()]))
setting_args = {"SNOT": {"num_tslots": 10, "evo_time": 1},
"SWAP": {"num_tslots": 30, "evo_time": 3},
"CNOT": {"num_tslots": 30, "evo_time": 3}}
tlist, coeffs = processor.load_circuit(qc,
setting_args=setting_args,
merge_gates=False)
Parameters
----------
qc: :class:`.QubitCircuit` or list of Qobj
The quantum circuit to be translated.
min_fid_err: float, optional
The minimal fidelity tolerance, if the fidelity error of any
gate decomposition is higher, a warning will be given.
Default is infinite.
merge_gates: boolean, optimal
If True, merge all gate/Qobj into one Qobj and then
find the optimal pulses for this unitary matrix. If False,
find the optimal pulses for each gate/Qobj.
setting_args: dict, optional
Only considered if merge_gates is False.
It is a dictionary containing keyword arguments
for different gates.
E.g.: ::
setting_args = {"SNOT": {"num_tslots": 10, "evo_time": 1},
"SWAP": {"num_tslots": 30, "evo_time": 3},
"CNOT": {"num_tslots": 30, "evo_time": 3}}
verbose: boolean, optional
If true, the information for each decomposed gate
will be shown. Default is False.
**kwargs
keyword arguments for `qutip.control.optimize_pulse_unitary`
Returns
-------
tlist: array_like
A NumPy array specifies the time of each coefficient
coeffs: array_like
A 2d NumPy array of the shape (len(ctrls), len(tlist)-1). Each
row corresponds to the control pulse sequence for
one Hamiltonian.
Notes
-----
``len(tlist) - 1 = coeffs.shape[1]`` since tlist gives the beginning
and the end of the pulses.
"""
if setting_args is None:
setting_args = {}
if isinstance(qc, QubitCircuit):
props = qc.propagators()
gates = [g.name for g in qc.gates]
elif isinstance(qc, Iterable):
props = qc
gates = None # using list of Qobj, no gates name
else:
raise ValueError(
"qc should be a "
"QubitCircuit or a list of Qobj")
if merge_gates: # merge all gates/Qobj into one Qobj
props = [gate_sequence_product(props)]
gates = None
time_record = [] # a list for all the gates
coeff_record = []
last_time = 0. # used in concatenation of tlist
for prop_ind, U_targ in enumerate(props):
U_0 = identity(U_targ.dims[0])
# If qc is a QubitCircuit and setting_args is not empty,
# we update the kwargs for each gate.
# keyword arguments in setting_arg have priority
if gates is not None and setting_args:
kwargs.update(setting_args[gates[prop_ind]])
full_drift_ham = self.drift.get_ideal_qobjevo(self.dims).cte
full_ctrls_hams = [pulse.get_ideal_qobj(self.dims)
for pulse in self.pulses]
result = cpo.optimize_pulse_unitary(
full_drift_ham, full_ctrls_hams, U_0, U_targ, **kwargs)
if result.fid_err > min_fid_err:
warnings.warn(
"The fidelity error of gate {} is higher "
"than required limit. Use verbose=True to see"
"the more detailed information.".format(prop_ind))
time_record.append(result.time[1:] + last_time)
last_time += result.time[-1]
coeff_record.append(result.final_amps.T)
if verbose:
print("********** Gate {} **********".format(prop_ind))
print("Final fidelity error {}".format(result.fid_err))
print("Final gradient normal {}".format(
result.grad_norm_final))
print("Terminated due to {}".format(result.termination_reason))
print("Number of iterations {}".format(result.num_iter))
tlist = np.hstack([[0.]] + time_record)
for i in range(len(self.pulses)):
self.pulses[i].tlist = tlist
coeffs = np.vstack([np.hstack(coeff_record)])
self.coeffs = coeffs
return tlist, coeffs