def map_loss_trial(self,
true_map_,
SAMPLE_GLOBAL_MODEL,
measurements_controls_=None,
autocontrol_="ON",
var_thres_=1.0):
'''Return an error vector for map reconstruction from one trial of an algorithm.'''
self.qslamobj = qs.ParticleFilter(SAMPLE_GLOBAL_MODEL)
self.qslamobj.QubitGrid.engineeredtruemap = true_map_
self.qslamobj.qslamr(
measurements_controls=measurements_controls_,
autocontrol=autocontrol_,
max_num_iterations=SAMPLE_GLOBAL_MODEL["MODELDESIGN"]
["MAX_NUM_ITERATIONS"],
var_thres=var_thres_)
posterior_map = self.qslamobj.QubitGrid.get_all_nodes(["f_state"])
posterior_corrs = self.qslamobj.QubitGrid.get_all_nodes(["r_state"])
map_residuals = self.loss(posterior_map, true_map_)
controlpath = self.qslamobj.QubitGrid.control_sequence
return posterior_map, map_residuals, posterior_corrs, controlpath
def qslam_trial(self,
measurements_controls_=None,
autocontrol_="ON",
var_thres_=1.0):
'''
Return posterior map, posterior lengthscales and control path for one
QSLAM run.
Parameters:
-----------
measurements_controls_ (`float` | dims: 2):
External input for the most recent control directive and measurement
outcome for the qubit, denoted by locaton index, node_j.
Default value: None.
autocontrol_ (`dtype` | string binary - "ON", "OFF"):
"OFF" - next qubit measured is specified as a user input via measurement_controls_
"ON" - next qubit measured is chosen by the algorithm.
Default value: "ON".
var_thres [NOT USED]:
Error variance threshold where if the variance
of length scales is less than interqubit separation, then algorithm
terminates.
Returns:
--------
posterior_f_state (`dtype` | numpy array):
Posterior dephasing noise field estimate at each qubit on grid in
vectorised form.
Dims: len(self.GLOBALDICT["GRIDDICT"])
posterior_r_state (`dtype` | numpy array):
Posterior correlation length estimate at each qubit on grid in
vectorised form.
Dims: len(self.GLOBALDICT["GRIDDICT"])
control_sequence (`dtype` | list ):
List of control actions (qubits to measure) for every iteration step of qslam.
Initialised as an empty list.
'''
qslamobj = qs.ParticleFilter(copy.deepcopy(self.GLOBALDICT),
real_data=True,
real_data_key=self.data_key)
qslamobj.qslamr(max_num_iterations=self.GLOBALDICT["MODELDESIGN"]
["MAX_NUM_ITERATIONS"],
measurements_controls=measurements_controls_,
autocontrol=autocontrol_,
var_thres=var_thres_)
posterior_f_state = qslamobj.QubitGrid.get_all_nodes(["f_state"])
posterior_r_state = qslamobj.QubitGrid.get_all_nodes(["r_state"])
control_sequence = qslamobj.QubitGrid.control_sequence
return posterior_f_state, posterior_r_state, control_sequence
def qslam_trial(self, measurements_controls_=None,
autocontrol_="ON",
var_thres_=1.0 ):
qslamobj = qs.ParticleFilter(copy.deepcopy(self.GLOBALDICT), real_data=True, real_data_key=self.data_key)
qslamobj.qslamr(max_num_iterations=self.GLOBALDICT["MODELDESIGN"]["MAX_NUM_ITERATIONS"],
measurements_controls=measurements_controls_,
autocontrol=autocontrol_,
var_thres=var_thres_)
posterior_map = qslamobj.QubitGrid.get_all_nodes(["f_state"])
posterior_corrs = qslamobj.QubitGrid.get_all_nodes(["r_state"])
controlpath = qslamobj.QubitGrid.control_sequence
return posterior_map, posterior_corrs, controlpath
def call_qslam(self):
qslamobj = 0
qslamobj = qs.ParticleFilter(
self.SAMPLE_GLOBAL_MODEL,
save_run=True,
beta_expansion_mode=self.beta_expansion_mode,
beta_skew_adjust=self.beta_skew_adjust)
qslamobj.QubitGrid.engineeredtruemap = self.true_map_
qslamobj.qslamr(
measurements_controls=self.measurements_controls_,
autocontrol=self.autocontrol_,
max_num_iterations=self.SAMPLE_GLOBAL_MODEL["MODELDESIGN"]
["MAX_NUM_ITERATIONS"],
var_thres=self.var_thres_)
return qslamobj
def map_loss_trial(self,
true_map_,
SAMPLE_GLOBAL_MODEL,
measurements_controls_=None,
autocontrol_="ON",
var_thres_=1.0):
'''Return posterior f_state, posterior r_state, residuals, and
control path for map reconstruction from one trial of QSLAM.
Parameters:
-----------
true_map_ (`dtype`| numpy array ):
A set of phase values associated to each qubit location in an arrangement,
where each phase take a value between [0, np.pi].
SAMPLE_GLOBAL_MODEL (`dtype`| numpy array ):
Dictionary object as given in qslamdesignparams.py that sets
parameters to conduct an empirical risk analysis and configure QSLAM.
measurements_controls_ (`dtype`| numpy array ):
A list containing a measurement set and the control
directive of the location of the measured qubit. Each control directive
is a single iteration of the algorithm.
autocontrol_ (`dtype`| string ):
"OFF" - next qubit measured is specified as a user input via measurement_controls_
"ON" - next qubit measured is chosen by the algorithm.
Default value: "OFF".
var_thres_ (`dtype`| numpy array ):
Error variance threshold where if the variance
of length scales is less than interqubit separation, then algorithm
terminates.
Returns:
-------
posterior_map (`type` | numpy array ):
Posterior f_state at the end of a QSLAM filtering run.
map_residuals (`type` | numpy array ):
Difference between posterior f_state and a true map at the end
of a QSLAM filtering run.
posterior_corrs (`type` | numpy array ):
Posterior r_state at the end of a QSLAM filtering run.
controlpath (`type` | list ):
List of qubit locations at which a physical measurement was performed.
A single element of this list corresponds to one iteration of QSLAM;
length of list output at the end of a QSLAM filtering run matches
total number of algorithm iterations.
'''
self.qslamobj = qs.ParticleFilter(SAMPLE_GLOBAL_MODEL)
self.qslamobj.QubitGrid.engineeredtruemap = true_map_
self.qslamobj.qslamr(
measurements_controls=measurements_controls_,
autocontrol=autocontrol_,
max_num_iterations=SAMPLE_GLOBAL_MODEL["MODELDESIGN"]
["MAX_NUM_ITERATIONS"],
var_thres=var_thres_)
posterior_map = self.qslamobj.QubitGrid.get_all_nodes(["f_state"])
posterior_corrs = self.qslamobj.QubitGrid.get_all_nodes(["r_state"])
map_residuals = self.loss(posterior_map, true_map_)
controlpath = self.qslamobj.QubitGrid.control_sequence
return posterior_map, map_residuals, posterior_corrs, controlpath