def __init__(self, model_function, setting_values, parameter_samples,
constants):
OptBayesExpt.__init__(self, model_function, setting_values,
parameter_samples, constants)
self.VALUES_PER_MEASUREMENT = 2
self.cost_of_changing_setting = 5.
self.noise_parameter_index = 3
def __init__(self, model_function, setting_values, parameter_samples,
constants):
OptBayesExpt.__init__(self, model_function, setting_values,
parameter_samples, constants)
self.noise_parameter_index = 3
self.sweep_settings = setting_values[0]
self.cost_of_new_sweep = 5.
self.start_stop_subsample = 3
self.start_stop_indices = self._generate_start_stop_indices()
self.start_stop_values = self.sweep_settings[self.start_stop_indices]
def setup():
pars = (np.array([0, 1, 2, 3]), np.array([1, 3, 2, 4]))
settings = (np.array([0, 1, 2]), )
cons = ()
def fakefunc(sets, pars, cons):
x, = sets
a, b = pars
return a + b * x
an_obe = OptBayesExpt(fakefunc, settings, pars, cons)
return an_obe
def __init__(self):
OptBayesExpt.__init__(self)
b_min = -1
b_max = 1
b_draws = np.random.uniform(b_min, b_max, n_particles)
# Pack the parameters into a tuple and incorporate it
pars = (x0_draws, a_draws, b_draws)
# note that the parameter order must correspond to how the
# values are unpacked in the model_function
# constant parameters -- just linewidth here.
dtrue = 0.15
cons = (dtrue, )
# Settings, parameters, constants and model all defined, so set it all up
myOBE = OptBayesExpt(my_model_function,
sets,
pars,
cons,
n_draws=30,
scale=False)
initial_pars = myOBE.parameters
#################################################################
# Measurement simulation
#################################################################
# secret stuff - to be used only by the measurement simulator
# pick the parameters of the true resonance
x0true = 2.5 # pick a random resonance x0
Atrue = 1
Btrue = 0
truevals = (x0true, Atrue, Btrue, dtrue)
b_samples = rng.normal(b_mean, b_sigma, n_samples)
# Pack the parameters into a tuple.
# Note that the order must correspond to how the values are unpacked in
# the model_function.
parameters = (x0_samples, a_samples, b_samples)
param_labels = ['Center', 'Amplitude', 'Background']
# Define Constants
#
dtrue = .1
constants = (dtrue, )
# make an instance of OptBayesExpt
#
my_obe = OptBayesExpt(my_model_function,
settings,
parameters,
constants,
scale=False,
use_jit=use_jit)
########################################################################
# MEASUREMENT LOOP
########################################################################
# Set up measurement simulator
#
# Randomly select "true" parameter values for simulation
true_pars = tuple([np.random.choice(param) for param in parameters])
# MeasurementSimulator.simdata() provides simulated data
# See optbayesexpt/obe_utils.py.
my_sim = MeasurementSimulator(my_model_function,
true_pars,
from optbayesexpt import OptBayesExpt
# Global variables
# number of measurement iterations to simulate
Nmeasure = 100
# use optbayesexpt (False uses least-squares fitting
smartmeasure = True
# optimum = True --> always measure at the maximum utility
# optimum = False --> select a high value of utility using the pickiness factor
optimum = False
pickiness = 6
"""
Create an instance of the OptBayesExpt class for our use
"""
myOBE = OptBayesExpt()
"""
Establish the experimental model -- a Lorentzian peak
"""
def lorentz_peak(x, x0, A, B, d):
"""
Calculate a Lorentzian function of x
All parameters may be scalars or they may be arrays
- as long as the arrays interact nicely
:param x: measurement setting
:param A: Amplitude parameter
:param B: background parameter
:param d: half-width at half-max parameter
:param x0: peak center value parameter
def __init__(self, model_function, setting_values, parameter_samples,
constants):
OptBayesExpt.__init__(self, model_function, setting_values,
parameter_samples, constants)
# identify the measurement noise parameter.
self.noise_parameter_index = 3
"""
Pi pulse tuner
"""
import numpy as np
import matplotlib.pyplot as plt
from optbayesexpt import OptBayesExpt
# make the instance that we'll use
myOBE = OptBayesExpt()
"""
Establish the experimental model
"""
def mxplusb(x, m, b):
# a straight line
return m * x + b
def mxplsub_wrapper(sets, pars, cons):
# unpack the experimental settings
x = sets[0]
# unpack model parameters
m = pars[0]
b = pars[1]
# unpack model constants
# N/A
return mxplusb(x, m, b)
"""
Pi pulse tuner
"""
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
from optbayesexpt import OptBayesExpt
myOBE = OptBayesExpt()
"""
ESTABLISH THE EXPERIMENTAL MODEL
"""
# this is the model of our experiment
def rabicounts(pulsetime, delta_f, B1, f_center, baseline, contrast, T1):
"""
A model of Rabi chevrons
:param pulsetime: Duration of microwave pulse
:param delta_f: detuning relative to a reference frequency
:param B1: Rabi frequency
:param f_center: true center of resonance relative to reference frequency
:param baseline: background signal
:param contrast: Maximum fractional change in signal for pi pulse
:param T1: coherence time
:return:
"""
zz = ((delta_f - f_center) / B1)**2
def __init__(self, model_function, settings, parameters, constants):
OptBayesExpt.__init__(self, model_function, settings,
parameters, constants)
a_draws = np.random.exponential(a_scale, n_particles) # background b_min = -1 b_max = 1 b_draws = np.random.uniform(b_min, b_max, n_particles) # Pack the parameters into a tuple and incorporate it pars = (x0_draws, a_draws, b_draws) # note that the parameter order must correspond to how the # values are unpacked in the model_function # constant parameters -- just linewidth here. dtrue = 0.15 cons = (dtrue, ) # Settings, parameters, constants and model all defined, so set it all up myOBE = OptBayesExpt(my_model_function, sets, pars, cons) myOBE.N_DRAWS = 30 initial_pars = myOBE.parameters ################################################################# # Measurement simulation ################################################################# # secret stuff - to be used only by the measurement simulator # pick the parameters of the true resonance x0true = 2.5 # pick a random resonance x0 Atrue = 1 Btrue = 0 truevals = (x0true, Atrue, Btrue, dtrue)