def __init__(self, sampler=None, params={}):
''' Define default parameters '''
super().__init__(sampler)
self.name = 'Online'
self.params['Presampled points'] = Parameter(
name='Presampled points',
value=15,
type=int,
description='Pre-sampled points before starting regression')
self.params['Iterations'] = Parameter(
name='Iterations',
type=int,
value=10,
description='Number of sample/fit cycles')
self.params['Batch size'] = Parameter(
name='Batch size',
value=10,
type=int,
description='Points sampled per iteration')
self.params['Tolerance'] = Parameter(
name='Tolerance',
value=0.01,
min=1e-4,
max=1e-1,
description='Relative tolerance required for convergence')
self.params['Mode'] = Parameter(
name='Mode',
value=['Optimizer', 'Explorer', 'Hybrid'],
type=str,
description='Peak-seeking behavior.')
for p in params:
self.params[p].value = params[p]
def __init__(self,
params={},
state=None,
bounds=None,
cost=None,
substate=None):
super().__init__()
self.params = {}
self.params['Population'] = Parameter(
name='Population',
value=20,
type=int,
description='Initial population size')
self.params['Tolerance'] = Parameter(
name='Tolerance', value=0.01, description='Convergence criterion')
self.params['Mutation'] = Parameter(
name='Population size',
value=1,
min=0,
max=2,
description=
'Mutation constant in [0,2]. Larger values increase search radius but slow down convergence.'
)
self.params['Recombination'] = Parameter(
name='Recombination',
value=0.7,
min=0,
max=1,
description=
'Crossover probability. Increasing this value allows a larger number of mutants to progress into the next generation, but at the risk of population stability.'
)
for p in params:
self.params[p].value = params[p]
super().__init__(state, bounds, cost, substate)
def __init__(self):
''' Define default parameters '''
self.name = 'PID'
self.params = {}
self.params['Proportional gain'] = Parameter(
name='Proportional gain',
value=1,
min=0,
max=100,
description='Proportional gain')
self.params['Integral gain'] = Parameter(name='Integral gain',
value=0.1,
min=0,
max=10,
description='Integral gain')
self.params['Derivative gain'] = Parameter(
name='Derivative gain',
value=0,
min=0,
max=100,
description='Derivative gain')
self.params['Sign'] = Parameter(name='Sign',
value=-1,
min=-1,
max=1,
description='Sign of correction')
def __init__(self, sampler=None, params={}):
''' Define default parameters '''
super().__init__(sampler)
self.name = 'Grid'
self.params['Steps'] = Parameter(name= 'Steps',
value = 20,
type = int,
description = 'Grid points per dimension')
self.params['Sweeps'] = Parameter(name= 'Sweeps',
value = 1,
type = int,
description = 'Number of sweeps to do')
for p in params:
self.params[p].value = params[p]
def __init__(self, params={}, state=None, bounds=None, cost=None, substate=None):
super().__init__()
self.params = {}
self.params['Tolerance'] = Parameter(name= 'Tolerance',
value = 0.01,
description = 'Convergence criterion')
for p in params:
self.params[p].value = params[p]
super().__init__(state, bounds, cost, substate)
def __init__(self):
super().__init__('GaussianProcess')
self.params['Amplitude'] = Parameter(name= 'Kernel amplitude',
value = 1,
min = 0,
max = 10,
description = 'Amplitude of modeled cost landscape')
self.params['Length scale'] = Parameter(name= 'Kernel length scale',
value = 1,
min = 0,
max = 10,
description = 'Characteristic size of cost landscape')
self.params['Noise'] = Parameter(name= 'Kernel noise',
value = 0.1,
min = 0,
max = 10,
description = 'Amplitude of modeled white noise process')
kernel = C(self.params['Amplitude'].value, (1e-3, 1e3)) * RBF(self.params['Length scale'].value, (1e-2, 1e2)) + WhiteKernel(self.params['Noise'].value)
self.model = GaussianProcessRegressor(kernel=kernel, n_restarts_optimizer=10)
self.extension = '.gp'
def __init__(self, name=None):
self.name = name
self.params = {}
self.params['Leash'] = Parameter(
name='Leash',
value=0.25,
min=0.01,
max=0.25,
description='Allowed search range relative to last best point')
self.imported = False
self.extension = None
def __init__(self,
params={},
state=None,
bounds=None,
cost=None,
substate=None):
self.params = {}
self.params['Steps'] = Parameter(
name='Steps',
type=int,
value=20,
description='Grid points per dimension')
self.params['Sweeps'] = Parameter(name='Sweeps',
value=1,
type=int,
description='Number of sweeps to do')
for p in params:
self.params[p].value = params[p]
super().__init__(state, bounds, cost, substate)
def __init__(self, params={}):
super().__init__()
self.params = {}
self.params['Optimizer'] = Parameter(name='Optimizer',
value=self.list_optimizers(),
type=str)
for p in params:
self.params[p].value = params[p]
if p == 'Optimizer':
self.optimizer = params['Optimizer']
self.optimizer.measure = self.measure
def __init__(self,
params={},
state=None,
bounds=None,
cost=None,
substate=None):
super().__init__()
''' Define default parameters '''
self.name = 'Adam'
self.params = {}
self.params['Learning rate'] = Parameter(
name='Learning rate',
value=0.1,
description='Scaling factor for steps along the gradient')
self.params['Tolerance'] = Parameter(
name='Tolerance',
value=0.01,
description=
'Convergence is detected when the gradient is this fraction of the max gradient'
)
self.params['Dither size'] = Parameter(
name='Dither size',
value=0.01,
description='Step size for gradient estimation')
self.params['Beta_1'] = Parameter(
name='Beta_1',
value=0.9,
min=0.8,
max=0.99,
description='Exponential decay rate for first-moment estimates')
self.params['Beta_2'] = Parameter(
name='Beta_2',
value=0.999,
min=0.9,
max=0.999,
description='Exponential decay rate for second-moment estimates')
self.params['Noise'] = Parameter(
name='Noise',
value=0,
description='Noise injection for stochastic optimization')
for p in params:
self.params[p].value = params[p]
super().__init__(state, bounds, cost, substate)
def __init__(self,
params={},
state=None,
bounds=None,
cost=None,
substate=None):
super().__init__()
self.params = {}
self.params['Steps'] = Parameter(name='Steps', type=int, value=10)
self.params['Particles'] = Parameter(name='Particles',
type=int,
value=10)
self.params['Velocity scale'] = Parameter(
name='Initial velocity scale', value=1)
self.params['Inertia'] = Parameter(name='Inertia', value=1)
self.params['Cognitive acceleration'] = Parameter(
name='Cognitive acceleration', value=0.1)
self.params['Social acceleration'] = Parameter(
name='Social acceleration', value=0.1)
for p in params:
self.params[p].value = params[p]
super().__init__(state, bounds, cost, substate)
def __init__(self, params={}):
super().__init__()
self.params = {'Threshold': Parameter(name='Threshold', value=0.5)}
for p in params:
self.params[p].value = params[p]
def __init__(self, params={'Loops': 1}):
super().__init__()
self.params = {}
self.params['Loops'] = Parameter(name='Loops', value=1, type=int)
for p in params:
self.params[p].value = params[p]