def __init__(self):
# we'll use a 2D Gaussian for the likelihood distribution
self.params = ['x0', 'x1']
self.mean = numpy.array([2., 5.])
self.std = numpy.array([1., 2.])
self.likelihood_dist = stats.norm(loc=self.mean, scale=self.std)
# we'll just use a uniform prior
xbnds = Boundaries((-20., 20.))
ybnds = Boundaries((-40., 40.))
self.prior_bounds = {'x0': xbnds, 'x1': ybnds}
self.prior_dist = {
'x0': stats.uniform(xbnds.lower, abs(xbnds)),
'x1': stats.uniform(ybnds.lower, abs(ybnds))
}
def __init__(self, lmbda=3):
# we'll use a Poission distribution for the likelihood
self.params = ['k']
self.likelihood_dist = stats.poisson(lmbda)
# we'll just use a uniform prior
kmin = 0
kmax = 10
self.prior_bounds = {'k': Boundaries((kmin, kmax))}
self.prior_dist = {'k': stats.randint(kmin, kmax)}
def __init__(self, phi0=0., std=1.):
self.phi0 = phi0
self.params = ['phi']
self.std = numpy.array([std])
# we'll just use a uniform prior
self.prior_bounds = {'phi': Boundaries((0., 2 * numpy.pi))}
pmin = self.prior_bounds['phi'].lower
dp = abs(self.prior_bounds['phi'])
self.prior_dist = {'phi': stats.uniform(pmin, dp)}
def __init__(self, seed=None):
# we'll use a 2D Gaussian for the likelihood distribution
self.params = ['x0', 'x1']
self.mean = numpy.array([2., 5.])
self.std = numpy.array([1., 2.])
self.likelihood_dist = stats.norm(loc=self.mean, scale=self.std)
# we'll just use a uniform prior
xbnds = Boundaries((-20., 20.))
ybnds = Boundaries((-40., 40.))
self.prior_bounds = {'x0': xbnds, 'x1': ybnds}
self.prior_dist = {
'x0': stats.uniform(xbnds.lower, abs(xbnds)),
'x1': stats.uniform(ybnds.lower, abs(ybnds))
}
# create an rng for drawing prior samples
if seed is None:
seed = MODEL_SEED
self.rng = numpy.random.default_rng(seed)
def __init__(self, lmbda=3, seed=None):
# we'll use a Poission distribution for the likelihood
self.params = ['k']
self.likelihood_dist = stats.poisson(lmbda)
# we'll just use a uniform prior
kmin = 0
kmax = 10
self.prior_bounds = {'k': Boundaries((kmin, kmax))}
self.prior_dist = {'k': stats.randint(kmin, kmax)}
# create an rng for drawing prior samples
if seed is None:
seed = MODEL_SEED
self.rng = numpy.random.default_rng(seed)
def __init__(self, phi0=0., std=1., seed=None):
self.phi0 = phi0
self.params = ['phi']
self.std = numpy.array([std])
# we'll just use a uniform prior
self.prior_bounds = {'phi': Boundaries((0., 2 * numpy.pi))}
pmin = self.prior_bounds['phi'].lower
dp = abs(self.prior_bounds['phi'])
self.prior_dist = {'phi': stats.uniform(pmin, dp)}
# create an rng for drawing prior samples
if seed is None:
seed = MODEL_SEED
self.rng = numpy.random.default_rng(seed)
def get_param_boundaries(params, opts):
"""Gets parameter boundaries for jump proposals.
The syntax for the options should be ``(min|max)_{param} = value``. Both
a minimum and maximum should be provided for every parameter in ``params``.
If the opts are created using ``load_opts``, then the options can be
formatted as ``(min|max)-{param}``, since that function will turn all ``-``
to ``_`` in option names.
Arguments will be popped from the given ``opts`` dictionary.
Parameters
----------
params : list of str
List of parameter names to get boundaries for.
opts : dict
Dictionary of option -> value that was loaded from a config file
section.
Returns
-------
dict :
Dictionary of parameter names -> :py:class:`epsie.proposals.Boundaries`
"""
boundaries = {}
for param in params:
minbound = opts.pop('min_{}'.format(param), None)
if minbound is None:
raise ValueError("Must provide a minimum bound for {p}."
"Syntax is min_{p} = val".format(p=param))
maxbound = opts.pop('max_{}'.format(param), None)
if maxbound is None:
raise ValueError("Must provide a maximum bound for {p}."
"Syntax is max_{p} = val".format(p=param))
boundaries[param] = Boundaries((float(minbound), float(maxbound)))
return boundaries