def fit_polychord(self,
n_live_points=1000,
n_chords=None,
basename='polychains/1-',
**kwargs):
import pypolychord
self._pchord_basename = basename
if hasattr(self, 'which'): self.n_dim = 9 + 6 * self.lc.n_planets
else: self.n_dim = 5 + 6 * self.lc.n_planets
if n_chords is not None:
self.n_chords = n_chords
else:
self.n_chords = 3 * self.n_dim
pypolychord.run(self.pchord_loglike,
self.pchord_prior,
self.n_dim,
n_live=n_live_points,
n_chords=self.n_chords,
output_basename=self._pchord_basename,
**kwargs)
def fit_polychord(self, n_live_points=1000, n_chords=None, basename="polychains/1-", **kwargs):
import pypolychord
self._pchord_basename = basename
if hasattr(self, "which"):
self.n_dim = 9 + 6 * self.lc.n_planets
else:
self.n_dim = 5 + 6 * self.lc.n_planets
if n_chords is not None:
self.n_chords = n_chords
else:
self.n_chords = 3 * self.n_dim
pypolychord.run(
self.pchord_loglike,
self.pchord_prior,
self.n_dim,
n_live=n_live_points,
n_chords=self.n_chords,
output_basename=self._pchord_basename,
**kwargs
)
prior_array = np.append(model.pmin, model.pmax)
def chord_like(ndim, theta, phi):
# check prior
if model.mark3LogPrior(theta) != -np.inf:
return loglike(theta, **loglkwargs)
else:
#print 'WARNING: Prior returns -np.inf!!'
return -np.inf
n_live = 1000
ndim = len(p0)
pypolychord.run(chord_like, ndim, prior_array, n_live=n_live, n_chords=5, \
output_basename=args.outDir+"/pcord-")
if args.sampler == 'multinest':
print('WARNING: Using MultiNest, will use uniform priors on all parameters')
import pymultinest
p0 = model.initParameters(startEfacAtOne=True, fixpstart=False)
# mark2 loglike
if args.incJitterEquad:
ndim = len(p0)
def myloglike(cube, ndim, nparams):
from __future__ import print_function
import numpy
from numpy import cos, pi
import pypolychord
def loglikelihood(params):
return (numpy.prod(cos(params / 2)) + 2)**5
def prior(pars):
return (4 + 1) * pi * pars
if __name__ == "__main__":
ndim = 2
import os
if not os.path.exists('chains'): os.mkdir('chains')
if not os.path.exists('chains/clusters'): os.mkdir('chains/clusters')
pypolychord.run(loglikelihood, prior, ndim, n_live=500, n_chords=1, output_basename='chains/eggbox-')
from __future__ import print_function
import numpy
from numpy import cos, pi
import pypolychord
def loglikelihood(params):
return (numpy.prod(cos(params / 2)) + 2)**5
def prior(pars):
return (4 + 1) * pi * pars
if __name__ == "__main__":
ndim = 2
import os
if not os.path.exists('chains'): os.mkdir('chains')
if not os.path.exists('chains/clusters'): os.mkdir('chains/clusters')
pypolychord.run(loglikelihood,
prior,
ndim,
n_live=500,
n_chords=1,
output_basename='chains/eggbox-')
prior_array = np.append(model.pmin, model.pmax)
def chord_like(ndim, theta, phi):
# check prior
if model.mark3LogPrior(theta) != -np.inf:
return loglike(theta, **loglkwargs)
else:
#print 'WARNING: Prior returns -np.inf!!'
return -np.inf
n_live = 1000
ndim = len(p0)
pypolychord.run(chord_like, ndim, prior_array, n_live=n_live, n_chords=5, \
output_basename=args.outDir+"/pcord-")
if args.sampler == 'multinest':
print 'WARNING: Using MultiNest, will use uniform priors on all parameters'
import pymultinest
p0 = model.initParameters(startEfacAtOne=True, fixpstart=False)
# mark2 loglike
if args.incJitterEquad:
ndim = len(p0)
def myloglike(cube, ndim, nparams):