def create_gaussian_prior_for(parameter, mean, std): """ Use for informed variables. The Gaussian prior weights by a Gaussian in the parameter. If you would like the logarithm of the parameter to be weighted by a Gaussian, create a ancillary parameter (see create_jeffreys_prior_for). :param parameter: Parameter to create a prior for. E.g. xspowerlaw.mypowerlaw.PhoIndex """ import invgauss lo = parameter.min hi = parameter.max f = invgauss.get_invgauss_func(mean, std) return lambda x: min(lo, max(hi, f(x)))
def create_gaussian_prior_for(parameter, mean, std):
"""
Use for informed variables.
The Gaussian prior weights by a Gaussian in the parameter. If you
would like the logarithm of the parameter to be weighted by
a Gaussian, create a ancillary parameter (see create_jeffreys_prior_for).
:param parameter: Parameter to create a prior for. E.g. xspowerlaw.mypowerlaw.PhoIndex
"""
import invgauss
lo = parameter.min
hi = parameter.max
f = invgauss.get_invgauss_func(mean, std)
return lambda x: max(lo, min(hi, f(x)))
# creating ancillary parameters for logarithmic treatment
print 'creating prior functions...'
from sherpa.models.parameter import Parameter
srclevel = Parameter('src', 'level', numpy.log10(src.norm.val), -8, -1, -8, -1)
srcnh = Parameter('src', 'nh', numpy.log10(abso.nH.val) + 22, 19, 24, 19, 24)
galnh = Parameter('gal', 'nh',
numpy.log10(galabso.nH.val) + 22, 19, 24, 19, 24)
src.norm = 10**srclevel
abso.nH = 10**(srcnh - 22)
galabso.nH = 10**(galnh - 22)
# example of a custom prior
import invgauss
f = invgauss.get_invgauss_func(numpy.log10(props['nhgal']), 0.15)
def limited_19_24(x):
v = f(x)
if v <= 19:
v = 19
if v >= 24:
v = 24
return v
priors = []
parameters = [srclevel, src.PhoIndex, srcnh, galnh]
import bxa.sherpa as bxa
galabso.nH.val = props['nhgal'] / 1e22
# creating ancillary parameters for logarithmic treatment
print('creating prior functions...')
from sherpa.models.parameter import Parameter
srclevel = Parameter('src', 'level', numpy.log10(src.norm.val), -8, -1, -8, -1)
srcnh = Parameter('src', 'nh', numpy.log10(abso.nH.val)+22, 19, 24, 19, 24)
galnh = Parameter('gal', 'nh', numpy.log10(galabso.nH.val)+22, 19, 24, 19, 24)
src.norm = 10**srclevel
abso.nH = 10**(srcnh - 22)
galabso.nH = 10**(galnh - 22)
# example of a custom prior
import invgauss
f = invgauss.get_invgauss_func(numpy.log10(props['nhgal']), 0.15)
def limited_19_24(x):
v = f(x)
if v <= 19:
v = 19
if v >= 24:
v = 24
return v
priors = []
parameters = [srclevel, src.PhoIndex, srcnh, galnh]
import bxa.sherpa as bxa
priors += [bxa.create_uniform_prior_for(srclevel)]
priors += [bxa.create_uniform_prior_for(src.PhoIndex)]
