set_prior

Synopsis

Set the prior function to use with a parameter.

Syntax

set_prior(par, prior)

par - a `Parameter` instance
prior - function or sherpa.models.model.Model instance

Description

The default prior used by `get_draws` for each parameter is flat, varying between the soft minimum and maximum values of the parameter (as given by the `min` and `max` attributes of the parameter object). The `set_prior` function is used to change the form of the prior for a parameter, and `get_prior` returns the current prior for a parameter.

Examples

Example 1

Set the prior for the kt parameter of the therm component to be a gaussian, centered on 1.7 keV and with a FWHM of 0.35 keV:

>>> create_model_component('xsapec', 'therm')
>>> create_model_component('gauss1d', 'p_temp')
>>> p_temp.pos = 1.7
>>> p_temp.fwhm = 0.35
>>> set_prior(therm.kT, p_temp)

Example 2

Create a function ( lognorm ) and use it as the prior of the nh parameter of the abs1 model component:

>>> def lognorm(x):
...     nh = 20
...     sigma = 0.5  # use a sigma of 0.5
...     # nH is in units of 10^-22 so convert
...     dx = np.log10(x) + 22 - nh
...     norm = sigma / np.sqrt(2 * np.pi)
...     return norm * np.exp(-0.5 * dx * dx / (sigma * sigma))
...
>>> create_model_component('xsphabs', 'abs1')
>>> set_prior(abs1.nH, lognorm)

PARAMETERS

The parameters for this function are:

Parameter	Definition
par	A parameter of a model instance.
prior	The function to use for a prior. It must accept a single argument and return a value of the same size as the input.

Bugs

See the bugs pages on the Sherpa website for an up-to-date listing of known bugs.