Navigation

Global	Documentation
Home	Manual
Readme	Glossary
Notes	Design
Examples	Classes
	Trouble
	Style

class ModelDistribution( ScaledErrorDistribution )	Source

To calculate the probability of a model M from a set of models S, given some data D, use Bayes rule

     P( M|DS ) = P( M|S ) * P( D|MS ) / P( D|S )
     posterior = prior * likelihood / evidence

This class calculates the likelihood P( D|MS ). On another level where we calculate the probability of the parameters p, we see this likelhood appear as evidence P( D|M ).

Again using Bayes

     P( p|DM ) = P( p|M ) * P( D|pM ) / P( D|M )

The evidence here is calculated as the integral over a Gausian approximation of the posterior.

Author Do Kester.

ModelDistribution( arbiter=None, scale=1.0, limits=None, copy=None, **kwargs )

Default Constructor.

Parameters

• arbiter : None or BaseFitter or str
       to provide the evidence
       None select fitter automatically
       BaseFiter Use this fitter
       str "fitter", "levenberg", "curve", "amoeba", "NestedSampler"

• scale : float
       noise scale

• limits : None or list of 2 floats [low,high]
       None : no limits implying fixed scale
       low low limit on scale (needs to be >0)
       high high limit on scale
       when limits are set, the scale is not fixed.

• copy : ModelDistribution
       distribution to be copied.

• kwargs : dict
       to be applied to arbiter

copy( )

Return copy of this.

acceptWeight( )

True if the distribution accepts weights. Always true for this distribution.

logLikelihood_alt( problem, allpars )

Return the log( likelihood ) for a Gaussian distribution.

Alternate calculation

Parameters

• problem : Problem
       to be solved
• allpars : array_like
       list of all parameters in the problem
       Return optimal parameters of the fit

logLdata( problem, allpars, mockdata=None )

Return the log( likelihood ) for each residual
    logL = sum( logLdata )

Parameters

• problem : Problem
       to be solved
• allpars : array_like
       list of all parameters in the problem
• mockdata : array_like
       as calculated by the model

Methods inherited from ScaledErrorDistribution

• setLimits( limits )

Methods inherited from ErrorDistribution

• getGaussianScale( problem, allpars=None )
• getResiduals( problem, allpars=None )
• getChisq( problem, allpars=None )
• toSigma( scale )
• isBound( )
• keepFixed( fixed=None )
• setPriors( priors )
• domain2Unit( dval, ks )
• unit2Domain( uval, ks )
• logCLhood( problem, allpars )
• logLhood( problem, allpars )
• partialLogL( problem, allpars, fitIndex )
• partialLogL_alt( problem, allpars, fitIndex )
• numPartialLogL( problem, allpars, fitIndex )
• updateLogL( problem, allpars, parval=None )
• setResult( )
• hyparname( k )