BayesicFitting

Model Fitting and Evidence Calculation

View project on GitHub



class NonLinearModel( Model )Source

Anchestor of all non-linear models.

The general non-linear model needs both the function value and the partials.

It contains provisions for mixed models. (TBC)

Attributes

  • _linear : list of int
          list of indices for the linear parameters (in case of a mixed model)

Attributes from Model

     parameters, stdevs, npchain, _next, _head, _operation, xUnit, yUnit

Attributes from FixedModel

     npmax, fixed, parlist, mlist

Attributes from BaseModel

     npbase, ndim, priors, posIndex, nonZero, tiny, deltaP, parNames

Author Do Kester

NonLinearModel( nparams, ndim=1, copy=None, **kwargs )

Parent class for all non linear models.

Parameters

  • nparams : int
         number of parameters in the model
  • ndim : int
         dimensionality of the input. Default 1
  • copy : NonLinearModel
         model to be copied.
  • kwargs : dict
         Possibly includes keywords from
             FixedModel : fixed, names
             BaseModel : posIndex, nonZero

setMixedModel( lindex )
Convert a NonLinear model into a Mixed model with linear and non-linear parameters.

Reset with SetMixedModel( null );

Parameters

  • lindex : list of int
         indices of the linear parameters

isMixed( )

Returns true when linear indices have been set

getNonLinearIndex( )

Returns the index of the non-linear parameters.

partial( xdata, param=None, useNum=False )
Return the partial derivatives for the model.

Parameters

  • xdata : array_like
         the value at which to calculate the partials
  • param : array_like
         the parameters of the model. Default the self.parameters
  • useNum : boolean
         if True use numeric partial derivatives. Default False
Methods inherited from Model
Methods inherited from FixedModel
Methods inherited from BaseModel