Problem = class Problem(builtins.object) |
|
Problem(model=None, xdata=None, ydata=None, weights=None, accuracy=None, copy=None)
Problem implements the common parts of specialized Problems.
A Problem is an optimization of parameters which does not involve
the fitting of data to a Model.
Problems can be solved by NestedSampler, with appropriate Engines and
ErrorDistributions.
The result of the function for certain x and p is given by
problem.result( x, p )
The parameters, p, are to be optimized while the x provide additional
information.
This class is a base class. Further specializations will define the
result method.
Attributes
----------
model : Model
to be optimized
xdata : array_like
independent variable (static)
ydata : array_like
dependent variable (static)
weights : array_like
weights associated with ydata
varyy : float or ndarry of shape (ndata,)
Variance in the errors of the ydata
npars : int
number of parameters in the model of the problem
partype : float | int
type of the parameters
Author : Do Kester |
|
Constructor:
- XXXweightedResiduals( param, mockdata=None, extra=False )
- Returns the (weighted) residuals, calculated at the xdata.
Optionally (extra=True) the weighted signs of the residuals are returned too.
Parameters
----------
param : array_like
values for the parameters.
mockdata : array_like
model fit at xdata
extra : bool (False)
true : return ( wgt * res, wgt * sign( res ) )
false : return wgt * res
Methods defined here:
- Problem( model=None, xdata=None, ydata=None, weights=None, accuracy=None, copy=None )
- Problem Constructor.
Parameters
----------
model : Model
the model to be solved
xdata : array_like or None
independent variable
ydata : array_like or None
dependent variable
weights : array_like or None
weights associated with ydata
accuracy : float or ndarray of shape (ndata,)
accuracy scale for the datapoints
all the same or one for each data point
copy : Problem
to be copied
- baseName()
- copy()
- Copy.
- cyclicCorrection( res )
- No correction.
Returns
-------
the residuals, unadultered
Parameters
----------
res : array_like
residuals
- cyclize( res, period )
- Apply correction on residuals which are cyclic in some
phase space.
If the model results in a phase value of +ε
while the data give that phase value as (p - ε)
to keep all data in the range [0,p], the naive residual
would be (p - 2 ε) while the actual distance should
be measured the other way around as (2 ε).
Here p = period and ε = small deviation.
Parameters
----------
res : array_like
original residuals
period : float
of the phase space
Returns
-------
corrected residuals.
- cycor1( res )
- Returns the residuals, all corrected for periodicity in residuals
Parameters
----------
res : array_like
residuals
- cycor2( res )
- Returns the residuals corrected for periodicity in residuals, only
the result dimensions listed in the model.cyclic dictionary.
Parameters
----------
res : array_like
residuals
- domain2Unit( dval, kpar )
- Return value in [0,1] for the selected parameter.
Parameters
----------
dval : float
domain value for the selected parameter
kpar : array_like
selected parameter index, where kp is index in [parameters, hyperparams]
- hasWeights()
- Return whether it has weights.
- isDynamic()
- residuals( param, mockdata=None )
- Returns the residuals, calculated at the xdata.
Parameters
----------
param : array_like
values for the parameters.
mockdata : array_like
model fit at xdata
- result( param )
- Returns the result using the parameters.
In this (base)class it is a placeholder.
Parameters
----------
param : array_like
values for the parameters.
- setAccuracy( accuracy=None )
- set the value for accuracy.
Paramaters
----------
accuracy : float or array of NDATA floats or None
Either one value for all or one for each data point
When None the value is set to 0 (for easy computational reasons)
- shortName()
- Return a short version the string representation: upto first non-letter.
- unit2Domain( uval, kpar )
- Return domain value for the selected parameter.
Parameters
----------
uval : array_like
unit value for the selected parameter
kpar : array_like
selected parameter indices, where kp is index in [parameters, hyperparams]
- weightedResSq( allpars, mockdata=None, extra=False )
- Returns the (weighted) squared residuals, calculated at the xdata.
Optionally (extra=True) the weighted residuals themselves are returned too.
Parameters
----------
allpars : array_like
values for the parameters.
mockdata : array_like
model fit at xdata
extra : bool (False)
true : return ( wgt * res^2, wgt * res )
false : return wgt * res^2
|
|