hep.aida.ref.pdf

Class NonParametricPdf

java.lang.Object

hep.aida.ref.pdf.Variable

hep.aida.ref.pdf.Function

hep.aida.ref.pdf.NonParametricPdf

All Implemented Interfaces:

IFunction, IModelFunction, FunctionDispatcher, HasName, HasUnits, HasValue, VariableListener

public class NonParametricPdf
extends Function

A Pdf builtg from a given data set. The Pdf is evaluated using the idea of adaptive kernel estimation presented at http://www-wisconsin.cern.ch/~cranmer/keys.html Mirroring is to be performed when distributions don't naturally taper to zero on one or both sides. With abrupt interruptions of the distribution there is a leak of probability due to the way the kernel is built (gaussian). By mirroring the distribution on the side of the abrupt edge the loss of probability is minimized.

Field Summary

Fields

Modifier and Type	Field and Description
static int	MIRROR_BOTH
static int	MIRROR_LEFT
static int	MIRROR_RIGHT
static int	NO_MIRROR

Fields inherited from class hep.aida.ref.pdf.Variable

DEPENDENT, FUNCTION, PARAMETER

Constructor Summary

Constructors

Constructor and Description
NonParametricPdf(java.lang.String name, FitData data, Dependent x)
NonParametricPdf(java.lang.String name, FitData data, Dependent x, int mirrorCode)

Method Summary

Modifier and Type	Method and Description
double	evaluateAnalyticalNormalization(Dependent dep)
double	functionValue()
boolean	hasAnalyticalNormalization(Dependent dep) FIXME Should the normalization be left to 1? or should we evaluate it numerically? The problem is the leaking of probability at the edge.
boolean	hasAnalyticalVariableGradient(Variable var) To be overwritten by classes extending Function.
void	variableChanged(Variable var) This method is invoked when a variable in the function has changed its value.

Methods inherited from class hep.aida.ref.pdf.Function

addFunctionListener, addVariable, addVariables, annotation, codeletString, dimension, evaluateAnalyticalVariableGradient, excludeNormalizationAll, functionMaxValue, functionValue, getDependent, getDependent, getNormalizationParameter, getParameter, getParameter, gradient, gradient, includeNormalizationAll, indexOfParameter, isComposite, isEqual, isNormalized, normalizationParameter, normalizationRange, normalizationRangeChanged, normalize, numberOfDependents, numberOfParameters, parameter, parameterGradient, parameterGradient, parameterNames, parameters, providesGradient, providesGradientWithRespectToVariable, providesNormalization, providesParameterGradient, removeFunctionListener, setCodeletString, setNormalizationParamter, setParameter, setParameters, setTitle, setValue, title, value, value, variableChangedName, variableChangedRange, variableChangedUnits, variableChangedValue, variableChangingName, variableChangingUnits, variableChangingValue, variableName, variableNames

Methods inherited from class hep.aida.ref.pdf.Variable

addVariableListener, name, removeVariableListener, setName, setUnits, type, units

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

NO_MIRROR

public static final int NO_MIRROR

See Also:

Constant Field Values

MIRROR_LEFT

public static final int MIRROR_LEFT

See Also:

Constant Field Values

MIRROR_RIGHT

public static final int MIRROR_RIGHT

See Also:

Constant Field Values

MIRROR_BOTH

public static final int MIRROR_BOTH

See Also:

Constant Field Values

Constructor Detail

NonParametricPdf

public NonParametricPdf(java.lang.String name,
                        FitData data,
                        Dependent x)

NonParametricPdf

public NonParametricPdf(java.lang.String name,
                        FitData data,
                        Dependent x,
                        int mirrorCode)

Method Detail

variableChanged

public void variableChanged(Variable var)

Description copied from class: Function

This method is invoked when a variable in the function has changed its value.

Overrides:

variableChanged in class Function

functionValue

public double functionValue()

Overrides:

functionValue in class Function

hasAnalyticalVariableGradient

public boolean hasAnalyticalVariableGradient(Variable var)

Description copied from class: Function

To be overwritten by classes extending Function. This method is used internally by this class to determine if a function has can provide an analytical gradient with respect to a given Variable.

Overrides:

hasAnalyticalVariableGradient in class Function

hasAnalyticalNormalization

public boolean hasAnalyticalNormalization(Dependent dep)

FIXME Should the normalization be left to 1? or should we evaluate it numerically? The problem is the leaking of probability at the edge.

Overrides:

hasAnalyticalNormalization in class Function

evaluateAnalyticalNormalization

public double evaluateAnalyticalNormalization(Dependent dep)

Overrides:

evaluateAnalyticalNormalization in class Function