net.finmath.montecarlo.interestrate.modelplugins

Class AbstractLIBORCovarianceModelParametric

java.lang.Object

net.finmath.montecarlo.interestrate.modelplugins.AbstractLIBORCovarianceModel

net.finmath.montecarlo.interestrate.modelplugins.AbstractLIBORCovarianceModelParametric

Direct Known Subclasses:

BlendedLocalVolatilityModel, LIBORCovarianceModelExponentialForm5Param, LIBORCovarianceModelExponentialForm7Param, LIBORCovarianceModelFromVolatilityAndCorrelation, LIBORCovarianceModelStochasticVolatility

public abstract class AbstractLIBORCovarianceModelParametric
extends AbstractLIBORCovarianceModel

Base class for parametric covariance models, see also AbstractLIBORCovarianceModel. Parametric models feature a parameter vector which can be inspected and modified for calibration purposes. The parameter vector may have zero length, which indicated that the model is not calibrateable. This class includes the implementation of a generic calibration algorithm. If you provide an arbitrary list of calibration products, the class can return a new instance where the parameters are chosen such that the (weighted) root-mean-square error of the difference of the value of the calibration products and given target values is minimized.

Constructor Summary

Constructors

Constructor and Description
AbstractLIBORCovarianceModelParametric(TimeDiscretizationInterface timeDiscretization, TimeDiscretizationInterface liborPeriodDiscretization, int numberOfFactors) Constructor consuming time discretizations, which are handled by the super class.

Method Summary

Modifier and Type	Method and Description
abstract java.lang.Object	clone()
AbstractLIBORCovarianceModelParametric	getCloneCalibrated(LIBORMarketModelInterface calibrationModel, AbstractLIBORMonteCarloProduct[] calibrationProducts, double[] calibrationTargetValues, double[] calibrationWeights)
AbstractLIBORCovarianceModelParametric	getCloneCalibrated(LIBORMarketModelInterface calibrationModel, AbstractLIBORMonteCarloProduct[] calibrationProducts, double[] calibrationTargetValues, double[] calibrationWeights, java.util.Map<java.lang.String,java.lang.Object> calibrationParameters) Performs a generic calibration of the parametric model by trying to match a given vector of calibration product to a given vector of target values using a given vector of weights.
AbstractLIBORCovarianceModelParametric	getCloneWithModifiedParameters(double[] parameters)
abstract double[]	getParameter() Get the parameters of determining this parametric covariance model.
abstract void	setParameter(double[] parameter)
java.lang.String	toString()

Methods inherited from class net.finmath.montecarlo.interestrate.modelplugins.AbstractLIBORCovarianceModel

getCovariance, getCovariance, getFactorLoading, getFactorLoading, getFactorLoading, getFactorLoadingPseudoInverse, getLiborPeriodDiscretization, getNumberOfFactors, getTimeDiscretization

Methods inherited from class java.lang.Object

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

Constructor Detail

AbstractLIBORCovarianceModelParametric

public AbstractLIBORCovarianceModelParametric(TimeDiscretizationInterface timeDiscretization,
                                              TimeDiscretizationInterface liborPeriodDiscretization,
                                              int numberOfFactors)

Constructor consuming time discretizations, which are handled by the super class.

Parameters:

timeDiscretization - The vector of simulation time discretization points.

liborPeriodDiscretization - The vector of tenor discretization points.

numberOfFactors - The number of factors to use (a factor reduction is performed)

Method Detail

getParameter

public abstract double[] getParameter()

Get the parameters of determining this parametric covariance model. The parameters are usually free parameters which may be used in calibration.

Returns:

Parameter vector.

setParameter

public abstract void setParameter(double[] parameter)

clone

public abstract java.lang.Object clone()

Overrides:

clone in class java.lang.Object

getCloneWithModifiedParameters

public AbstractLIBORCovarianceModelParametric getCloneWithModifiedParameters(double[] parameters)

getCloneCalibrated

public AbstractLIBORCovarianceModelParametric getCloneCalibrated(LIBORMarketModelInterface calibrationModel,
                                                                 AbstractLIBORMonteCarloProduct[] calibrationProducts,
                                                                 double[] calibrationTargetValues,
                                                                 double[] calibrationWeights)
                                                          throws CalculationException

Throws:

CalculationException

getCloneCalibrated

public AbstractLIBORCovarianceModelParametric getCloneCalibrated(LIBORMarketModelInterface calibrationModel,
                                                                 AbstractLIBORMonteCarloProduct[] calibrationProducts,
                                                                 double[] calibrationTargetValues,
                                                                 double[] calibrationWeights,
                                                                 java.util.Map<java.lang.String,java.lang.Object> calibrationParameters)
                                                          throws CalculationException

Performs a generic calibration of the parametric model by trying to match a given vector of calibration product to a given vector of target values using a given vector of weights. Optional calibration parameters may be passed using the map calibrationParameters. The keys are (Strings):

• brownianMotion: Under this key an object implementing BrownianMotionInterface may be provided. If so, this Brownian motion is used to build the valuation model.
• maxIterations: Under this key an object of type Integer may be provided specifying the maximum number of iterations.
• accuracy: Under this key an object of type Double may be provided specifying the desired accuracy. Note that this is understood in the sense that the solver will stop if the iteration does not improve by more than this number.

Parameters:

calibrationModel - The LIBOR market model to be used for calibrations (specifies forward curve and tenor discretization).

calibrationProducts - The array of calibration products.

calibrationTargetValues - The array of target values.

calibrationWeights - The array of weights.

calibrationParameters - A map of type Map<String, Object> specifying some (optional) calibration parameters.

Returns:

A new parametric model of the same type than this one, but with calibrated parameters.

Throws:

CalculationException - Thrown if calibration has failed.

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object