Module net.finmath.lib
Package net.finmath.montecarlo.conditionalexpectation

Class MonteCarloConditionalExpectationRegressionLocalizedOnDependents

java.lang.Object
net.finmath.montecarlo.conditionalexpectation.MonteCarloConditionalExpectationRegression
net.finmath.montecarlo.conditionalexpectation.MonteCarloConditionalExpectationRegressionLocalizedOnDependents
All Implemented Interfaces:
ConditionalExpectationEstimator
public class MonteCarloConditionalExpectationRegressionLocalizedOnDependents extends MonteCarloConditionalExpectationRegression
A service that allows to estimate conditional expectation via regression. This implementation uses a localization weight derived from the dependent variable. The regression only considers sample paths where \( - M < y_{i} < M \) where M is a multiple of the standard deviation of y. In oder to estimate the conditional expectation, basis functions have to be specified. The class can either estimate and predict the conditional expectation within the same simulation (which will eventually introduce a small foresight bias) or use a different simulation for estimation (using basisFunctionsEstimator) to predict conditional expectation within another simulation (using basisFunctionsPredictor). In the latter case, the basis functions have to correspond to the same entities, however, generated in different simulations (number of path, etc., may be different).
Version:
1.0
Author:
Christian Fries

  • Constructor Details

    • MonteCarloConditionalExpectationRegressionLocalizedOnDependents

      public MonteCarloConditionalExpectationRegressionLocalizedOnDependents(RandomVariable[] basisFunctionsEstimator, RandomVariable[] basisFunctionsPredictor, double standardDeviations)
      Creates a class for conditional expectation estimation.
      Parameters:
      basisFunctionsEstimator - A vector of random variables to be used as basis functions for estimation.
      basisFunctionsPredictor - A vector of random variables to be used as basis functions for prediction.
      standardDeviations - A standard deviation parameter for the weight function.

    • MonteCarloConditionalExpectationRegressionLocalizedOnDependents

      public MonteCarloConditionalExpectationRegressionLocalizedOnDependents(RandomVariable[] basisFunctionsEstimator, double standardDeviations)
      Creates a class for conditional expectation estimation.
      Parameters:
      basisFunctionsEstimator - A vector of random variables to be used as basis functions for estimation.
      standardDeviations - A standard deviation parameter for the weight function.

    • MonteCarloConditionalExpectationRegressionLocalizedOnDependents

      public MonteCarloConditionalExpectationRegressionLocalizedOnDependents()

    • MonteCarloConditionalExpectationRegressionLocalizedOnDependents

      public MonteCarloConditionalExpectationRegressionLocalizedOnDependents(RandomVariable[] basisFunctions)
      Creates a class for conditional expectation estimation.
      Parameters:
      basisFunctions - A vector of random variables to be used as basis functions.

    • MonteCarloConditionalExpectationRegressionLocalizedOnDependents

      public MonteCarloConditionalExpectationRegressionLocalizedOnDependents(RandomVariable[] basisFunctionsEstimator, RandomVariable[] basisFunctionsPredictor)
      Creates a class for conditional expectation estimation.
      Parameters:
      basisFunctionsEstimator - A vector of random variables to be used as basis functions for estimation.
      basisFunctionsPredictor - A vector of random variables to be used as basis functions for prediction.

  • Method Details

    • getLinearRegressionParameters

      public double[] getLinearRegressionParameters(RandomVariable dependents)
      Return the solution x of XTX x = XT y for a given y.
      Overrides:
      getLinearRegressionParameters in class MonteCarloConditionalExpectationRegression
      Parameters:
      dependents - The sample vector of the random variable y.
      Returns:
      The solution x of XTX x = XT y.
      To dos:
      Performance upon repeated call can be optimized by caching XTX.