org.akutan.optimization
Class BayesSteinSolver

java.lang.Object
  org.akutan.optimization.ActiveSetSolver
      org.akutan.optimization.ReturnSolver
          org.akutan.optimization.BayesSteinSolver

All Implemented Interfaces:

Solver

public class BayesSteinSolver
extends ReturnSolver
implements Solver

This class encapsulates an example of shrinking the returns and covariance using Bayes-Stein shrinkage as descried in Jorion, 1986. NOTE: As near as I can tell this code is correct, however I don't see any improvements in the results when using this algorithm.

Since:

24 February 2006

Field Summary

protected java.text.DecimalFormat	df
protected cern.colt.matrix.DoubleMatrix1D	orig_Er
protected cern.colt.matrix.DoubleMatrix2D	orig_V

Constructor Summary

BayesSteinSolver(cern.colt.matrix.DoubleMatrix1D e_r, cern.colt.matrix.DoubleMatrix2D V_p)
Constructs

Method Summary

protected java.lang.String	formatOutput(int ct, double ra, cern.colt.matrix.DoubleMatrix1D x, cern.colt.matrix.DoubleMatrix1D e_r, cern.colt.matrix.DoubleMatrix2D V) Override method in ActiveSet so that we can use the un-shrunk E(r) and V
protected SolvedPoint	makePoint(double ra, cern.colt.matrix.DoubleMatrix1D x, cern.colt.matrix.DoubleMatrix1D e_r, cern.colt.matrix.DoubleMatrix2D V) Override method in ActiveSet so that we can use the un-shrunk E(r) and V
protected void	shrink() This is the method that actually does the shrinking of the mean and covariance and replaces the original inputs with the shrunk values.

Methods inherited from class org.akutan.optimization.ReturnSolver

cleanConstraints, initialGuess, solve, solveForReturn

Methods inherited from class org.akutan.optimization.ActiveSetSolver

algorithm, dumpConstraints, generateConstraints, generateValues, initialGuess, make2D, setMatrix, setMatrix, solve, solveSingle

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.akutan.optimization.Solver

solve

Field Detail

df

protected final java.text.DecimalFormat df

orig_Er

protected cern.colt.matrix.DoubleMatrix1D orig_Er

orig_V

protected cern.colt.matrix.DoubleMatrix2D orig_V

Constructor Detail

BayesSteinSolver

public BayesSteinSolver(cern.colt.matrix.DoubleMatrix1D e_r,
                        cern.colt.matrix.DoubleMatrix2D V_p)

Constructs

Method Detail

shrink

protected void shrink()

This is the method that actually does the shrinking of the mean and covariance and replaces the original inputs with the shrunk values.

makePoint

protected SolvedPoint makePoint(double ra,
                                cern.colt.matrix.DoubleMatrix1D x,
                                cern.colt.matrix.DoubleMatrix1D e_r,
                                cern.colt.matrix.DoubleMatrix2D V)

Override method in ActiveSet so that we can use the un-shrunk E(r) and V

Overrides:

makePoint in class ActiveSetSolver

formatOutput

protected java.lang.String formatOutput(int ct,
                                        double ra,
                                        cern.colt.matrix.DoubleMatrix1D x,
                                        cern.colt.matrix.DoubleMatrix1D e_r,
                                        cern.colt.matrix.DoubleMatrix2D V)

Override method in ActiveSet so that we can use the un-shrunk E(r) and V

Overrides:

formatOutput in class ActiveSetSolver