dev.nm.stat.covariance.nlshrink

Class TauEstimator

java.lang.Object

dev.nm.stat.covariance.nlshrink.TauEstimator

public class TauEstimator
extends Object

Non-linear shrinkage estimator of population eigenvalues.

See Also:

"Olivier Ledoit and Michael Wolf, "Equation 43, Section 10.2, Numerical Implementation of the QuEST Function", 2016."

Constructor Summary

Constructors

Constructor and Description
TauEstimator(int p, int n, Vector sample_lambdas)
TauEstimator(int p, int n, Vector sample_lambdas, int max_optimizer_iterations, double epsilon)

Method Summary

Modifier and Type	Method and Description
Vector	estimatedPopulationEigenvalues(Vector initials) Estimates population eigenvalues from given sample eigenvalues.

Methods inherited from class java.lang.Object

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

Constructor Detail

TauEstimator

public TauEstimator(int p,
                    int n,
                    Vector sample_lambdas)

TauEstimator

public TauEstimator(int p,
                    int n,
                    Vector sample_lambdas,
                    int max_optimizer_iterations,
                    double epsilon)

Method Detail

estimatedPopulationEigenvalues

public Vector estimatedPopulationEigenvalues(Vector initials)

Estimates population eigenvalues from given sample eigenvalues. The primary search uses SQP. If singular QuEST derivatives prevent that search, a projected, backtracking search minimizes the same objective within the same eigenvalue bounds. Failure of both searches is reported with the original cause rather than returning the starting estimate.

Parameters:

initials - start values for the iterative search

Returns:

population eigenvalues in ascending order