org.broadinstitute.hellbender.utils

Class OptimizationUtils

java.lang.Object

org.broadinstitute.hellbender.utils.OptimizationUtils

public final class OptimizationUtils
extends java.lang.Object

Created by davidben on 4/27/16.

Field Summary

Fields

Modifier and Type	Field and Description
protected static org.apache.commons.math3.optim.MaxEval	DEFAULT_MAX_EVAL
protected static org.apache.commons.math3.optim.univariate.BrentOptimizer	DEFAULT_OPTIMIZER

Method Summary

Modifier and Type	Method and Description
static double	argmax(java.util.function.Function<java.lang.Double,java.lang.Double> function, double min, double max, double guess)
static double	argmax(java.util.function.Function<java.lang.Double,java.lang.Double> function, double min, double max, double guess, double relativeTolerance, double absoluteTolerance, int maxEvaluations)
static double	singleNewtonArgmaxUpdate(java.util.function.Function<java.lang.Double,java.lang.Double> function, double min, double max, double guess)
static double	singleNewtonArgmaxUpdate(java.util.function.Function<java.lang.Double,java.lang.Double> function, double min, double max, double guess, double epsilon, double derivativeThreshold) One iteration of Newton's method for univariate optimization (i.e.

Methods inherited from class java.lang.Object

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

Field Detail

DEFAULT_MAX_EVAL

protected static final org.apache.commons.math3.optim.MaxEval DEFAULT_MAX_EVAL

DEFAULT_OPTIMIZER

protected static final org.apache.commons.math3.optim.univariate.BrentOptimizer DEFAULT_OPTIMIZER

Method Detail

argmax

public static double argmax(java.util.function.Function<java.lang.Double,java.lang.Double> function,
                            double min,
                            double max,
                            double guess)

argmax

public static double argmax(java.util.function.Function<java.lang.Double,java.lang.Double> function,
                            double min,
                            double max,
                            double guess,
                            double relativeTolerance,
                            double absoluteTolerance,
                            int maxEvaluations)

singleNewtonArgmaxUpdate

public static double singleNewtonArgmaxUpdate(java.util.function.Function<java.lang.Double,java.lang.Double> function,
                                              double min,
                                              double max,
                                              double guess,
                                              double epsilon,
                                              double derivativeThreshold)

One iteration of Newton's method for univariate optimization (i.e. not repeated until convergence)

Parameters:

function - the function to maximize

min - minimum allowable value

max - maximum allowable value

guess - current guess

epsilon - infinitesimal for numerical first and second derivatives

derivativeThreshold - first derivative below which current guess is considered stationary

singleNewtonArgmaxUpdate

public static double singleNewtonArgmaxUpdate(java.util.function.Function<java.lang.Double,java.lang.Double> function,
                                              double min,
                                              double max,
                                              double guess)