hex.glm

Class GLM.GLMGradientSolver

java.lang.Object

hex.glm.GLM.GLMGradientSolver

All Implemented Interfaces:

OptimizationUtils.GradientSolver

Enclosing class:

GLM

public static final class GLM.GLMGradientSolver
extends java.lang.Object
implements OptimizationUtils.GradientSolver

Gradient and line search computation for L_BFGS and also L_BFGS solver wrapper (for ADMM)

Constructor Summary

Constructors

Constructor and Description
GLMGradientSolver(water.Job job, GLMModel.GLMParameters glmp, DataInfo dinfo, double l2pen, GLM.BetaConstraint bc, GLM.BetaInfo bi)
GLMGradientSolver(water.Job job, GLMModel.GLMParameters glmp, DataInfo dinfo, double l2pen, GLM.BetaConstraint bc, GLM.BetaInfo bi, double[][][] penaltyMat, int[][] gamColInd)

Method Summary

Modifier and Type	Method and Description
GLM.GLMGradientInfo	getGradient(double[] beta) Evaluate ginfo at solution beta.
GLM.GLMGradientInfo	getGradient(double[] beta, ComputationState state) This method calculates the gradient for constrained GLM without taking into account the contribution of the constraints in this case.
GLM.GLMGradientInfo	getMultinomialLikelihood(double[] beta)
OptimizationUtils.GradientInfo	getObjective(double[] beta)

Methods inherited from class java.lang.Object

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

Constructor Detail

GLMGradientSolver

public GLMGradientSolver(water.Job job,
                         GLMModel.GLMParameters glmp,
                         DataInfo dinfo,
                         double l2pen,
                         GLM.BetaConstraint bc,
                         GLM.BetaInfo bi)

GLMGradientSolver

public GLMGradientSolver(water.Job job,
                         GLMModel.GLMParameters glmp,
                         DataInfo dinfo,
                         double l2pen,
                         GLM.BetaConstraint bc,
                         GLM.BetaInfo bi,
                         double[][][] penaltyMat,
                         int[][] gamColInd)

Method Detail

getMultinomialLikelihood

public GLM.GLMGradientInfo getMultinomialLikelihood(double[] beta)

getGradient

public GLM.GLMGradientInfo getGradient(double[] beta,
                                       ComputationState state)

This method calculates the gradient for constrained GLM without taking into account the contribution of the constraints in this case. The likelihood, objective are calculated without the contribution of the constraints either.

getGradient

public GLM.GLMGradientInfo getGradient(double[] beta)

Description copied from interface: OptimizationUtils.GradientSolver

Evaluate ginfo at solution beta.

Specified by:

getGradient in interface OptimizationUtils.GradientSolver

Returns:

getObjective

public OptimizationUtils.GradientInfo getObjective(double[] beta)

Specified by:

getObjective in interface OptimizationUtils.GradientSolver