Class LBFGSArray

Implements the LBFGS compact Quasi-Newton solver on an array of Vectors (which may be sparse)

Inheritance

Object

BFGS

LBFGS

LBFGSArray

Inherited Members

LBFGS.rho

LBFGS.approxDim

LBFGS.ApproximationDimension

LBFGS.Run(Vector, Double, FunctionEval)

BFGS.lineSearch

BFGS.epsilon

BFGS.iter

BFGS.stepMax

BFGS.maxIterations

BFGS.initialStep

BFGS.cancel

BFGS.dimension

BFGS.currentObj

BFGS.debug

BFGS.linesearchDebug

BFGS.convergenceCriteria

BFGS.IterationsPerformed

BFGS.Epsilon

BFGS.MaximumStep

BFGS.InitialStep

BFGS.MaximumIterations

BFGS.Cancel

BFGS.SetWolfeConstants(Double, Double)

BFGS.TryLineSearch(Double, Double, Double)

BFGS.RaiseIterationEvent(Int32, Double, Double)

BFGS.OnIteration

Object.Equals(Object)

Object.Equals(Object, Object)

Object.GetHashCode()

Object.GetType()

Object.MemberwiseClone()

Object.ReferenceEquals(Object, Object)

Object.ToString()

Namespace: Microsoft.ML.Probabilistic.Math

Assembly: Microsoft.ML.Probabilistic.dll

Syntax

public class LBFGSArray : LBFGS

Constructors

LBFGSArray(Int32)

Constructor

Declaration

public LBFGSArray(int approxDim)

Parameters

Type	Name	Description
Int32	approxDim	Rank of the hessian approximation

Fields

currentDeriv

Declaration

protected Vector[] currentDeriv

Field Value

Type	Description
Vector[]

currentX

Declaration

protected Vector[] currentX

Field Value

Type	Description
Vector[]

ess

Declaration

protected List<Vector[]> ess

Field Value

Type	Description
List<Vector[]>

feval

Declaration

protected LBFGSArray.FunctionEvalArray feval

Field Value

Type	Description
LBFGSArray.FunctionEvalArray

searchDir

Declaration

protected Vector[] searchDir

Field Value

Type	Description
Vector[]

wye

Declaration

protected List<Vector[]> wye

Field Value

Type	Description
List<Vector[]>

Methods

LSEval(Double, Boolean, out Double)

Declaration

protected override double LSEval(double step, bool calcDeriv, out double deriv)

Parameters

Type	Name	Description
Double	step
Boolean	calcDeriv
Double	deriv

Returns

Type	Description
Double

Overrides

BFGS.LSEval(Double, Boolean, out Double)

Run(Vector[], Double, LBFGSArray.FunctionEvalArray)

Run an unconstrained minimization using LBFGS

Declaration

public Vector[] Run(Vector[] x0, double normOfFirstStep, LBFGSArray.FunctionEvalArray func)

Parameters

Type	Name	Description
Vector[]	x0	Starting value
Double	normOfFirstStep	Norm of first step
LBFGSArray.FunctionEvalArray	func	Multivariate function and derivative evaluator

Returns

Type	Description
Vector[]	The local minimum point