Package opennlp.tools.ml.maxent.quasinewton

Class LineSearch.LineSearchResult

java.lang.Object
opennlp.tools.ml.maxent.quasinewton.LineSearch.LineSearchResult
Enclosing class:
LineSearch
public static class LineSearch.LineSearchResult extends Object
Represents a LineSearch result encapsulating the relevant data at a point in time during computation.

  • Constructor Details

    • LineSearchResult

      public LineSearchResult(double stepSize, double valueAtCurr, double valueAtNext, double[] gradAtCurr, double[] gradAtNext, double[] currPoint, double[] nextPoint, int fctEvalCount)
      Initializes a LineSearch.LineSearchResult object with the specified parameters.

    • LineSearchResult

      public LineSearchResult(double stepSize, double valueAtCurr, double valueAtNext, double[] gradAtCurr, double[] gradAtNext, double[] pseudoGradAtNext, double[] currPoint, double[] nextPoint, double[] signVector, int fctEvalCount)
      Initializes a LineSearch.LineSearchResult object with the specified parameters.

  • Method Details

    • setAll

      public void setAll(double stepSize, double valueAtCurr, double valueAtNext, double[] gradAtCurr, double[] gradAtNext, double[] currPoint, double[] nextPoint, int fctEvalCount)
      Updates line search elements.

    • setAll

      public void setAll(double stepSize, double valueAtCurr, double valueAtNext, double[] gradAtCurr, double[] gradAtNext, double[] pseudoGradAtNext, double[] currPoint, double[] nextPoint, double[] signVector, int fctEvalCount)
      Updates line search elements.

    • getFuncChangeRate

      public double getFuncChangeRate()

    • getStepSize

      public double getStepSize()

    • setStepSize

      public void setStepSize(double stepSize)

    • getValueAtCurr

      public double getValueAtCurr()

    • setValueAtCurr

      public void setValueAtCurr(double valueAtCurr)

    • getValueAtNext

      public double getValueAtNext()

    • setValueAtNext

      public void setValueAtNext(double valueAtNext)

    • getGradAtCurr

      public double[] getGradAtCurr()

    • setGradAtCurr

      public void setGradAtCurr(double[] gradAtCurr)

    • getGradAtNext

      public double[] getGradAtNext()

    • setGradAtNext

      public void setGradAtNext(double[] gradAtNext)

    • getPseudoGradAtNext

      public double[] getPseudoGradAtNext()

    • setPseudoGradAtNext

      public void setPseudoGradAtNext(double[] pseudoGradAtNext)

    • getCurrPoint

      public double[] getCurrPoint()

    • setCurrPoint

      public void setCurrPoint(double[] currPoint)

    • getNextPoint

      public double[] getNextPoint()

    • setNextPoint

      public void setNextPoint(double[] nextPoint)

    • getSignVector

      public double[] getSignVector()

    • setSignVector

      public void setSignVector(double[] signVector)

    • getFctEvalCount

      public int getFctEvalCount()

    • setFctEvalCount

      public void setFctEvalCount(int fctEvalCount)

    • getInitialObject

      public static LineSearch.LineSearchResult getInitialObject(double valueAtX, double[] gradAtX, double[] x)
      Initial linear search object for L1-regularization.
      Parameters:
      valueAtX - The value at x.
      gradAtX - The gradient at x.
      x - The input double[] vector.
      Returns:
      The LineSearch.LineSearchResult holding the results.

    • getInitialObjectForL1

      public static LineSearch.LineSearchResult getInitialObjectForL1(double valueAtX, double[] gradAtX, double[] pseudoGradAtX, double[] x)
      Initial linear search object for L1-regularization.
      Parameters:
      valueAtX - The value at x.
      gradAtX - The gradient at x.
      pseudoGradAtX - The pseudo-gradient at x.
      x - The input double[] vector.
      Returns:
      The LineSearch.LineSearchResult holding the results.

    • getInitialObject

      public static LineSearch.LineSearchResult getInitialObject(double valueAtX, double[] gradAtX, double[] pseudoGradAtX, double[] x, double[] signX, int fctEvalCount)
      Initial linear search object for L1-regularization.
      Parameters:
      valueAtX - The value at x.
      gradAtX - The gradient at x.
      pseudoGradAtX - The pseudo-gradient at x.
      x - The input double[] vector.
      signX - The sign double[] vector for x.
      fctEvalCount - The number of function evaluations. Must be equal to or greater than 0.
      Returns:
      The LineSearch.LineSearchResult holding the results.