breeze.optimize

ProjectedQuasiNewton

class ProjectedQuasiNewton extends FirstOrderMinimizer[DenseVector[Double], DiffFunction[DenseVector[Double]]] with Projecting[DenseVector[Double]] with SerializableLogging

Linear Supertypes
Projecting[DenseVector[Double]], FirstOrderMinimizer[DenseVector[Double], DiffFunction[DenseVector[Double]]], SerializableLogging, Serializable, Serializable, Minimizer[DenseVector[Double], DiffFunction[DenseVector[Double]]], AnyRef, Any
Ordering
  1. Alphabetic
  2. By inheritance
Inherited
  1. ProjectedQuasiNewton
  2. Projecting
  3. FirstOrderMinimizer
  4. SerializableLogging
  5. Serializable
  6. Serializable
  7. Minimizer
  8. AnyRef
  9. Any
  1. Hide All
  2. Show all
Learn more about member selection
Visibility
  1. Public
  2. All

Instance Constructors

  1. new ProjectedQuasiNewton(tolerance: Double = 1e-6, m: Int = 10, initFeas: Boolean = false, testOpt: Boolean = true, maxIter: Int = 1, maxSrchIt: Int = 50, gamma: Double = 1e-4, projection: (DenseVector[Double]) ⇒ DenseVector[Double] = identity)(implicit space: MutableInnerProductModule[DenseVector[Double], Double])

Type Members

  1. type BDV = DenseVector[Double]

  2. type History = CompactHessian

    Any history the derived minimization function needs to do its updates.

    Any history the derived minimization function needs to do its updates. typically an approximation to the second derivative/hessian matrix.

    Definition Classes
    ProjectedQuasiNewton → FirstOrderMinimizer

  3. case class State(x: T, value: Double, grad: T, adjustedValue: Double, adjustedGradient: T, iter: Int, initialAdjVal: Double, history: History, fVals: IndexedSeq[Double] = Vector(Double.PositiveInfinity), numImprovementFailures: Int = 0, searchFailed: Boolean = false) extends Product with Serializable

    Tracks the information about the optimizer, including the current point, its value, gradient, and then any history.

    Tracks the information about the optimizer, including the current point, its value, gradient, and then any history. Also includes information for checking convergence.

    x

    the current point being considered

    value

    f(x)

    grad

    f.gradientAt(x)

    adjustedValue

    f(x) + r(x), where r is any regularization added to the objective. For LBFGS, this is f(x).

    adjustedGradient

    f'(x) + r'(x), where r is any regularization added to the objective. For LBFGS, this is f'(x).

    iter

    what iteration number we are on.

    initialAdjVal

    f(x_0) + r(x_0), used for checking convergence

    history

    any information needed by the optimizer to do updates.

    fVals

    the sequence of the last minImprovementWindow values, used for checking if the "value" isn't improving

    numImprovementFailures

    the number of times in a row the objective hasn't improved, mostly for SGD

    searchFailed

    did the line search fail?

    Definition Classes
    FirstOrderMinimizer

Value Members

  1. final def !=(arg0: Any): Boolean

    Definition Classes
    AnyRef → Any

  2. final def ##(): Int

    Definition Classes
    AnyRef → Any

  3. final def ==(arg0: Any): Boolean

    Definition Classes
    AnyRef → Any

  4. def adjust(newX: DenseVector[Double], newGrad: DenseVector[Double], newVal: Double): (Double, DenseVector[Double])

    Attributes
    protected
    Definition Classes
    ProjectedQuasiNewton → FirstOrderMinimizer

  5. def adjustFunction(f: DiffFunction[DenseVector[Double]]): DiffFunction[DenseVector[Double]]

    Attributes
    protected
    Definition Classes
    FirstOrderMinimizer

  6. final def asInstanceOf[T0]: T0

    Definition Classes
    Any

  7. def calculateObjective(f: DiffFunction[DenseVector[Double]], x: DenseVector[Double], history: History): (Double, DenseVector[Double])

    Attributes
    protected
    Definition Classes
    FirstOrderMinimizer

  8. def chooseDescentDirection(state: State, fn: DiffFunction[DenseVector[Double]]): DenseVector[Double]

    Attributes
    protected
    Definition Classes
    ProjectedQuasiNewton → FirstOrderMinimizer

  9. def clone(): AnyRef

    Attributes
    protected[java.lang]
    Definition Classes
    AnyRef
    Annotations
    @throws( ... )

  10. def determineStepSize(state: State, f: DiffFunction[DenseVector[Double]], dir: DenseVector[Double]): Double

    Given a direction, perform a Strong Wolfe Line Search

    Given a direction, perform a Strong Wolfe Line Search

    TO DO: Compare performance with Cubic Interpolation based line search from Mark's PQN paper

    state

    the current state

    f

    The objective

    dir

    The step direction

    returns

    stepSize

    Attributes
    protected
    Definition Classes
    ProjectedQuasiNewton → FirstOrderMinimizer

  11. final def eq(arg0: AnyRef): Boolean

    Definition Classes
    AnyRef

  12. def equals(arg0: Any): Boolean

    Definition Classes
    AnyRef → Any

  13. def finalize(): Unit

    Attributes
    protected[java.lang]
    Definition Classes
    AnyRef
    Annotations
    @throws( classOf[java.lang.Throwable] )

  14. val gamma: Double

  15. final def getClass(): Class[_]

    Definition Classes
    AnyRef → Any

  16. def hashCode(): Int

    Definition Classes
    AnyRef → Any

  17. def infiniteIterations(f: DiffFunction[DenseVector[Double]], state: State): Iterator[State]

    Definition Classes
    FirstOrderMinimizer

  18. val initFeas: Boolean

  19. def initialHistory(f: DiffFunction[DenseVector[Double]], init: DenseVector[Double]): History

    Attributes
    protected
    Definition Classes
    ProjectedQuasiNewton → FirstOrderMinimizer

  20. def initialState(f: DiffFunction[DenseVector[Double]], init: DenseVector[Double]): State

    Attributes
    protected
    Definition Classes
    FirstOrderMinimizer

  21. val innerOptimizer: SpectralProjectedGradient[BDV]

  22. final def isInstanceOf[T0]: Boolean

    Definition Classes
    Any

  23. def iterations(f: DiffFunction[DenseVector[Double]], init: DenseVector[Double]): Iterator[State]

    Definition Classes
    FirstOrderMinimizer

  24. def logger: LazyLogger

    Attributes
    protected
    Definition Classes
    SerializableLogging

  25. val m: Int

  26. val maxSrchIt: Int

  27. val minImprovementWindow: Int

    How many iterations to improve function by at least improvementTol

    How many iterations to improve function by at least improvementTol

    Definition Classes
    FirstOrderMinimizer

  28. def minimize(f: DiffFunction[DenseVector[Double]], init: DenseVector[Double]): DenseVector[Double]

    Definition Classes
    FirstOrderMinimizer → Minimizer

  29. def minimizeAndReturnState(f: DiffFunction[DenseVector[Double]], init: DenseVector[Double]): State

    Definition Classes
    FirstOrderMinimizer

  30. final def ne(arg0: AnyRef): Boolean

    Definition Classes
    AnyRef

  31. final def notify(): Unit

    Definition Classes
    AnyRef

  32. final def notifyAll(): Unit

    Definition Classes
    AnyRef

  33. val numberOfImprovementFailures: Int

    Definition Classes
    FirstOrderMinimizer

  34. def projectedVector(x: DenseVector[Double], g: DenseVector[Double])(implicit vspace: Module[DenseVector[Double], Double]): DenseVector[Double]

    Definition Classes
    Projecting

  35. val projection: (DenseVector[Double]) ⇒ DenseVector[Double]

    Definition Classes
    ProjectedQuasiNewton → Projecting

  36. final def synchronized[T0](arg0: ⇒ T0): T0

    Definition Classes
    AnyRef

  37. def takeStep(state: State, dir: DenseVector[Double], stepSize: Double): DenseVector[Double]

    Attributes
    protected
    Definition Classes
    ProjectedQuasiNewton → FirstOrderMinimizer

  38. val testOpt: Boolean

  39. def toString(): String

    Definition Classes
    AnyRef → Any

  40. def updateFValWindow(oldState: State, newAdjVal: Double): IndexedSeq[Double]

    Attributes
    protected
    Definition Classes
    FirstOrderMinimizer

  41. def updateHistory(newX: DenseVector[Double], newGrad: DenseVector[Double], newVal: Double, f: DiffFunction[DenseVector[Double]], oldState: State): History

    Attributes
    protected
    Definition Classes
    ProjectedQuasiNewton → FirstOrderMinimizer

  42. final def wait(): Unit

    Definition Classes
    AnyRef
    Annotations
    @throws( ... )

  43. final def wait(arg0: Long, arg1: Int): Unit

    Definition Classes
    AnyRef
    Annotations
    @throws( ... )

  44. final def wait(arg0: Long): Unit

    Definition Classes
    AnyRef
    Annotations
    @throws( ... )

Inherited from Projecting[DenseVector[Double]]

Inherited from FirstOrderMinimizer[DenseVector[Double], DiffFunction[DenseVector[Double]]]

Inherited from SerializableLogging

Inherited from Serializable

Inherited from Serializable

Inherited from Minimizer[DenseVector[Double], DiffFunction[DenseVector[Double]]]

Inherited from AnyRef

Inherited from Any

Ungrouped