SpectralProjectedGradient

Instance Constructors

new SpectralProjectedGradient(projection: (T) ⇒ T = (t: T) => t, tolerance: Double = 1e-6, suffDec: Double = 1e-4, minImprovementWindow: Int = 30, alphaMax: Double = 1e10, alphaMin: Double = 1e-10, bbMemory: Int = 10, maxIter: Int = 1, initFeas: Boolean = false, curvilinear: Boolean = false, bbType: Int = 1, maxSrcht: Int = 30)(implicit space: MutableVectorField[T, Double])

projection
projection operations
tolerance
termination criterion: tolerance for norm of projected gradient
suffDec
sufficient decrease parameter
alphaMax
longest step
alphaMin
shortest step
bbMemory
number of history entries for linesearch
maxIter
maximum number of iterations
initFeas
is the initial guess feasible, or should it be projected?
curvilinear
if curvilinear true, do the projection inside line search in place of doing it in chooseDescentDirection
maxSrcht
maximum number of iterations inside line search

Type Members

case class History(alphaBB: Double, fvals: IndexedSeq[Double]) extends Product with Serializable
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

final def !=(arg0: Any): Boolean

Definition Classes
AnyRef → Any
final def ##(): Int

Definition Classes
AnyRef → Any
final def ==(arg0: Any): Boolean

Definition Classes
AnyRef → Any
def adjust(newX: T, newGrad: T, newVal: Double): (Double, T)

Attributes
protected
Definition Classes
FirstOrderMinimizer
def adjustFunction(f: DiffFunction[T]): DiffFunction[T]

Attributes
protected
Definition Classes
FirstOrderMinimizer
final def asInstanceOf[T0]: T0

Definition Classes
Any
def bbAlpha(s: T, y: T): Double

From Mark Schmidt's Matlab code if bbType == 1 alpha = (s'*s)/(s'*y); else alpha = (s'*y)/(y'*y);
From Mark Schmidt's Matlab code if bbType == 1 alpha = (s'*s)/(s'*y); else alpha = (s'*y)/(y'*y);

Attributes
protected
val bbType: Int
def calculateObjective(f: DiffFunction[T], x: T, history: History): (Double, T)

Attributes
protected
Definition Classes
FirstOrderMinimizer
def chooseDescentDirection(state: State, f: DiffFunction[T]): T

Attributes
protected
Definition Classes
SpectralProjectedGradient → FirstOrderMinimizer
def clone(): AnyRef

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( ... )
val curvilinear: Boolean

if curvilinear true, do the projection inside line search in place of doing it in chooseDescentDirection
def determineStepSize(state: State, f: DiffFunction[T], direction: T): Double

Attributes
protected
Definition Classes
SpectralProjectedGradient → FirstOrderMinimizer
final def eq(arg0: AnyRef): Boolean

Definition Classes
AnyRef
def equals(arg0: Any): Boolean

Definition Classes
AnyRef → Any
def finalize(): Unit

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( classOf[java.lang.Throwable] )
final def getClass(): Class[_]

Definition Classes
AnyRef → Any
def hashCode(): Int

Definition Classes
AnyRef → Any
def infiniteIterations(f: DiffFunction[T], state: State): Iterator[State]

Definition Classes
FirstOrderMinimizer
val initFeas: Boolean

is the initial guess feasible, or should it be projected?
def initialHistory(f: DiffFunction[T], init: T): History

Attributes
protected
Definition Classes
SpectralProjectedGradient → FirstOrderMinimizer
def initialState(f: DiffFunction[T], init: T): State

Attributes
protected
Definition Classes
FirstOrderMinimizer
final def isInstanceOf[T0]: Boolean

Definition Classes
Any
def iterations(f: DiffFunction[T], init: T): Iterator[State]

Definition Classes
FirstOrderMinimizer
def logger: LazyLogger

Attributes
protected
Definition Classes
SerializableLogging
val maxSrcht: Int

maximum number of iterations inside line search
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
def minimize(f: DiffFunction[T], init: T): T

Definition Classes
FirstOrderMinimizer → Minimizer
def minimizeAndReturnState(f: DiffFunction[T], init: T): State

Definition Classes
FirstOrderMinimizer
final def ne(arg0: AnyRef): Boolean

Definition Classes
AnyRef
final def notify(): Unit

Definition Classes
AnyRef
final def notifyAll(): Unit

Definition Classes
AnyRef
val numberOfImprovementFailures: Int

Definition Classes
FirstOrderMinimizer
def projectedVector(x: T, g: T)(implicit vspace: Module[T, Double]): T

Definition Classes
Projecting
val projection: (T) ⇒ T

projection operations
projection operations

Definition Classes
SpectralProjectedGradient → Projecting
final def synchronized[T0](arg0: ⇒ T0): T0

Definition Classes
AnyRef
def takeStep(state: State, dir: T, stepSize: Double): T

Attributes
protected
Definition Classes
SpectralProjectedGradient → FirstOrderMinimizer
def toString(): String

Definition Classes
AnyRef → Any
def updateFValWindow(oldState: State, newAdjVal: Double): IndexedSeq[Double]

Attributes
protected
Definition Classes
FirstOrderMinimizer
def updateHistory(newX: T, newGrad: T, newVal: Double, f: DiffFunction[T], oldState: State): History

Attributes
protected
Definition Classes
SpectralProjectedGradient → FirstOrderMinimizer
final def wait(): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
final def wait(arg0: Long, arg1: Int): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
final def wait(arg0: Long): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )

class SpectralProjectedGradient[T] extends FirstOrderMinimizer[T, DiffFunction[T]] with Projecting[T] with SerializableLogging

Instance Constructors

Type Members

case class History(alphaBB: Double, fvals: IndexedSeq[Double]) extends Product with Serializable

Value Members

final def !=(arg0: Any): Boolean

final def ##(): Int

final def ==(arg0: Any): Boolean

def adjust(newX: T, newGrad: T, newVal: Double): (Double, T)

def adjustFunction(f: DiffFunction[T]): DiffFunction[T]

final def asInstanceOf[T0]: T0

def bbAlpha(s: T, y: T): Double

val bbType: Int

def calculateObjective(f: DiffFunction[T], x: T, history: History): (Double, T)

def chooseDescentDirection(state: State, f: DiffFunction[T]): T

def clone(): AnyRef

val curvilinear: Boolean

def determineStepSize(state: State, f: DiffFunction[T], direction: T): Double

final def eq(arg0: AnyRef): Boolean

def equals(arg0: Any): Boolean

def finalize(): Unit

final def getClass(): Class[_]

def hashCode(): Int

def infiniteIterations(f: DiffFunction[T], state: State): Iterator[State]

val initFeas: Boolean

def initialHistory(f: DiffFunction[T], init: T): History

def initialState(f: DiffFunction[T], init: T): State

final def isInstanceOf[T0]: Boolean

def iterations(f: DiffFunction[T], init: T): Iterator[State]

def logger: LazyLogger

val maxSrcht: Int

val minImprovementWindow: Int

def minimize(f: DiffFunction[T], init: T): T

def minimizeAndReturnState(f: DiffFunction[T], init: T): State

final def ne(arg0: AnyRef): Boolean

final def notify(): Unit

final def notifyAll(): Unit

val numberOfImprovementFailures: Int

def projectedVector(x: T, g: T)(implicit vspace: Module[T, Double]): T

val projection: (T) ⇒ T

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

def takeStep(state: State, dir: T, stepSize: Double): T

def toString(): String

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

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

final def wait(): Unit

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

final def wait(arg0: Long): Unit

Inherited from Projecting[T]

Inherited from FirstOrderMinimizer[T, DiffFunction[T]]

Inherited from SerializableLogging

Inherited from Serializable

Inherited from Serializable

Inherited from Minimizer[T, DiffFunction[T]]

Inherited from AnyRef

Inherited from Any

Ungrouped