PenaltyFunctionModel

Overview

Package

Class

Use

Tree

Deprecated

Index

Help

PREV CLASS NEXT CLASS

FRAMES NO FRAMES

SUMMARY: NESTED | FIELD | CONSTR | METHOD

DETAIL: FIELD | CONSTR | METHOD

model
Class PenaltyFunctionModel

java.lang.Object
  function.BasicScalarFunction
      function.VariableDomainScalarFunction
          model.BasicApproxModel
              model.PenaltyFunctionModel

All Implemented Interfaces:: DataListener, Computable, ScalarFunction, VariableBounds, VariableInputDimension, ApproxModel, DataFit

public class PenaltyFunctionModel
extends BasicApproxModel
extends BasicApproxModel

Computes mean and variance for an external penalty function augmented Lagrangian. Computes first and second moments of the distributions for the constraint violations, then uses those values as the values of a Gaussian. This is basically to find the closest Gaussian to these non-Gaussian posterior distributions over constraint violations. Note: although the posterior over constraint values is Gaussian, the posterior over violations is not Gaussian. Even so, the approximation above is made, then mean and variance for L = f + \rho^T max(0, g) are easy to compute. Read my thesis for the details.

Author:: dgorur

Field Summary

Fields inherited from class function.BasicScalarFunction
`DEFAULT_BOUNDS, DEFAULT_DIM, VALUE_OUTSIDE_BOUNDS`

Constructor Summary
`PenaltyFunctionModel()`

Method Summary
`void`	`calibrate()` Calibrates this DataFit.
`void`	`calibrateOptional()` Optional calibration, based on some measure of calibration being required.
`void`	`clear()` Notification that all data has been removed.
`int`	`dataAdded(DataBlock block)` Notification that a `DataBlock` has been added.
`boolean`	`dataAdded(DataPoint dataPoint)` Notification that a single `DataPoint` has been added.
`int`	`dataRemoved(DataBlock dataBlock)` Notification that a `DataBlock` has been removed.
`boolean`	`dataRemoved(DataPoint dataPoint)` Notification that a single `DataPoint` has been removed.
`VectorFunction`	`getCheapConstraints()`
`java.util.ArrayList<ApproxModel>`	`getConstraintModels()`
`cern.colt.matrix.DoubleMatrix1D`	`getConstraintTargets()` Returns the constraintTargets.
`double`	`getMaxUncertainty()`
`ApproxModel`	`getObjectiveModel()`
`cern.colt.matrix.DoubleMatrix1D`	`getPenaltyVector()`
`void`	`init()` Initialization method.
`protected ScalarFunction`	`makeUncertaintyEstimator()`
`double`	`quickCompute(double[] x)` Compute method without dimension checking.
`void`	`setCheapConstraints(VectorFunction cheapConstraints)`
`void`	`setConstraintModels(java.util.ArrayList<ApproxModel> constraintModels)`
`void`	`setConstraintTargets(cern.colt.matrix.DoubleMatrix1D constraintTargets)` Sets the constraintTargets.
`void`	`setObjectiveModel(ApproxModel objectiveModel)`
`void`	`setPenaltyVector(cern.colt.matrix.DoubleMatrix1D penaltyVector)`

Methods inherited from class model.BasicApproxModel
`getOutputNumber, getUncertaintyEstimator, setOutputNumber, setUncertaintyEstimator`

Methods inherited from class function.VariableDomainScalarFunction
`getBounds, getInputDimension, setBounds, setInputDimension`

Methods inherited from class function.BasicScalarFunction
`checkDimensions, compute, compute, compute, compute, compute, compute, getOutputDimension, isWithinBounds, makeBounds, quickCompute`

Methods inherited from class java.lang.Object
`clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait`

Methods inherited from interface function.ScalarFunction
`compute, compute, compute, compute, quickCompute`

Methods inherited from interface function.Computable
`compute, compute, getBounds, getInputDimension, getOutputDimension, isWithinBounds`

Methods inherited from interface function.VariableInputDimension
`setInputDimension`

Constructor Detail

PenaltyFunctionModel

public PenaltyFunctionModel()

Method Detail

calibrate

public void calibrate()

Description copied from interface: DataFit

Calibrates this DataFit.

calibrateOptional

public void calibrateOptional()

Description copied from interface: DataFit

Optional calibration, based on some measure of calibration being required.