Module ojalgo

Package org.ojalgo.optimisation.convex

Class ConvexSolver.Configuration

java.lang.Object

org.ojalgo.optimisation.convex.ConvexSolver.Configuration

Enclosing class:

ConvexSolver

public static final class ConvexSolver.Configuration
extends Object

Field Summary

Fields

Modifier and Type	Field	Description
private boolean	myCombinedScaleFactor
private boolean	myExtendedPrecision
private NumberContext	myIterativeAccuracy
private Supplier<Preconditioner>	myIterativePreconditioner
private Supplier<IterativeSolverTask>	myIterativeSolver
private double	mySmallDiagonal
private Function<Structure2D,MatrixDecomposition.Solver<Double>>	mySolverGeneral
private Function<Structure2D,MatrixDecomposition.Solver<Double>>	mySolverSPD

Constructor Summary

Constructors

Constructor	Description
Configuration()

Method Summary

Modifier and Type	Method	Description
ConvexSolver.Configuration	combinedScaleFactor(boolean combinedScaleFactor)	Only relevant with extended precision.
ConvexSolver.Configuration	extendedPrecision(boolean extendedPrecision)	With extended precision the usual solver is wrapped by a master algorithm, implemented in Quadruple precision, that iteratively refines (zoom and shift) the problem to be solved by the delegate solver.
boolean	isCombinedScaleFactor()
boolean	isExtendedPrecision()
NumberContext	iterative()	Deprecated. Since v56 It's applied for you when calling newIterativeSolver(int)
ConvexSolver.Configuration	iterative(Supplier<IterativeSolverTask> solver, Supplier<Preconditioner> preconditioner, NumberContext accuracy)
ConvexSolver.Configuration	iterative(Supplier<IterativeSolverTask> solver, NumberContext accuracy)	Select which iterative linear system solver to use for the Schur-complement step in IterativeASS.
ConvexSolver.Configuration	iterative(NumberContext accuracy)	The accuracy of the iterative Schur complement solver used in IterativeASS.
IterativeSolverTask	newIterativeSolver(int maxIterations)	Returns a new iterative solver instance configured with the current accuracy, maximum iterations, and preconditioner settings.
MatrixDecomposition.Solver<Double>	newSolverGeneral(Structure2D structure)
MatrixDecomposition.Solver<Double>	newSolverSPD(Structure2D structure)
double	smallDiagonal()
ConvexSolver.Configuration	smallDiagonal(double factor)	The [Q] matrix (of quadratic terms) is "inverted" using a matrix decomposition returned by newSolverSPD(Structure2D).
ConvexSolver.Configuration	solverGeneral(Function<Structure2D,MatrixDecomposition.Solver<Double>> factory)	This matrix decomposition should be able to "invert" the full KKT systsem body matrix (which is symmetric) and/or its Schur complement with regards to the [Q] matrix (of quadratic terms).
ConvexSolver.Configuration	solverSPD(Function<Structure2D,MatrixDecomposition.Solver<Double>> factory)	The [Q] matrix (of quadratic terms) is supposed to be symmetric positive definite (or at least semidefinite), but in reality there are usually many deficiencies.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Details

myCombinedScaleFactor

private boolean myCombinedScaleFactor

myExtendedPrecision

private boolean myExtendedPrecision

myIterativeAccuracy

private NumberContext myIterativeAccuracy

myIterativePreconditioner

private Supplier<Preconditioner> myIterativePreconditioner

myIterativeSolver

private Supplier<IterativeSolverTask> myIterativeSolver

mySmallDiagonal

private double mySmallDiagonal

mySolverGeneral

private Function<Structure2D,MatrixDecomposition.Solver<Double>> mySolverGeneral

mySolverSPD

private Function<Structure2D,MatrixDecomposition.Solver<Double>> mySolverSPD

Constructor Details

Configuration

public Configuration()

Method Details

combinedScaleFactor

public ConvexSolver.Configuration combinedScaleFactor(boolean combinedScaleFactor)

Only relevant with extended precision. With the extended precision solver the primal and dual variables are scaled (shift and zoom) to iteratively generate subproblems. In theory there are different scaling factors for the primal and dual variables, but forcing them to be the same enables simplifications resulting in significant performance gains.

The default is to use the same scaling factor for both primal and dual variables. By setting this to false, you switch to a slower more complex, but theoretically more accurate and flexible algorithm.

See Also:

• extendedPrecision(boolean)

extendedPrecision

public ConvexSolver.Configuration extendedPrecision(boolean extendedPrecision)

With extended precision the usual solver is wrapped by a master algorithm, implemented in Quadruple precision, that iteratively refines (zoom and shift) the problem to be solved by the delegate solver. This enables to handle constraints with very high accuracy.

The iterative refinement solver cannot handle general inequality constraints, only simple variable bounds (modelled as inequality constraints).

This is an experimental feature!

Setting this to true you should most likely also set the Optimisation.Options.solution to something matching that allows for higher precision.

isCombinedScaleFactor

public boolean isCombinedScaleFactor()

isExtendedPrecision

public boolean isExtendedPrecision()

iterative

@Deprecated
public NumberContext iterative()

Deprecated.

Since v56 It's applied for you when calling newIterativeSolver(int)

iterative

public ConvexSolver.Configuration iterative(NumberContext accuracy)

The accuracy of the iterative Schur complement solver used in IterativeASS. This is the step that calculates the Lagrange multipliers (dual variables). The iterative solver used is a ConjugateGradientSolver.

iterative

public ConvexSolver.Configuration iterative(Supplier<IterativeSolverTask> solver,
 NumberContext accuracy)

Select which iterative linear system solver to use for the Schur-complement step in IterativeASS. Default is ConjugateGradientSolver. You may set e.g. new QMRSolver().

iterative

public ConvexSolver.Configuration iterative(Supplier<IterativeSolverTask> solver,
 Supplier<Preconditioner> preconditioner,
 NumberContext accuracy)

newIterativeSolver

public IterativeSolverTask newIterativeSolver(int maxIterations)

Returns a new iterative solver instance configured with the current accuracy, maximum iterations, and preconditioner settings.

newSolverGeneral

public MatrixDecomposition.Solver<Double> newSolverGeneral(Structure2D structure)

newSolverSPD

public MatrixDecomposition.Solver<Double> newSolverSPD(Structure2D structure)

smallDiagonal

public double smallDiagonal()

smallDiagonal

public ConvexSolver.Configuration smallDiagonal(double factor)

The [Q] matrix (of quadratic terms) is "inverted" using a matrix decomposition returned by newSolverSPD(Structure2D). If, after decomposition, MatrixDecomposition.Solver.isSolvable() returns false a small constant is added to the diagonal.

The small constant will be the largest absolute element times this small diagonal factor.

This is only meant to handle minor, unexpected, deficiencies.

solverGeneral

public ConvexSolver.Configuration solverGeneral(Function<Structure2D,MatrixDecomposition.Solver<Double>> factory)

This matrix decomposition should be able to "invert" the full KKT systsem body matrix (which is symmetric) and/or its Schur complement with regards to the [Q] matrix (of quadratic terms).

solverSPD

public ConvexSolver.Configuration solverSPD(Function<Structure2D,MatrixDecomposition.Solver<Double>> factory)

The [Q] matrix (of quadratic terms) is supposed to be symmetric positive definite (or at least semidefinite), but in reality there are usually many deficiencies. This matrix decomposition should handle "inverting" the [Q] matrix.