fr.cnes.sirius.patrius.math.optim.joptimizer.optimizers

Class OptimizationRequestHandler

java.lang.Object

fr.cnes.sirius.patrius.math.optim.joptimizer.optimizers.OptimizationRequestHandler

Direct Known Subclasses:

AbstractLPOptimizationRequestHandler, BarrierMethod, NewtonLEConstrainedFSP, NewtonLEConstrainedISP, NewtonUnconstrained, PrimalDualMethod

public class OptimizationRequestHandler
extends Object

Generic class for optimization process.

Since:

4.6

Field Summary

Fields

Modifier and Type	Field and Description
protected int	dim dimension
static double	MAX_ITERATIONS Maximum number of iterations
protected int	meq number of equalities
protected int	mieq number of inequalities
protected OptimizationRequest	request optimization request
protected OptimizationResponse	response optimization response
static double	SCALAR Scalar
protected OptimizationRequestHandler	successor optimization request handler

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	OptimizationRequestHandler() Protected constructor

Method Summary

Modifier and Type	Method and Description
protected boolean	checkCustomExitConditions(RealVector y) Overriding this, a subclass can define some extra condition for exiting the iteration loop.
protected RealVector	findEqFeasiblePoint(RealMatrix aMatrix, RealVector bVector) Find a solution of the linear (equalities) system A.x = b.
protected RealVector	findEqFeasiblePoint2(RealMatrix aMatrix, RealVector bVector) Find a solution of the linear (equalities) system A.x = b.
protected int	forwardOptimizationRequest() Forward optimization request
protected RealMatrix	getA() Get the equalities constraints matrix
protected double	getAlpha() Get the calibration parameter for line search
protected RealMatrix	getAT() Get the equalities constraints matrix transposed
protected RealVector	getB() Get the equalities constraints vector
protected double	getBeta() Get the calibration parameter for line search
protected int	getDim() Get the number of variables.
protected ConvexMultivariateRealFunction	getF0() Objective function.
protected double	getF0(RealVector x) Objective function value at X.
protected ConvexMultivariateRealFunction[]	getFi() Inequality functions.
protected RealVector	getFi(RealVector vecX) Inequality functions values at X.
protected RealVector	getGradF0(RealVector x) Objective function gradient at X.
protected RealMatrix	getGradFi(RealVector vecX) Inequality functions gradients values at X.
protected RealMatrix	getHessF0(RealVector x) Objective function hessian at X.
protected RealMatrix[]	getHessFi(RealVector vecX) Inequality functions hessians values at X.
protected RealVector	getInitialLagrangian() Get the initial Lagrangian
protected RealVector	getInitialPoint() Get the initial point
protected String	getInteriorPointMethod() Get the chosen interior point method.
protected int	getMaxIteration() Get the maximum number of iteration in the search algorithm
protected int	getMeq() Get the number of equalities.
protected int	getMieq() Get the number of inequalities.
protected double	getMu() Get the calibration parameter for line search
protected RealVector	getNotFeasibleInitialPoint() Get the not feasible initial point
protected OptimizationRequest	getOptimizationRequest() Get the optimization request
OptimizationResponse	getOptimizationResponse() Get the optimization response
protected double	getTolerance() Get the tolerance for the minimum value
protected double	getToleranceFeas() Get the tolerance for the constraints satisfaction
protected double	getToleranceInnerStep() Get the tolerance for inner iterations in the barrier-method.
protected double	getToleranceKKT() Get the acceptable tolerance for KKT system resolution
protected boolean	isCheckKKTSolutionAccuracy() Check the accuracy of the solution of KKT system during iterations.
protected boolean	isCheckProgressConditions() If true, a progress in the relevant algorithm norms is required during iterations, otherwise the iteration will be exited with a warning
protected boolean	isInDomainF0(RealVector x) Objective function domain.
int	optimize() Optimizer
protected RealVector	rPri(RealVector x) rPri := Ax - b
void	setOptimizationRequest(OptimizationRequest req) Set the optimization request
protected void	setOptimizationResponse(OptimizationResponse res) Set the optimization response

Methods inherited from class java.lang.Object

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

Field Detail

SCALAR

public static final double SCALAR

Scalar

See Also:

Constant Field Values

MAX_ITERATIONS

public static final double MAX_ITERATIONS

Maximum number of iterations

See Also:

Constant Field Values

successor

protected OptimizationRequestHandler successor

optimization request handler

request

protected OptimizationRequest request

optimization request

response

protected OptimizationResponse response

optimization response

dim

protected int dim

dimension

meq

protected int meq

number of equalities

mieq

protected int mieq

number of inequalities

Constructor Detail

OptimizationRequestHandler

protected OptimizationRequestHandler()

Protected constructor

Method Detail

setOptimizationRequest

public void setOptimizationRequest(OptimizationRequest req)

Set the optimization request

Parameters:

req - request

getOptimizationRequest

protected OptimizationRequest getOptimizationRequest()

Get the optimization request

Returns:

request

setOptimizationResponse

protected void setOptimizationResponse(OptimizationResponse res)

Set the optimization response

Parameters:

res - response

getOptimizationResponse

public OptimizationResponse getOptimizationResponse()

Get the optimization response

Returns:

response

optimize

public int optimize()
             throws PatriusException

Optimizer

Returns:

retCode

Throws:

PatriusException - if an error occurs

forwardOptimizationRequest

protected int forwardOptimizationRequest()
                                  throws PatriusException

Forward optimization request

Returns:

retCode

Throws:

PatriusException - if an error occurs

getDim

protected final int getDim()

Get the number of variables.

Returns:

dimension

getMeq

protected final int getMeq()

Get the number of equalities.

Returns:

number

getMieq

protected final int getMieq()

Get the number of inequalities.

Returns:

number

getInitialPoint

protected RealVector getInitialPoint()

Get the initial point

Returns:

point

getNotFeasibleInitialPoint

protected RealVector getNotFeasibleInitialPoint()

Get the not feasible initial point

Returns:

point

getInitialLagrangian

protected RealVector getInitialLagrangian()

Get the initial Lagrangian

Returns:

Lagrangian

getA

protected final RealMatrix getA()

Get the equalities constraints matrix

Returns:

matrix

getAT

protected final RealMatrix getAT()

Get the equalities constraints matrix transposed

Returns:

transposed matrix

getB

protected final RealVector getB()

Get the equalities constraints vector

Returns:

vector

getMaxIteration

protected final int getMaxIteration()

Get the maximum number of iteration in the search algorithm

Returns:

max number

getTolerance

protected final double getTolerance()

Get the tolerance for the minimum value

Returns:

tolerance

getToleranceFeas

protected final double getToleranceFeas()

Get the tolerance for the constraints satisfaction

Returns:

tolerance

getToleranceInnerStep

protected final double getToleranceInnerStep()

Get the tolerance for inner iterations in the barrier-method.

Returns:

tolerance

getAlpha

protected final double getAlpha()

Get the calibration parameter for line search

Returns:

alpha

getBeta

protected final double getBeta()

Get the calibration parameter for line search

Returns:

beta

getMu

protected final double getMu()

Get the calibration parameter for line search

Returns:

mu

isCheckProgressConditions

protected final boolean isCheckProgressConditions()

If true, a progress in the relevant algorithm norms is required during iterations, otherwise the iteration will be exited with a warning

Returns:

true/false

isCheckKKTSolutionAccuracy

protected final boolean isCheckKKTSolutionAccuracy()

Check the accuracy of the solution of KKT system during iterations. If true, every inversion of the system must have an accuracy that satisfy the given toleranceKKT

Returns:

true/false

getToleranceKKT

protected final double getToleranceKKT()

Get the acceptable tolerance for KKT system resolution

Returns:

tolerance

getInteriorPointMethod

protected final String getInteriorPointMethod()

Get the chosen interior point method.

Returns:

point

getF0

protected final ConvexMultivariateRealFunction getF0()

Objective function.

Returns:

objective funtion

isInDomainF0

protected boolean isInDomainF0(RealVector x)

Objective function domain.

Parameters:

x - vector X

Returns:

true/false

getF0

protected double getF0(RealVector x)

Objective function value at X.

Parameters:

x - value X

Returns:

objective function

getGradF0

protected RealVector getGradF0(RealVector x)

Objective function gradient at X.

Parameters:

x - gradient X

Returns:

objective function

getHessF0

protected RealMatrix getHessF0(RealVector x)

Objective function hessian at X.

Parameters:

x - hessian X

Returns:

objective function

getFi

protected ConvexMultivariateRealFunction[] getFi()

Inequality functions.

Returns:

inequalities constraints array

getFi

protected RealVector getFi(RealVector vecX)

Inequality functions values at X.

Parameters:

vecX - values X

Returns:

inequality function

getGradFi

protected RealMatrix getGradFi(RealVector vecX)

Inequality functions gradients values at X.

Parameters:

vecX - gradients X

Returns:

inequality function

getHessFi

protected RealMatrix[] getHessFi(RealVector vecX)

Inequality functions hessians values at X.

Parameters:

vecX - hessians X

Returns:

inequality function

checkCustomExitConditions

protected boolean checkCustomExitConditions(RealVector y)

Overriding this, a subclass can define some extra condition for exiting the iteration loop.

Parameters:

y - vector

Returns:

false

findEqFeasiblePoint

protected RealVector findEqFeasiblePoint(RealMatrix aMatrix,
                                         RealVector bVector)
                                  throws PatriusRuntimeException

Find a solution of the linear (equalities) system A.x = b. A is a pxn matrix, with rank(A) = p < n.

Parameters:

aMatrix - matrix A

bVector - vector B

Returns:

solution

Throws:

PatriusRuntimeException - if an error occurs

See Also:

"S.Boyd and L.Vandenberghe, Convex Optimization, p. 682" NB: we are waiting for Csparsej to fix its qr decomposition issues.

findEqFeasiblePoint2

protected RealVector findEqFeasiblePoint2(RealMatrix aMatrix,
                                          RealVector bVector)
                                   throws PatriusRuntimeException

Find a solution of the linear (equalities) system A.x = b. A is a pxn matrix, with rank(A) = p < n.

Parameters:

aMatrix - matrix A

bVector - vector B

Returns:

solution

Throws:

PatriusRuntimeException - if an error occurs

See Also:

"S.Boyd and L.Vandenberghe, Convex Optimization, p. 682"

rPri

protected RealVector rPri(RealVector x)

rPri := Ax - b

Parameters:

x - vector

Returns:

solution