Class RecursiveSolver

java.lang.Object

solvers.queueingnet.QNSolver

solvers.queueingnet.RecursiveSolver

All Implemented Interfaces:

QNSolverInterface

Direct Known Subclasses:

ConvolutionSolver, RECALNonRecursiveSolver, RECALSolver

public class RecursiveSolver
extends QNSolver

This class implements the RecursiveSolver object, which contains common methods for the Convolution and RECAL algorithms.

Author:

Michail Makaronidis, 2010

Field Summary

Fields

Modifier and Type	Field	Description
protected Map<Integer, BigRational>	factorial	The factorials for some numbers are pre-computed for efficiency reasons.
protected Map<PopulationVector, Map<MultiplicitiesVector, BigRational>>	Gmap	Here is stored every normalising constant computed.

Fields inherited from class QNSolver

G, memUsage, qnm, totalTimer

Constructor Summary

Constructors

Constructor	Description
RecursiveSolver(QNModel qnm)	Creates and initialises a RecursiveSolver object.

Method Summary

Modifier and Type	Method	Description
protected BigRational	compute(MultiplicitiesVector m, PopulationVector p)	Computes any normalising constant using a simple recursion and taking advantage of any already computed and stored value.
void	computePerformanceMeasures()	Computes and stores the performance measures (mean throughputs and mean queue lengths for the current model.
protected BigRational	findG(MultiplicitiesVector m, PopulationVector p)	Searches for a normalising constant in the memory.
protected BigRational	initialConditionFor(PopulationVector n)	Returns the initial condition corresponding to a particular PopulationVector object.
protected BigRational	recallG(MultiplicitiesVector m, PopulationVector p)	Searches for a normalising constant in the memory.
protected void	storeG(MultiplicitiesVector m, PopulationVector p, BigRational Gval)	Stores a computed normalising constant value for future re-use.
protected void	storeG(MultiplicitiesVector m, PopulationVector p, BigRational Gval, Map<PopulationVector, Map<MultiplicitiesVector, BigRational>> map)	Stores at a particular Map<PopulationVector, Map<MultiplicitiesVector, BigRational>> data structure a computed normalising constant value for future re-use.

Methods inherited from class QNSolver

computeNormalisingConstant, getNormalisingConstant, printMemUsage, printTimeStatistics, printWelcome

Methods inherited from class Object

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

Field Details

Gmap

protected Map<PopulationVector, Map<MultiplicitiesVector, BigRational>> Gmap

Here is stored every normalising constant computed. This speeds up the computation.

factorial

protected Map<Integer, BigRational> factorial

The factorials for some numbers are pre-computed for efficiency reasons.

Constructor Details

RecursiveSolver

public RecursiveSolver(QNModel qnm)
                throws InternalErrorException

Creates and initialises a RecursiveSolver object.

Parameters:

qnm - The QNModel object that we are working on

Throws:

InternalErrorException - Thrown when the solver cannot be initialised

Method Details

initialConditionFor

protected BigRational initialConditionFor(PopulationVector n)

Returns the initial condition corresponding to a particular PopulationVector object.

Parameters:

n - The PopulationVector object

Returns:

The corresponding initial condition

storeG

protected void storeG(MultiplicitiesVector m,
 PopulationVector p,
 BigRational Gval)

Stores a computed normalising constant value for future re-use.

Parameters:

m - The MultiplicitiesVector

p - The PopulationVector

Gval - The corresponding normalising constant that has been computed

storeG

protected void storeG(MultiplicitiesVector m,
 PopulationVector p,
 BigRational Gval,
 Map<PopulationVector, Map<MultiplicitiesVector, BigRational>> map)

Stores at a particular Map<PopulationVector, Map<MultiplicitiesVector, BigRational>> data structure a computed normalising constant value for future re-use.

Parameters:

m - The MultiplicitiesVector

p - The PopulationVector

Gval - The corresponding normalising constant that has been computed

map - The data structure at which the constant will be stored

recallG

protected BigRational recallG(MultiplicitiesVector m,
 PopulationVector p)

Searches for a normalising constant in the memory. If it's not found, it returns null.

Parameters:

m - The MultiplicitiesVector

p - The PopulationVector

Returns:

The normalising constant as a BigRational object. Null if it has not been precomputed in the past.

findG

protected BigRational findG(MultiplicitiesVector m,
 PopulationVector p)
                     throws InternalErrorException

Searches for a normalising constant in the memory. If it's not found, it computes and returns it.

Parameters:

m - The MultiplicitiesVector

p - The PopulationVector

Returns:

The normalising constant as a BigRational object

Throws:

InternalErrorException - An exception is thrown if any internal error is encountered during computations.

compute

protected BigRational compute(MultiplicitiesVector m,
 PopulationVector p)
                       throws InternalErrorException

Computes any normalising constant using a simple recursion and taking advantage of any already computed and stored value.

Parameters:

m - The MultiplicitiesVector

p - The PopulationVector

Returns:

The normalising constant as a BigRational object

Throws:

InternalErrorException - An exception is thrown if any internal error is encountered during computations.

computePerformanceMeasures

public void computePerformanceMeasures()
                                throws InternalErrorException

Description copied from class: QNSolver

Computes and stores the performance measures (mean throughputs and mean queue lengths for the current model.

Specified by:

computePerformanceMeasures in interface QNSolverInterface

Overrides:

computePerformanceMeasures in class QNSolver

Throws:

InternalErrorException - Thrown when any computation fails