Class MultiThreadsByPieceCoreProcessor<RW extends MaterializedData & Writable>

java.lang.Object

fr.inria.rules.integraal.core.MultiThreadsByPieceCoreProcessor<RW>

Type Parameters:

RW - the writable materialized fact-base type processed by the core algorithm

All Implemented Interfaces:

CoreProcessor<RW>

public class MultiThreadsByPieceCoreProcessor<RW extends MaterializedData & Writable>
extends Object
implements CoreProcessor<RW>

Compute the core of a fact base with a multithreading algorithm that computes the core by retracting pieces. In the exhaustive variant, for each piece, we search all the homomorphisms that retract the piece into the atom set, and we keep the homomorphism that minimizes the number of remaining variables in the piece. In the "by specialization" variant, we specialize the piece when we find a homomorphism h that reduces the number of the variables in the piece (the specialization is done by sending using h on the piece and then search new homomorphisms that retract more variables of the initial piece). In the "by deletion" variant, when we found such a homomorphism, we remove the retracted variables from the piece and the atom set and search new homomorphisms on the result.

Author:

Guillaume Pérution-Kihli

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static enum	MultiThreadsByPieceCoreProcessor.Variant	Variant of the multithreaded by-piece core algorithm.

Constructor Summary

Constructors

Constructor	Description
MultiThreadsByPieceCoreProcessor()	Creates a processor using the deletion variant, the default evaluator, and a default thread cap.
MultiThreadsByPieceCoreProcessor(long limitNbThreads, MultiThreadsByPieceCoreProcessor.Variant variant)	Creates a processor with the default evaluator, a thread cap, and a variant.
MultiThreadsByPieceCoreProcessor(long limitNbThreads, FOQueryEvaluator<FOFormula, ? super RW> evaluator)	Creates a processor using the deletion variant and a custom evaluator.
MultiThreadsByPieceCoreProcessor(FOQueryEvaluator<FOFormula, ? super RW> evaluator, long limitNbThreads, MultiThreadsByPieceCoreProcessor.Variant variant)	Creates a processor with a custom evaluator, thread cap, and variant.

Method Summary

Modifier and Type	Method	Description
void	computeCore(RW fb, Set<Variable> frozenVariables)	Compute the core of a fact base by removing all the redundant atoms in it, in considering the frozen variables as constants, these variables must belong to the core.

Methods inherited from class Object

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

Methods inherited from interface CoreProcessor

computeCore

Constructor Details

MultiThreadsByPieceCoreProcessor

public MultiThreadsByPieceCoreProcessor(FOQueryEvaluator<FOFormula, ? super RW> evaluator,
 long limitNbThreads,
 MultiThreadsByPieceCoreProcessor.Variant variant)

Creates a processor with a custom evaluator, thread cap, and variant.

Parameters:

evaluator - Query evaluator

limitNbThreads - Maximum number of threads

variant - Variant of the algorithm

MultiThreadsByPieceCoreProcessor

public MultiThreadsByPieceCoreProcessor(long limitNbThreads,
 MultiThreadsByPieceCoreProcessor.Variant variant)

Creates a processor with the default evaluator, a thread cap, and a variant.

Parameters:

limitNbThreads - Maximum number of threads

variant - Variant of the algorithm

MultiThreadsByPieceCoreProcessor

public MultiThreadsByPieceCoreProcessor(long limitNbThreads,
 FOQueryEvaluator<FOFormula, ? super RW> evaluator)

Creates a processor using the deletion variant and a custom evaluator.

Parameters:

limitNbThreads - Maximum number of threads

evaluator - Query evaluator

MultiThreadsByPieceCoreProcessor

public MultiThreadsByPieceCoreProcessor()

Creates a processor using the deletion variant, the default evaluator, and a default thread cap.

Method Details

computeCore

public void computeCore(RW fb,
 Set<Variable> frozenVariables)

Description copied from interface: CoreProcessor

Compute the core of a fact base by removing all the redundant atoms in it, in considering the frozen variables as constants, these variables must belong to the core.

Specified by:

computeCore in interface CoreProcessor<RW extends MaterializedData & Writable>

Parameters:

fb - the writable data on which we want to compute the core

frozenVariables - Variables that will be treated as constants