Package org.ddolib.ddo.core.compilation

Class CompilationConfig<T>

java.lang.Object

org.ddolib.ddo.core.compilation.CompilationConfig<T>

Type Parameters:

T - the type representing the state of the problem

public class CompilationConfig<T>
extends Object

Represents the configuration parameters used during the compilation of a Multi-valued Decision Diagram (MDD) or similar decision structure.

A CompilationConfig object centralizes all components, heuristics, and algorithmic options required to guide the compilation process. It defines how the search space is explored, bounded, reduced, and pruned.

This configuration supports multiple compilation strategies such as exact, relaxed, approximate, or hybrid approaches. It also integrates advanced mechanisms like dominance checking, caching, and large neighborhood search (LNS).

See Also:

• CompilationType
• Problem
• Relaxation
• VariableHeuristic
• StateRanking
• SubProblem
• FastLowerBound
• DominanceChecker
• SimpleCache
• CutSetType
• DebugLevel
• ReductionStrategy

Field Summary

Fields

Modifier and Type	Field	Description
Double	bestUB	Best known upper bound on the objective value at compilation time.
Optional<SimpleCache<T>>	cache	Optional cache to store previously computed states and avoid redundant work.
CompilationType	compilationType	Specifies how the MDD is compiled.
CutSetType	cutSetType	Defines the cut set strategy used to control node expansion and pruning.
DebugLevel	debugLevel	Debugging level controlling logging and internal consistency checks.
DominanceChecker<T>	dominance	Dominance checker used to prune dominated states and reduce the size of the search space.
Boolean	exportAsDot	Indicates whether the compiled MDD should be exported as a DOT file (Graphviz format) for visualization purposes.
FastLowerBound<T>	flb	Fast lower bound heuristic used to estimate the best achievable objective value from a given state.
int[]	initialSolution	Initial solution provided to guide the search (e.g., for heuristics or LNS).
Integer	maxWidth	Maximum allowed width (number of nodes) per layer in the MDD.
double	probability	Probability parameter used in randomized strategies (e.g., LNS or heuristics).
Problem<T>	problem	Reference to the optimization or constraint problem being solved.
ReductionStrategy<T>	reductionStrategy	Strategy used to reduce the width of the MDD by merging or discarding nodes.
Relaxation<T>	relaxation	Relaxation model used to approximate or merge states in relaxed MDD compilation.
SubProblem<T>	residual	The residual (sub)problem defining the remaining search space to explore during compilation.
int[]	solution	Stores the current or best solution found during compilation.
StateRanking<T>	stateRanking	Ranking heuristic used to prioritize states when pruning nodes in width-limited MDD layers.
Boolean	useLNS	Indicates whether Large Neighborhood Search (LNS) should be used to improve solutions during compilation.
VariableHeuristic<T>	variableHeuristic	Heuristic used to select the next variable to branch on during compilation.

Constructor Summary

Constructors

Constructor	Description
CompilationConfig(Model<T> model)	Constructs a new compilation configuration for the given model.

Method Summary

Modifier and Type	Method	Description
CompilationConfig<T>	copy()	Creates a shallow copy of this configuration.
String	toString()	Returns a human-readable string representation of this configuration.

Methods inherited from class java.lang.Object

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

Field Details

compilationType

public CompilationType compilationType

Specifies how the MDD is compiled.

Determines whether the compilation is exact, relaxed, approximate, or hybrid depending on the selected algorithm.

problem

public Problem<T> problem

Reference to the optimization or constraint problem being solved.

variableHeuristic

public VariableHeuristic<T> variableHeuristic

Heuristic used to select the next variable to branch on during compilation.

stateRanking

public StateRanking<T> stateRanking

Ranking heuristic used to prioritize states when pruning nodes in width-limited MDD layers.

residual

public SubProblem<T> residual

The residual (sub)problem defining the remaining search space to explore during compilation.

maxWidth

public Integer maxWidth

Maximum allowed width (number of nodes) per layer in the MDD.

Smaller values improve performance but may reduce solution quality.

flb

public FastLowerBound<T> flb

Fast lower bound heuristic used to estimate the best achievable objective value from a given state.

dominance

public DominanceChecker<T> dominance

Dominance checker used to prune dominated states and reduce the size of the search space.

bestUB

public Double bestUB

Best known upper bound on the objective value at compilation time.

exportAsDot

public Boolean exportAsDot

Indicates whether the compiled MDD should be exported as a DOT file (Graphviz format) for visualization purposes.

debugLevel

public DebugLevel debugLevel

Debugging level controlling logging and internal consistency checks.

Higher levels provide more detailed diagnostics but may impact performance.

reductionStrategy

public ReductionStrategy<T> reductionStrategy

Strategy used to reduce the width of the MDD by merging or discarding nodes.

relaxation

public Relaxation<T> relaxation

Relaxation model used to approximate or merge states in relaxed MDD compilation.

cache

public Optional<SimpleCache<T>> cache

Optional cache to store previously computed states and avoid redundant work.

cutSetType

public CutSetType cutSetType

Defines the cut set strategy used to control node expansion and pruning.

initialSolution

public int[] initialSolution

Initial solution provided to guide the search (e.g., for heuristics or LNS).

solution

public int[] solution

Stores the current or best solution found during compilation.

probability

public double probability

Probability parameter used in randomized strategies (e.g., LNS or heuristics).

useLNS

public Boolean useLNS

Indicates whether Large Neighborhood Search (LNS) should be used to improve solutions during compilation.

Constructor Details

CompilationConfig

public CompilationConfig(Model<T> model)

Constructs a new compilation configuration for the given model.

Parameters:

model - the model containing problem-specific components (heuristics, bounds, dominance, etc.)

Method Details

toString

public String toString()

Returns a human-readable string representation of this configuration.

Overrides:

toString in class Object

Returns:

a formatted string containing the compilation type, residual problem, and best known upper bound

copy

public CompilationConfig<T> copy()

Creates a shallow copy of this configuration.

Some components are re-fetched from the underlying model (e.g., heuristics, bounds, dominance), while others are directly copied.

Returns:

a new CompilationConfig instance with the same parameters