Google OR-Tools v9.9
a fast and portable software suite for combinatorial optimization
Loading...
Searching...
No Matches
operations_research::sat::ConstraintGraphNeighborhoodGenerator Class Reference

#include <cp_model_lns.h>

Inheritance diagram for operations_research::sat::ConstraintGraphNeighborhoodGenerator:
operations_research::sat::NeighborhoodGenerator

Public Member Functions

 ConstraintGraphNeighborhoodGenerator (NeighborhoodGeneratorHelper const *helper, absl::string_view name)
 
Neighborhood Generate (const CpSolverResponse &initial_solution, double difficulty, absl::BitGenRef random) final
 
- Public Member Functions inherited from operations_research::sat::NeighborhoodGenerator
 NeighborhoodGenerator (absl::string_view name, NeighborhoodGeneratorHelper const *helper)
 
virtual ~NeighborhoodGenerator ()=default
 
virtual bool ReadyToGenerate () const
 Returns true if the neighborhood generator can generate a neighborhood.
 
double GetUCBScore (int64_t total_num_calls) const
 
void AddSolveData (SolveData data)
 
double Synchronize ()
 
std::string name () const
 Returns a short description of the generator.
 
int64_t num_calls () const
 Number of times this generator was called.
 
int64_t num_fully_solved_calls () const
 Number of time the neighborhood was fully solved (OPTIMAL/INFEASIBLE).
 
int64_t num_improving_calls () const
 Out of num_calls(), how many improved the given solution.
 
int64_t num_consecutive_non_improving_calls () const
 
double difficulty () const
 The current difficulty of this generator.
 
double deterministic_limit () const
 The current time limit that the sub-solve should use on this generator.
 

Additional Inherited Members

- Protected Attributes inherited from operations_research::sat::NeighborhoodGenerator
const std::string name_
 
const NeighborhoodGeneratorHelperhelper_
 
absl::Mutex generator_mutex_
 

Detailed Description

Pick a random subset of constraint and relax all of their variables. We are a bit smarter than this because after the first constraint is selected, we only select constraints that share at least one variable with the already selected constraints. The variable from the "last" constraint are selected randomly.

Definition at line 552 of file cp_model_lns.h.

Constructor & Destructor Documentation

◆ ConstraintGraphNeighborhoodGenerator()

operations_research::sat::ConstraintGraphNeighborhoodGenerator::ConstraintGraphNeighborhoodGenerator ( NeighborhoodGeneratorHelper const * helper,
absl::string_view name )
inlineexplicit

Definition at line 554 of file cp_model_lns.h.

Member Function Documentation

◆ Generate()

Neighborhood operations_research::sat::ConstraintGraphNeighborhoodGenerator::Generate ( const CpSolverResponse & initial_solution,
double difficulty,
absl::BitGenRef random )
finalvirtual
Note
even if difficulty means full neighborhood, we go through the generation process to never get out of a connected components.

Start by a random constraint.

Stop if we have a full connected component.

Pick a random unprocessed constraint.

Add all the variable of this constraint and increase the set of next possible constraints.

Implements operations_research::sat::NeighborhoodGenerator.

Definition at line 1482 of file cp_model_lns.cc.


The documentation for this class was generated from the following files: