DDOLib-Scala
API
Generated with
DDOLib-Scala/org.ddolibscala/org.ddolibscala.tools/org.ddolibscala.tools.ddo/org.ddolibscala.tools.ddo.heuristics/org.ddolibscala.tools.ddo.heuristics.cluster/Hybrid

Hybrid

org.ddolibscala.tools.ddo.heuristics.cluster.Hybrid
object Hybrid

Hybrid reduction strategy that combines cost-based and distance-based clustering for decision diagram layers. This strategy is a hybridation between cost based selection and GHP. It preserves the w * alpha best nodes (alpha between 0 and 1) and merge the other nodes using clustering. It requires a problem-specific StateRanking comparator to break the ties between nodes of same cost, and a problem-specif StateDistance to quantify the dissimilarity between states.

This class implements ReductionStrategy and merges two strategies:

  • CostBased: preserves a fraction of nodes based on their ranking
  • GHP: clusters the remaining nodes based on a distance metric

The combination is controlled by a weighting parameter alpha:

  • alpha fraction of the clusters are preserved using cost-based ranking ;
  • 1 - alpha fraction of the clusters are formed using the GHP distance-based method.

Attributes

See also
Graph
Supertypes
class Object
trait Matchable
class Any
Self type
Hybrid.type

Members list

Value members

Concrete methods

def apply[T](ranking: StateRanking[T], distance: StateDistance[T], alpha: Double, seed: Long): Hybrid[T]

Constructs a Hybrid reduction strategy with specified ranking, distance, alpha, and seed.

Constructs a Hybrid reduction strategy with specified ranking, distance, alpha, and seed.

Type parameters

T

the type of states in the decision diagram

Value parameters

alpha

fraction of clusters preserved using cost-based strategy

distance

state distance used for GHP clustering

ranking

state ranking used for cost-based preservation

seed

random seed for distance-based clustering

Attributes

Returns

a Hybrid reduction strategy

Generated with