Cluster

Classes
class	TAdmCondition
	Defines basic interface for admissibility conditions. More...
class	TAlgAdmCond
	base class for algebraic admissibility conditions More...
class	TStdAlgAdmCond
	Standard admissibility condition based on matrix graph criteria. More...
class	TAlgCTBuilder
	Base class for cluster tree construction algorithms based on graph partitioning with graph defined by a sparse matrix. More...
class	TAlgNDCTBuilder
	Enhances algebraic clustering by nested dissection. More...
class	TPartAlgCTBuilder
	Enhances algebraic clustering by allowing the user to define the first level of index partitioning, e.g. define which index belongs to which son cluster. More...
class	TAlgPartStrat
	Base class for partitioning strategies for algebraic clustering. More...
class	TBFSAlgPartStrat
	Graph partitioning using BFS algorithm and FM optimisation. More...
class	TMLAlgPartStrat
	Multi level graph partitioning. More...
class	TMETISAlgPartStrat
	Graph partitioning using METIS. More...
class	TScotchAlgPartStrat
	Graph partitioning using Scotch. More...
class	TChacoAlgPartStrat
	Graph partitioning using CHACO. More...
class	TBCBuilder
	Recursively build block cluster tree with supplied admissibility condition. More...
class	TBlockCluster
	Representing a node in a block cluster tree as product of two clusters. More...
class	TBlockClusterTree
	Represents a block cluster tree. More...
class	TGeomCTBuilder
	Base class for all cluster tree constructors based on geometry data. More...
class	TBSPCTBuilder
	Base class for all cluster tree constructors based on BSP. More...
class	TBSPNDCTBuilder
	Combines binary space partitioning with nested dissection based on connectivity defined by a sparse matrix. More...
class	TGeomPartCTBuilder
	Enhances other geometrical ct builders by allowing the user to define the first level of index partitioning by a given vector. More...
class	TGeomGroupCTBuilder
	Enhances geometrical ct builder by allowing to group indices, e.g. the groups are clustered and later expanded, ensuring that all indices in a group are in the same cluster NOTE: bounding boxes per index are not yet supported (bb_min/bb_max) More...
class	TBSPPartStrat
	Base class for partitioning strategies for geometrical BSP clustering. More...
class	TGeomBSPPartStrat
	Partition according to geometrical volume of index sets. More...
class	TCardBSPPartStrat
	Partition according to cardinality of index sets. More...
class	TPCABSPPartStrat
	Partition according to principle component analysis. More...
class	TNDBSPPartStrat
	Special partition strategy to optimized nested dissection clustering. More...
class	TAutoBSPPartStrat
	Automatic choice of best partitioning strategy. More...
class	TCluster
	Represents a node in a cluster tree with an arbitrary number of sons. More...
class	TClusterBasis< T >
	class representing a nested cluster basis More...
class	TDenseClusterBasisBuilder< T >
	class for constructing cluster bases using dense matrices More...
class	THClusterBasisBuilder< T >
	class for constructing cluster bases using H-matrices More...
class	TClusterTree
	Represents a cluster tree with permutation of index sets. More...
class	TDiGraph
	Class for directed graph represented by adjacency matrix in sparse format (assuming sparse graph!) More...
class	TFVS
	Uses a heuristic algorithm to compute feedback vertex set of a directed graph represented by a sparse matrix. More...
class	TStdGeomAdmCond
	Standard admissibility for FEM/BEM applications normal : adm iff min( diam(τ), diam(σ) ) ≤ η·dist(τ,σ) use_max: adm iff max( diam(τ), diam(σ) ) ≤ η·dist(τ,σ) More...
class	TWeakStdGeomAdmCond
	Combination of standard and weak admissibility. More...
class	THiLoFreqGeomAdmCond
	Admissibility for high and low frequency regimes. More...
class	TGeomCluster
	Extend standard cluster by bounding box. More...
class	TGraph
	Class for a undirected graph stored as adjacency matrix in CRS representation. More...
class	TEWGraph
	Represents undirected graph with edge weights. More...
class	TNodeSet
	Represents a set of nodes by an array. More...

Typedefs
using	weight_t = int
	type for edge weights
using	node_t = idx_t
	Basic type for storing node names.

Enumerations
enum	adjust_depth_mode_t { adjust_depth_off, adjust_depth_on }
	turns on/off depth adjustment of sub trees
enum	edgecut_weights_mode_t { edgecut_weights_off, edgecut_weights_on }
	turn on/off usage of weights in edge cut
enum	edge_weights_mode_t { edge_weights_off, edge_weights_on, edge_weights_on_sym }
	turn on/off usage of edge weights
enum	cluster_level_mode_t { cluster_level_equal, cluster_level_all }
	turn on/off enforcement of same cluster level in block cluster
enum	diam_mode_t { use_min_diam, use_max_diam }
	choose between minimum and maximum of cluster diameters in std. adm.

Detailed Description

This modules provides all classes for index set and cluster management, e.g. cluster trees, block cluster trees.

#include <hlib-cluster.hh>