Incremental graph algorithms for parallel random access machines

Thesis/Dissertation · Wed Jan 01 04:00:00 EST 1986

OSTI ID:6918186

Pawagi, S R

Incremental graph algorithms deal with recomputing (updating) properties of a graph after a minor modification is made to it such as insertion or deletion of an edge or a vertex. Algorithms are developed for updating three classes of graph properties on a parallel random access machine (PRAM) in which processors share a common memory. The first class comprises of minimum spanning trees, connected components, cycle basis, bridges, bridge-connected components, cut points, and biconnected components of an undirected graph. Incremental algorithms for this class require updating a spanning tree/forest for the original graph. The second class of properties includes the distance matrix and shortest paths for all vertex pairs of an undirected graph; and the transitive closure, a topological ordering, and the dominators of a directed acyclic graph (DAG). This class of properties is characterized by start-over algorithms (one that does not have access to the previous solution) that are based on plus-min multiplication of the adjacency matrix for the graph. In the third class, the update problem for a maximum weight matching in trees is considered. An important feature of these incremental algorithms is their versatility, that is, they are adaptable on concurrent read, exclusive write and concurrent write models of PRAM.

• Research Organization: Maryland Univ., College Park (USA)
• OSTI ID: 6918186
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS
990210* -- Supercomputers-- (1987-1989)
ALGORITHMS
ARRAY PROCESSORS
COMPUTER GRAPHICS
MATHEMATICAL LOGIC
MEMORY DEVICES
PARALLEL PROCESSING
PROGRAMMING
RANDOMNESS