The Bipartite data structure implements a distributed bipartite graph, where the nodes of the graph fall into two classes designated e and h. The user builds the Bipartite graph by specifying on each processor the to-edges for local nodes and then connecting the graph to establish the corresponding, possibly remote, from-edges. The user locally computes new node data values based on the edge weight and value of connected nodes. These new values are then globally propagated for an entire node class by validating the node values.
The Bipartite data structure can be used in bulk-synchronous simulations where the underlying problem structure is a bipartite graph.
DistObj, Snapshot.