LAGraph Documentation

The LAGraph library is a collection of high level graph algorithms based on the GraphBLAS C API. These algorithms construct graph algorithms expressed in the language of linear algebra. Graphs are expressed as matrices, and the operations over these matrices are generalized through the use of a semiring algebraic structure.

LAGraph is available at https://github.com/GraphBLAS/LAGraph. LAGraph requires SuiteSparse:GraphBLAS, available at https://github.com/DrTimothyAldenDavis/GraphBLAS.

Contents:

• Introduction
• LAGraph context and error handling
  • LAGraph Context Functions
  • Error handling
• The Graph Object
  • Basic Graph Functions
• Algorithms
  • Basic
  • Advanced
• Utility Functions
  • Input/Output Functions
  • Matrix Structure Functions
  • Matrix Comparison Functions
  • Introspecting Types
  • Printing
  • Pre-defined semirings
• Experimental Algorithms
• Installation
• Acknowledgements
• References

Example Usage

Note that this simple example does not check any error conditions.

#include "LAGraph.h"

int main (void)
{
    // initialize LAGraph
    char msg [LAGRAPH_MSG_LEN] ;
    LAGraph_Init (msg) ;
    GrB_Matrix A = NULL ;
    GrB_Vector centrality = NULL ;
    LAGraph_Graph G = NULL ;

    // read a Matrix Market file from stdin and create a graph
    LAGraph_MMRead (&A, stdin, msg) ;
    LAGraph_New (&G, &A, LAGraph_ADJACENCY_UNDIRECTED, msg) ;

    // compute the out-degree of every node
    LAGraph_Cached_OutDegree (G, msg) ;

    // compute the pagerank
    int niters = 0 ;
    LAGr_PageRank (&centrality, &niters, G, 0.85, 1e-4, 100, msg) ;

    // print the result
    LAGraph_Vector_Print (centrality, LAGraph_COMPLETE, stdout, msg) ;

    // free the graph, the pagerank, and finish LAGraph
    LAGraph_Delete (&G, msg) ;
    GrB_free (&centrality) ;
    LAGraph_Finalize (msg) ;
}

Index