This is a library plus a test harness for collecting algorithms that use the GraphBLAS. For test coverage reports, see https://graphblas.org/LAGraph/ . Documentation: https://lagraph.readthedocs.org
Other
229
stars
61
forks
source link
Added type handling to LAGraph_Coarsen_Matching #203
Previously, the coarsening algorithm could overflow when combine_weights = true when two edges are combined during a coarsening step. In addition, matrices of type GrB_BOOL would not combine edge weights at all since all integers cast to true. To fix this, non-float input matrices are not used as is, instead a new GrB_INT64 matrix using their entries is built.
Added description of outputs to coarsening code
Fixed type of S matrix in LAGraph_Parent_to_S to be GrB_BOOL.
combine_weights = truewhen two edges are combined during a coarsening step. In addition, matrices of typeGrB_BOOLwould not combine edge weights at all since all integers cast totrue. To fix this, non-float input matrices are not used as is, instead a newGrB_INT64matrix using their entries is built.Smatrix inLAGraph_Parent_to_Sto beGrB_BOOL.