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
225
stars
59
forks
source link
Fixing roundoff errors in coarsening, incidence matrix construction #233
Previously, some operations in LAGraph_IncidenceMatrix and LAGraph_Coarsen_Matching were being done by casting large integers to floating point, which will cause potential roundoff errors. This PR will fix that by ensuring integers and floats cast only to other integers and floats respectively of at least the same size.
Also performed miscellaneous code cleanups. Importantly, removed GxB function calls and fixed error codes in LG_check_coarsen (which may be used for checking against any coarsening method), and made slight changes to the Coarsen_Matching API.
LAGraph_IncidenceMatrixandLAGraph_Coarsen_Matchingwere being done by casting large integers to floating point, which will cause potential roundoff errors. This PR will fix that by ensuring integers and floats cast only to other integers and floats respectively of at least the same size.