adding louvain partitioning implementation

Author: Udit Agarwal

lonestar/analytics/cpu/CMakeLists.txt

@@ -15,3 +15,5 @@ add_subdirectory(pointstoanalysis)
 add_subdirectory(preflowpush)
 add_subdirectory(sssp)
 add_subdirectory(triangle-counting)
+
+add_subdirectory(louvain-partitioning)

lonestar/analytics/cpu/louvain-partitioning/CMakeLists.txt

@@ -0,0 +1,5 @@
+add_executable(louvain-partitioning-cpu louvain-partitioning.cpp serial-partitioning.cpp)
+add_dependencies(apps louvain-partitioning-cpu)
+target_link_libraries(louvain-partitioning-cpu PRIVATE Galois::shmem lonestar)
+install(TARGETS louvain-partitioning-cpu DESTINATION "${CMAKE_INSTALL_BINDIR}" COMPONENT apps EXCLUDE_FROM_ALL)
+add_test_scale(small1 louvain-partitioning-cpu -symmetricGraph "${BASEINPUT}/scalefree/symmetric/rmat10.sgr")

lonestar/analytics/cpu/louvain-partitioning/README.md

@@ -0,0 +1,28 @@
+Partitioning using Louvain
+================================================================================
+
+DESCRIPTION 
+--------------------------------------------------------------------------------
+
+This directory contains a graph partitioning algorithm that uses a hierarchical community detection algorithm for coarsening. The algorithm uses the well-known multi-level partitioning approach, similar to GMetis.
+
+INPUT
+--------------------------------------------------------------------------------
+
+This application takes in symmetric Galois .gr graphs.
+You must specify the -symmetricGraph flag when running this benchmark.
+
+BUILD
+--------------------------------------------------------------------------------
+
+1. Run cmake at BUILD directory (refer to top-level README for cmake instructions).
+
+2. Run `cd <BUILD>/lonestar/analytics/cpu/louvain_partitioning; make -j`
+
+RUN
+--------------------------------------------------------------------------------
+
+The following is an example command line.
+
+-`$  ./louvain-partitioning-cpu -input <path-to-graph> -t 4 -tolerance=15 -threshold=0.000001 -max_iter 1000  -symmetricGraph`
+