support for ternary

Author: N-Maas

mt-kahypar/application/mt_kahypar.cc

@@ -125,10 +125,16 @@ int main(int argc, char* argv[]) {
   // Read Frequencies
   vec<EdgeMetadata> edge_metadata;
   if ( context.partition.frequencies_filename != "" ) {
-    edge_metadata = io::getEdgeMetadataFromFile(hypergraph, context.partition.frequencies_filename, context.coarsening.rating.guiding_binary_treshold);
+    edge_metadata = io::getEdgeMetadataFromFile(hypergraph, context.partition.frequencies_filename,
+                                                context.coarsening.rating.guiding_binary_treshold,
+                                                context.coarsening.rating.guiding_ternary_treshold,
+                                                context.coarsening.rating.guiding_ternary_value);
   } else if ( context.partition.frequencies_default_file ) {
     std::string frequencies_file = context.partition.graph_filename + ".freq.csv";
-    edge_metadata = io::getEdgeMetadataFromFile(hypergraph, frequencies_file, context.coarsening.rating.guiding_binary_treshold);
+    edge_metadata = io::getEdgeMetadataFromFile(hypergraph, frequencies_file,
+                                                context.coarsening.rating.guiding_binary_treshold,
+                                                context.coarsening.rating.guiding_ternary_treshold,
+                                                context.coarsening.rating.guiding_ternary_value);
   }
 
   if ( context.partition.fixed_vertex_filename != "" ) {

mt-kahypar/io/command_line_options.cpp

@@ -378,6 +378,14 @@ namespace mt_kahypar {
              po::value<double>(&context.coarsening.rating.guiding_binary_treshold)->value_name(
                      "<double>")->default_value(-1.0),
              "Binary treshold for guided coarsening (e.g. by ML with classification).")
+            ("c-guiding-ternary-threshold",
+             po::value<double>(&context.coarsening.rating.guiding_ternary_treshold)->value_name(
+                     "<double>")->default_value(-1.0),
+             "Ternary treshold for guided coarsening (e.g. by ML with classification).")
+            ("c-guiding-ternary-value",
+             po::value<double>(&context.coarsening.rating.guiding_ternary_value)->value_name(
+                     "<double>")->default_value(-1.0),
+             "Value used as ternary between 0 and 1 when using ternary threshold.")
             ("c-guiding-threshold-max",
              po::value<double>(&context.coarsening.rating.guiding_treshold_max)->value_name(
                      "<double>")->default_value(0.8),

mt-kahypar/io/hypergraph_factory.cpp

@@ -293,9 +293,11 @@ vec<EdgeMetadata> getEdgeMetadata(const Hypergraph& hypergraph,
 
 vec<EdgeMetadata> getEdgeMetadataFromFile(mt_kahypar_hypergraph_t hypergraph,
                                           const std::string& filename,
-                                          double binary_threshold) {
+                                          double binary_threshold,
+                                          double ternary_threshold,
+                                          double ternary_value) {
   ds::DynamicSparseMap<__uint128_t, float> frequencies;
-  io::readFrequencyFile(filename, frequencies, binary_threshold);
+  io::readFrequencyFile(filename, frequencies, binary_threshold, ternary_threshold, ternary_value);
   switch ( hypergraph.type ) {
     // case STATIC_HYPERGRAPH:
     //   return getEdgeMetadata(utils::cast<ds::StaticHypergraph>(hypergraph), frequencies);

mt-kahypar/io/hypergraph_factory.h

@@ -63,7 +63,9 @@ void removeFixedVertices(mt_kahypar_hypergraph_t hypergraph);
 
 vec<EdgeMetadata> getEdgeMetadataFromFile(mt_kahypar_hypergraph_t hypergraph,
                                           const std::string& filename,
-                                          double binary_threshold);
+                                          double binary_threshold,
+                                          double ternary_threshold,
+                                          double ternary_value);
 
 }  // namespace io
 }  // namespace mt_kahypar

mt-kahypar/io/hypergraph_io.cpp

@@ -681,7 +681,11 @@ namespace mt_kahypar::io {
     }
   }
 
-  void readFrequencyFile(const std::string& filename, ds::DynamicSparseMap<__uint128_t, float>& frequencies, double binary_threshold) {
+  void readFrequencyFile(const std::string& filename,
+                         ds::DynamicSparseMap<__uint128_t, float>& frequencies,
+                         double binary_threshold,
+                         double ternary_threshold,
+                         double ternary_value) {
     ALWAYS_ASSERT(!filename.empty(), "No filename for frequency file specified");
     std::ifstream file(filename);
     if (file) {
@@ -716,7 +720,13 @@ namespace mt_kahypar::io {
           ALWAYS_ASSERT(!frequencies.contains(key) && f <= max);
           frequencies[key] = f / static_cast<double>(max);
           if (binary_threshold >= 0) {
-            frequencies[key] = frequencies[key] > binary_threshold ? 1.0 : 0.0;
+            if (ternary_threshold >= 0) {
+              ALWAYS_ASSERT(binary_threshold < ternary_threshold && ternary_threshold < 1);
+              const double mid = ternary_value >= 0 ? ternary_value : ternary_threshold;
+              frequencies[key] = frequencies[key] > binary_threshold ? (frequencies[key] > ternary_threshold ? 1.0 : mid) : 0.0;
+            } else {
+              frequencies[key] = frequencies[key] > binary_threshold ? 1.0 : 0.0;
+            }
           }
       }
       file.close();

mt-kahypar/io/hypergraph_io.h

@@ -59,7 +59,11 @@ namespace io {
   void readPartitionFile(const std::string& filename, HypernodeID num_nodes, std::vector<PartitionID>& partition);
   void readPartitionFile(const std::string& filename, HypernodeID num_nodes, PartitionID* partition);
 
-  void readFrequencyFile(const std::string& filename, ds::DynamicSparseMap<__uint128_t, float>& frequencies, double binary_threshold);
+  void readFrequencyFile(const std::string& filename,
+                         ds::DynamicSparseMap<__uint128_t, float>& frequencies,
+                         double binary_threshold,
+                         double ternary_threshold,
+                         double ternary_value);
 
   inline __uint128_t hashedEdgeKey(const utils_tm::hash_tm::murmur2_hash& hasher, HypernodeID u, HypernodeID v) {
     // this is a bit ugly, but we need it to avoid collisions (since we don't have a good hashmap with a non-trivial hash function)

mt-kahypar/partition/coarsening/policies/rating_degree_similarity_policy.h

@@ -364,7 +364,7 @@ class GuidedCoarseningPolicy final : public kahypar::meta::PolicyBase {
         sum += value;
         break;
       case GuidedEdgeAccumulation::quadratic:
-        sum += value / static_cast<double>(edge_weight) * value;
+        sum += value == 0 ? 0 : value / static_cast<double>(edge_weight) * value;
         break;
       case GuidedEdgeAccumulation::max:
         sum = std::max(sum, value / static_cast<float>(edge_weight));

mt-kahypar/partition/context.h

@@ -122,6 +122,8 @@ struct RatingParameters {
   double guiding_treshold = 1.0;
   double guiding_treshold_max = 1.0;
   double guiding_binary_treshold = -1.0;
+  double guiding_ternary_treshold = -1.0;
+  double guiding_ternary_value = -1.0;
   size_t num_guided_subrounds = 1;
   size_t guided_coarsening_levels = std::numeric_limits<size_t>::max();
   bool consider_edges_deleted = false;