Merge pull request jlothian#24 from spowers/master

expansion; doubles; graph stats
MattBBaker · Jun 14, 2013 · 531a201 · 531a201
2 parents b079c45 + 5fa77e0
commit 531a201
Show file tree

Hide file tree

Showing 7 changed files with 138 additions and 14 deletions.
diff --git a/lib_graphd/inc/GraphProperties.h b/lib_graphd/inc/GraphProperties.h
@@ -85,7 +85,12 @@ namespace Graph {
     /**
      * \brief returns frequency distributions of eccentricities
      */
-    void eccentricity_dist(Graph *g, vector<int> &ecc, vector<float> &freq_ecc);
+    void eccentricity_dist(Graph *g, vector<int> &ecc, vector<double> &freq_ecc);
+
+    /**
+     * \brief returns normalized expansion (distance distribution)
+     */
+    void expansion(Graph *g, vector<double> &norm_hops);
 
     /**
      * \brief Calculates the diameter of the specified graph

diff --git a/lib_graphd/src/GraphProperties.cpp b/lib_graphd/src/GraphProperties.cpp
@@ -745,7 +745,7 @@ namespace Graph {
      * \param[in] ecc vector of eccentricities (empty or pre-computed)
      * \param[out] freq_ecc vector of eccentricity frequencies
      */
-    void GraphProperties::eccentricity_dist(Graph *g, vector<int> &ecc, vector<float> &freq_ecc){
+    void GraphProperties::eccentricity_dist(Graph *g, vector<int> &ecc, vector<double> &freq_ecc){
         const int n = g->get_num_nodes();
         int bestAll = 0;
 
@@ -768,6 +768,34 @@ namespace Graph {
         }
     } //eccentricity_dist
 
+    /**
+     * Calcuates the expansion (hop-distance) from each vertex: [see Tangmunarunkit (2002)]
+     * compute #nodes reachable in h hops starting at each vertex, average, normalize
+     * \param[in] g input graph
+     * \param[out] norm_hops normalized distribution
+     */
+    void GraphProperties::expansion(Graph *g, vector<double> &norm_hops){
+        const int n = g->get_num_nodes();
+        vector< vector<int> > short_paths = g->get_shortest_path_dist_ref();   //computes if not already
+        vector <int> hops(n,0);  //largest possible dist is n-1
+        norm_hops.resize(n);
+
+        //for each vertex, find the number of vertices reachable for all hops; add to tally
+        for(int i = 0; i < n; i++){
+            for(int j = 0; j < n; j++){
+                hops[ short_paths[i][j] ]++;
+            }
+        }
+
+        //average (divide by n) and normalize (divide by n-1)
+        norm_hops[0] = 0.0; //no one is reachable in 0 hops (not counting self)
+
+        for(int h = 1; h < n; h++){
+            norm_hops[h] = (double)hops[h] / (n * (n - 1));
+            //printf("h = %d and number is %d; norm value is %f\n",h,hops[h],norm_hops[h]);
+        }
+    } //expansion
+
     /**
      * Calculates the diameter of graph g
      * \param[in] g Pointer to a graph

diff --git a/util/Makefile b/util/Makefile
@@ -42,7 +42,11 @@ ECC_SRCS = ./src/v_eccentricity.cpp
 ECC_OBJ = $(ECC_SRCS:.cpp=.o)
 ECC_EXE = $(INDDGO_BIN)/v_eccentricity
 
-all: $(PKT_EXE) $(TD_STATS_EXE) $(E2D_EXE) $(E2A_EXE) $(M2D_EXE) $(G_STATS_EXE) $(TRI_EXE) $(SRT_EXE) $(ECC_EXE)
+EXP_SRCS = ./src/expansion.cpp
+EXP_OBJ = $(EXP_SRCS:.cpp=.o)
+EXP_EXE = $(INDDGO_BIN)/expansion
+
+all: $(PKT_EXE) $(TD_STATS_EXE) $(E2D_EXE) $(E2A_EXE) $(M2D_EXE) $(G_STATS_EXE) $(TRI_EXE) $(SRT_EXE) $(ECC_EXE) $(EXP_EXE)
 
 .cpp.o:
 	$(CXX) $(CPPFLAGS) $(CFLAGS) -c $(INC_DIR)  $< -o $@
@@ -74,8 +78,11 @@ $(SRT_EXE): $(SRT_OBJ) ../lib/libgraphd.a
 $(ECC_EXE): $(ECC_OBJ) ../lib/libgraphd.a
 	$(CXX) $(CPPFLAGS) $(CFLAGS) $^ $(INC_DIR) $(LIBS) $(LDFLAGS) -o $@
 
+$(EXP_EXE): $(EXP_OBJ) ../lib/libgraphd.a
+	$(CXX) $(CPPFLAGS) $(CFLAGS) $^ $(INC_DIR) $(LIBS) $(LDFLAGS) -o $@
+
 clean: 
-	-rm -rf $(PKT_OBJ) $(PKT_EXE) $(TD_STATS_OBJ) $(TD_STATS_EXE) $(E2D_OBJ) $(E2D_EXE) $(M2D_EXE) $(M2D_OBJ) $(G_STATS_EXE) $(G_STATS_OBJ) $(TRI_OBJ) $(TRI_EXE) $(SRT_OBJ) $(SRT_EXE) $(ECC_OBJ) $(ECC_EXE) $(E2A_EXE) $(E2A_OBJ)
+	-rm -rf $(PKT_OBJ) $(PKT_EXE) $(TD_STATS_OBJ) $(TD_STATS_EXE) $(E2D_OBJ) $(E2D_EXE) $(M2D_EXE) $(M2D_OBJ) $(G_STATS_EXE) $(G_STATS_OBJ) $(TRI_OBJ) $(TRI_EXE) $(SRT_OBJ) $(SRT_EXE) $(ECC_OBJ) $(ECC_EXE) $(EXP_OBJ) $(EXP_EXE) $(E2A_EXE) $(E2A_OBJ)
 
 doc: $(DOXYFILE)
 	doxygen $(DOXYFILE)

diff --git a/util/src/expansion.cpp b/util/src/expansion.cpp
@@ -0,0 +1,69 @@
+/*
+   This file is part of INDDGO.
+
+   Copyright (C) 2012, Oak Ridge National Laboratory
+
+   This product includes software produced by UT-Battelle, LLC under Contract No.
+   DE-AC05-00OR22725 with the Department of Energy.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the New BSD 3-clause software license (LICENSE).
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+   LICENSE for more details.
+
+   For more information please contact the INDDGO developers at:
+   [email protected]
+
+ */
+
+#include "GraphReader.h"
+#include "GraphProperties.h"
+#include <ctime>
+
+using namespace std;
+
+void usage(const char *s){
+    fprintf(stderr, "Usage: %s filename\n", s);
+}
+
+int main(int argc, char **argv){
+    //usage check
+    if((argc == 1) ||
+       ((argc == 2) && (strcmp(argv[1],"-h") == 0)) ||
+       ((argc == 2) && (strcmp(argv[1], "--help") == 0))){
+        usage(argv[0]);
+        exit(-1);
+    }
+
+    Graph::Graph *g;
+
+    clock_t begin, end;
+
+    Graph::GraphProperties prop;
+    Graph::GraphReader ngr;
+
+    //create the graph object
+    g = new Graph::Graph();
+
+    //read the graph from the filename, assume it is an edgelist
+    ngr.read_graph(g, argv[1], "Edge", false);
+    //ngr.read_graph(g, argv[1], "ADJLIST", false);
+    printf("Read %d vertices and %d edges\n", g->get_num_nodes(), g->get_num_edges());
+
+    printf("Simplifying graph\n");
+    begin = clock();
+    prop.make_simple(g);    //remove self loops and duplicate edges
+    end = clock();
+    printf("Time: %f\n", double(end - begin) / CLOCKS_PER_SEC);
+
+    //compute normalized expansion
+    begin = clock();
+    vector<double> norm_hops;
+    prop.expansion(g, norm_hops);
+    end = clock();
+    printf("Alg Time (expansion): %f\n", double(end - begin) / CLOCKS_PER_SEC);
+} // main
+
diff --git a/util/src/gen_pkt.cpp b/util/src/gen_pkt.cpp
@@ -97,7 +97,7 @@ int main(int argc, char **argv){
             ktree_p = atoi(argv[i + 1]);
         }
         if(strcmp(argv[i],"-o") == 0){
-            out_format = string(argv[i+1]);
+            out_format = string(argv[i + 1]);
         }
         if(strcmp(argv[i],"-s") == 0){
             seed = atoi(argv[i + 1]);
@@ -106,23 +106,23 @@ int main(int argc, char **argv){
             prefix = argv[i + 1];
         }
         if(strcmp(argv[i],"-fe") == 0){
-            prefix = argv[i+1];
+            prefix = argv[i + 1];
             exact_filename = true;
         }
-        if(strcmp(argv[i],"-m") ==0){
+        if(strcmp(argv[i],"-m") == 0){
             timings = true;
         }
         if(strcmp(argv[i],"-r") == 0){
             random = true;
         }
     }
 
-    if(true == exact_filename && t != 1 ){
+    if((true == exact_filename) && (t != 1)){
         cerr << "Error: cannot specify both -t and -fe flags\n";
         exit(1);
     }
 
-    if("adjlist" != out_format && "dimacs" != out_format){
+    if(("adjlist" != out_format) && ("dimacs" != out_format)){
         cerr << "Error: only allowed output formats are: adjlist, dimacs\n";
         exit(1);
     }
@@ -170,7 +170,7 @@ int main(int argc, char **argv){
         //write it to a file
         if(!exact_filename){
             sprintf(filename, "%s.%d.%d.%d_%d.dimacs", prefix, ktree_n,ktree_k,ktree_p, i);
-        } 
+        }
         else {
             strncpy(filename,prefix, 99);
         }

diff --git a/util/src/graph_stats.cpp b/util/src/graph_stats.cpp
@@ -41,7 +41,7 @@ void print_time(string prefix, ORB_t start, ORB_t end){
     cout << prefix + ": " << ORB_seconds(end, start) << "\n";
 }
 
-const string allowed_methods ("edge_density,avg_degree,degree_dist,global_cc,avg_cc,local_ccs,shortest_paths,assortativity");
+const string allowed_methods ("edge_density,avg_degree,degree_dist,global_cc,avg_cc,local_ccs,shortest_paths,assortativity,eccentricity,eccentricity_dist");
 
 /**
  * Creates a map from a comma-separated string
@@ -139,9 +139,9 @@ int main(int argc, char **argv){
     print_time("Time(read_graph)", t1, t2);
 
     double global_cc, avg_cc, assortativity;
-    vector<double> local_cc;
+    vector<double> local_cc, freq_ecc;
     float edge_density, avg_degree;
-    vector<int> deg_dist;
+    vector<int> deg_dist, ecc;
     vector< vector<int> > shortest_path_distances;
 
     outfile.open(outfilename.c_str());
@@ -218,6 +218,21 @@ int main(int argc, char **argv){
         ORB_read(t2);
         print_time("Time(shortest_paths_dijkstra)", t1, t2);
     }
+    if(req_methods["eccentricity"] == true){
+        cout << "Calculating eccentricities\n";
+        ORB_read(t1);
+        gp.eccentricity(&g, ecc);
+        ORB_read(t2);
+        print_time("Time(eccentricity)",t1,t2);
+    }
+    if(req_methods["eccentricity_dist"] == true){
+        cout << "Calculating distribution of eccentricities\n";
+        ORB_read(t1);
+        gp.eccentricity_dist(&g, ecc, freq_ecc);
+        ORB_read(t2);
+        print_time("Time(eccentricity distribution)",t1,t2);
+    }
+
     outfile.close();
     exit(0);
 } // main

diff --git a/util/src/v_eccentricity.cpp b/util/src/v_eccentricity.cpp
@@ -68,7 +68,7 @@ int main(int argc, char **argv){
 
     //compute eccentricity distribution
     begin = clock();
-    vector<float> ecc_dist;
+    vector<double> ecc_dist;
     prop.eccentricity_dist(g,ecc, ecc_dist);
     end = clock();
     printf("Alg Time (freq dist eccentricity): %f\n", double(end - begin) / CLOCKS_PER_SEC);