// Test edge subgraph conversion
TEST(subgraph, TestConvertESubGraphs) {
PNEGraph NEGraph;
PNGraph NGraph;
TIntV NIdV;
TIntV EIdV;
int i;
NGraph = GetTestTNGraph();
EXPECT_EQ(20,NGraph->GetNodes());
EXPECT_EQ(60,NGraph->GetEdges());
for (i = 0; i < 20; i += 2) {
NIdV.Add(i);
}
// TODO: fix TSnap::ConvertSubGraph<PUNGraph>(NGraph, NIdV, true), it fails
// UNGraph = TSnap::ConvertSubGraph<PUNGraph>(NGraph, NIdV, true);
NEGraph = TSnap::ConvertGraph<PNEGraph>(NGraph);
EXPECT_EQ(20,NEGraph->GetNodes());
EXPECT_EQ(60,NEGraph->GetEdges());
// select every second edge
i = 0;
for (TNEGraph::TEdgeI EI = NEGraph->BegEI(); EI < NEGraph->EndEI(); EI++) {
if (i == 0) {
EIdV.Add(EI.GetId());
}
i = (i + 1) % 2;
}
NGraph = TSnap::ConvertESubGraph<PNGraph>(NEGraph, EIdV);
EXPECT_EQ(20,NGraph->GetNodes());
EXPECT_EQ(30,NGraph->GetEdges());
}
void OnlyD3CEdges(PNGraph& dir_graph, PNGraph& d3c_graph, bool recip_edges) {
// Add all of the nodes into the new graph
for (TNGraph::TNodeI node = dir_graph->BegNI(); node < dir_graph->EndNI();
node++) {
int curr_node = node.GetId();
d3c_graph->AddNode(curr_node);
}
for (TNGraph::TNodeI node = dir_graph->BegNI(); node < dir_graph->EndNI();
node++) {
int curr_node = node.GetId();
auto curr_node_it = dir_graph->GetNI(curr_node);
for (int out_edge = 0; out_edge < curr_node_it.GetOutDeg(); ++out_edge) {
int out_node = curr_node_it.GetOutNId(out_edge);
for (int in_edge = 0; in_edge < curr_node_it.GetInDeg(); ++in_edge) {
int in_node = curr_node_it.GetInNId(in_edge);
if (out_node == in_node && !recip_edges) { continue; }
if (dir_graph->IsEdge(out_node, in_node) || recip_edges) {
if (!d3c_graph->IsEdge(out_node, in_node)) {
d3c_graph->AddEdge(out_node, in_node);
}
if (!d3c_graph->IsEdge(in_node, curr_node)) {
d3c_graph->AddEdge(in_node, curr_node);
}
if (!d3c_graph->IsEdge(curr_node, out_node)) {
d3c_graph->AddEdge(curr_node, out_node);
}
}
}
}
}
#ifdef _VERBOSE_
std::cout << "Original graph edge count: " << dir_graph->GetEdges() << std::endl
<< "D3C graph edge count: " << d3c_graph->GetEdges() << std::endl;
#endif
}
// Test node subgraph conversion
void TestConvertSubGraphs() {
PNGraph NGraph;
PUNGraph UNGraph;
int N1, N2, N3;
int E1, E2, E3;
TIntV NIdV;
int i;
NGraph = GetTestTNGraph();
N1 = NGraph->GetNodes();
E1 = NGraph->GetEdges();
for (i = 0; i < 20; i += 2) {
NIdV.Add(i);
}
// TODO: fix TSnap::ConvertSubGraph<PUNGraph>(NGraph, NIdV, true), it fails
// UNGraph = TSnap::ConvertSubGraph<PUNGraph>(NGraph, NIdV, true);
UNGraph = TSnap::ConvertSubGraph<PUNGraph>(NGraph, NIdV);
N2 = UNGraph->GetNodes();
E2 = UNGraph->GetEdges();
NGraph = TSnap::ConvertSubGraph<PNGraph>(UNGraph, NIdV);
N3 = NGraph->GetNodes();
E3 = NGraph->GetEdges();
printf("---- TestConvertSubGraphs -----\n");
printf("nodes: %d,%d,%d, edges: %d,%d,%d\n", N1, N2, N3, E1, E2, E3);
printf("\n");
}
int main(int argc, char* argv[]) {
Env = TEnv(argc, argv, TNotify::StdNotify);
Env.PrepArgs(TStr::Fmt("Node Centrality. build: %s, %s. Time: %s", __TIME__, __DATE__, TExeTm::GetCurTm()));
TExeTm ExeTm;
Try
const TStr InFNm = Env.GetIfArgPrefixStr("-i:", "../as20graph.txt", "Input un/directed graph");
const TStr OutFNm = Env.GetIfArgPrefixStr("-o:", "node_centrality.tab", "Output file");
printf("Loading %s...", InFNm.CStr());
PNGraph Graph = TSnap::LoadEdgeList<PNGraph>(InFNm);
//PNGraph Graph = TSnap::GenRndGnm<PNGraph>(10, 10);
//TGraphViz::Plot(Graph, gvlNeato, InFNm+".gif", InFNm, true);
printf("nodes:%d edges:%d\n", Graph->GetNodes(), Graph->GetEdges());
PUNGraph UGraph = TSnap::ConvertGraph<PUNGraph>(Graph); // undirected version of the graph
TIntFltH BtwH, EigH, PRankH, CcfH, CloseH, HubH, AuthH;
//printf("Computing...\n");
printf("Treat graph as DIRECTED: ");
printf(" PageRank... "); TSnap::GetPageRank(Graph, PRankH, 0.85);
printf(" Hubs&Authorities..."); TSnap::GetHits(Graph, HubH, AuthH);
printf("\nTreat graph as UNDIRECTED: ");
printf(" Eigenvector..."); TSnap::GetEigenVectorCentr(UGraph, EigH);
printf(" Clustering..."); TSnap::GetNodeClustCf(UGraph, CcfH);
printf(" Betweenness (SLOW!)..."); TSnap::GetBetweennessCentr(UGraph, BtwH, 1.0);
printf(" Constraint (SLOW!)..."); TNetConstraint<PUNGraph> NetC(UGraph, true);
printf(" Closeness (SLOW!)...");
for (TUNGraph::TNodeI NI = UGraph->BegNI(); NI < UGraph->EndNI(); NI++) {
const int NId = NI.GetId();
CloseH.AddDat(NId, TSnap::GetClosenessCentr<PUNGraph>(UGraph, NId, false));
}
printf("\nDONE! saving...");
FILE *F = fopen(OutFNm.CStr(), "wt");
fprintf(F,"#Network: %s\n", InFNm.CStr());
fprintf(F,"#Nodes: %d\tEdges: %d\n", Graph->GetNodes(), Graph->GetEdges());
fprintf(F,"#NodeId\tDegree\tCloseness\tBetweennes\tEigenVector\tNetworkConstraint\tClusteringCoefficient\tPageRank\tHubScore\tAuthorityScore\n");
for (TUNGraph::TNodeI NI = UGraph->BegNI(); NI < UGraph->EndNI(); NI++) {
const int NId = NI.GetId();
const double DegCentr = UGraph->GetNI(NId).GetDeg();
const double CloCentr = CloseH.GetDat(NId);
const double BtwCentr = BtwH.GetDat(NId);
const double EigCentr = EigH.GetDat(NId);
const double Constraint = NetC.GetNodeC(NId);
const double ClustCf = CcfH.GetDat(NId);
const double PgrCentr = PRankH.GetDat(NId);
const double HubCentr = HubH.GetDat(NId);
const double AuthCentr = AuthH.GetDat(NId);
fprintf(F, "%d\t%f\t%f\t%f\t%f\t%f\t%f\t%f\t%f\t%f\n", NId,
DegCentr, CloCentr, BtwCentr, EigCentr, Constraint, ClustCf, PgrCentr, HubCentr, AuthCentr);
}
fclose(F);
Catch
printf("\nrun time: %s (%s)\n", ExeTm.GetTmStr(), TSecTm::GetCurTm().GetTmStr().CStr());
return 0;
}
void PlotSngVec(const PNGraph& Graph, const TStr& FNmPref, TStr DescStr) {
TFltV LeftSV, RightSV;
TSnap::GetSngVec(Graph, LeftSV, RightSV);
LeftSV.Sort(false);
RightSV.Sort(false);
TFltV BinV;
if (DescStr.Empty()) { DescStr = FNmPref; }
TGUtil::MakeExpBins(LeftSV, BinV, 1.01);
TGnuPlot::PlotValV(BinV, "sngVecL."+FNmPref, TStr::Fmt("%s. G(%d, %d). Left signular vector",
DescStr.CStr(), Graph->GetNodes(), Graph->GetEdges()), "Rank", "Component of left singular vector", gpsLog10XY, false, gpwLinesPoints);
TGnuPlot::PlotValV(BinV, "sngVecL."+FNmPref, TStr::Fmt("%s. G(%d, %d). Right signular vector",
DescStr.CStr(), Graph->GetNodes(), Graph->GetEdges()), "Rank", "Component of right singular vector", gpsLog10XY, false, gpwLinesPoints);
}
void TGraphKey::TakeSig(const PNGraph& Graph, const int& MnSvdGraph, const int& MxSvdGraph) {
const int Edges = Graph->GetEdges();
Nodes = Graph->GetNodes();
VariantId = 0;
SigV.Gen(2+Nodes, 0);
// degree sequence
TIntPrV DegV(Nodes, 0);
for (TNGraph::TNodeI NodeI = Graph->BegNI(); NodeI < Graph->EndNI(); NodeI++) {
DegV.Add(TIntPr(NodeI.GetInDeg(), NodeI.GetOutDeg()));
}
DegV.Sort(false);
SigV.Add(TFlt(Nodes));
SigV.Add(TFlt(Edges));
for (int i = 0; i < DegV.Len(); i++) {
SigV.Add(DegV[i].Val1());
SigV.Add(DegV[i].Val2());
}
// singular values signature
// it turns out that it is cheaper to do brute force isomorphism
// checking than to calculate SVD and then check isomorphism
if (Nodes >= MnSvdGraph && Nodes < MxSvdGraph) {
// perform full SVD
TFltVV AdjMtx(Nodes+1, Nodes+1);
TFltV SngValV;
TFltVV LSingV, RSingV;
TIntH NodeIdH;
// create adjecency matrix
for (TNGraph::TNodeI NodeI = Graph->BegNI(); NodeI < Graph->EndNI(); NodeI++) {
NodeIdH.AddKey(NodeI.GetId());
}
for (TNGraph::TNodeI NodeI = Graph->BegNI(); NodeI < Graph->EndNI(); NodeI++) {
const int NodeId = NodeIdH.GetKeyId(NodeI.GetId()) + 1;
for (int e = 0; e < NodeI.GetOutDeg(); e++) {
const int DstNId = NodeIdH.GetKeyId(NodeI.GetOutNId(e)) + 1; // no self edges
if (NodeId != DstNId) AdjMtx.At(NodeId, DstNId) = 1;
}
}
try { // can fail to converge but results seem to be good
TSvd::Svd(AdjMtx, LSingV, SngValV, RSingV);
} catch(...) {
printf("\n***No SVD convergence: G(%d, %d): SngValV.Len():%d\n", Nodes(), Graph->GetEdges(), SngValV.Len());
}
// round singular values
SngValV.Sort(false);
for (int i = 0; i < SngValV.Len(); i++) {
SigV.Add(TMath::Round(SngValV[i], RoundTo));
}
}
//printf("SIG:\n"); for (int i = 0; i < SigV.Len(); i++) { printf("\t%f\n", SigV[i]); }
SigV.Pack();
}
void GetGraphs(const vector <TStr>& Parameters, const TStr& ModelGen, const TStr&ModelPlt)
{
PNGraph G;
size_t PSize = Parameters.size();
if (GRAPHGEN >= PSize || MTXGEN >= PSize || KRONGEN >= PSize || KRONFIT >= PSize)
Error("GetGraphs", "Wrong index in array of parameters");
GetModel(Parameters[GRAPHGEN], G);
if (G->GetNodes() == 0)
Error("GetGraphs", "Empty graph");
TFltPrV MDegIn, MDegOut;
TSnap::GetInDegCnt(G, MDegIn);
TSnap::GetOutDegCnt(G, MDegOut);
PlotDegrees(Parameters, MDegIn, MDegOut, "model");
TFile << "Model nodes: " << G->GetNodes() << ", model edges: " << G->GetEdges() << endl;
TFile << "Maximum output degree in model graph: " << MDegOut[MDegOut.Len()-1].GetVal1() << endl;
TFile << "Maximum input degree in model graph: " << MDegIn[MDegIn.Len()-1].GetVal1() << endl;
if (ModelGen == "model+kron"){
// generate (or read) Kronecker initiator matrix
TKronMtx FitMtxM;
if (!GetMtx(Parameters[MTXGEN], FitMtxM))
GenNewMtx(G, Parameters[KRONFIT], FitMtxM);
PrintMtx(FitMtxM, TFile);
TFile << "Scaling for the number of edges... " << endl;
FitMtxM.SetForEdges(G->GetNodes(), G->GetEdges());
int ModelNodes = G->GetNodes(), ModelEdges = G->GetEdges();
Env = TEnv(Parameters[KRONGEN], TNotify::NullNotify);
TStr IsDir = Env.GetIfArgPrefixStr("-isdir:", "false", "Produce directed graph (true, false)");
const TInt NIter = Env.GetIfArgPrefixInt("-i:", 1, "Number of iterations of Kronecker product");
if (pow(FitMtxM.GetDim(), static_cast<double>(NIter)) != ModelNodes)
Error("GetGraphs", "Inconsistent value of -i: parameter, KronNodes != ModelNodes");
// in and out average degrees of Kronecker graphs
TFltPrV KronDegAvgIn, KronDegAvgOut;
GenKron(Parameters[KRONGEN], FitMtxM, KronDegAvgIn, KronDegAvgOut);
PlotDegrees(Parameters, KronDegAvgIn, KronDegAvgOut, "kron");
}
}
double DirectedModularity(PNGraph& graph, std::vector<int>& communities) {
if (graph->GetNodes() != communities.size()) {
throw std::logic_error("Number of nodes does not match community size.");
}
int num_edges = graph->GetEdges();
double score = 0.0;
int num_unique = 10;
std::map<int, double> outdeg_sums;
std::map<int, double> indeg_sums;
for (TNGraph::TNodeI node = graph->BegNI(); node < graph->EndNI(); node++) {
int comm = communities[node.GetId()];
outdeg_sums[comm] += node.GetOutDeg();
indeg_sums[comm] += node.GetInDeg();
}
for (auto& kv : outdeg_sums) {
score -= (kv.second / num_edges) * indeg_sums[kv.first];
}
for (TNGraph::TNodeI node = graph->BegNI(); node < graph->EndNI(); node++) {
int node_ID = node.GetId();
for (int e = 0; e < node.GetOutDeg(); ++e) {
int nbr = node.GetOutNId(e);
if (communities[node_ID] == communities[nbr]) {
score += 1.0;
}
}
}
return score / num_edges;
}
void analyzeSimNetProps() {
const char *eFName = "epidemicRoutingNetEdges.txt";
const char *pFName = "prophetRoutingNetEdges.txt";
PNGraph eGraph = TSnap::LoadEdgeListStr<PNGraph>(eFName, 0, 1);
PNEGraph pGraph = TSnap::LoadEdgeListStr<PNEGraph>(pFName, 0, 1);
PNGraph randGraph = TSnap::GenRndGnm<PNGraph>(eGraph->GetNodes(), eGraph->GetEdges(), true, TInt::Rnd);
chdir("dot");
for (int i=0; i<10; i++) {
TIntV NIdV;
for (int j = 0; j < 10; j++) {
int randNode = eGraph->GetRndNId();
NIdV.AddUnique(randNode);
}
// Plot the mesage propagtion in Endroy-Renyi graphs
PNGraph randFlow = TSnap::GetSubGraph<PNGraph>(randGraph, NIdV);
char randf[50]; sprintf(randf,"%d-erdos.dot",i);
TSnap::SaveGViz(randFlow, randf, TStr("Edros-Renyi random graph"));
// Now plot epidemic routing
PNGraph epidemicFlow = TSnap::GetSubGraph<PNGraph>(eGraph, NIdV);
char epf[50]; sprintf(epf,"%d-epidemic.dot",i);
TSnap::SaveGViz(epidemicFlow, epf, TStr("Epidemic routing"));
}
}
void TGStat::TakeStat(const PNGraph& Graph, const TSecTm& _Time, TFSet StatFSet, const TStr& GraphName) {
printf("\n===TakeStat: G(%u, %u)\n", Graph->GetNodes(), Graph->GetEdges());
TExeTm ExeTm, FullTm;
Time = _Time;
GraphNm = GraphName;
if (StatFSet.In(gsvNone)) { return; }
TakeBasicStat(Graph, false);
TakeDiam(Graph, StatFSet, false);
if (StatFSet.In(gsdWcc) || StatFSet.In(gsdWccHops) || StatFSet.In(gsvFullDiam) || StatFSet.In(gsvEffWccDiam)) {
PNGraph WccGraph = TSnap::GetMxWcc(Graph);
TakeBasicStat(WccGraph, true);
TakeDiam(WccGraph, StatFSet, true);
}
// degrees
TakeDegDistr(Graph, StatFSet);
// components
TakeConnComp(Graph, StatFSet);
// spectral
TakeSpectral(Graph, StatFSet, -1);
// clustering coeffient
if (StatFSet.In(gsdClustCf) || StatFSet.In(gsvClustCf)) {
TakeClustCf(Graph); }
if (StatFSet.In(gsdTriadPart)) {
TakeTriadPart(Graph); }
printf(" [%s]\n", FullTm.GetTmStr());
}
void GetSngVec(const PNGraph& Graph, TFltV& LeftSV, TFltV& RightSV) {
const int Nodes = Graph->GetNodes();
TFltVV LSingV, RSingV;
TFltV SngValV;
if (Nodes < 500) {
// perform full SVD
TFltVV AdjMtx(Nodes+1, Nodes+1);
TIntH NodeIdH;
// create adjecency matrix
for (TNGraph::TNodeI NodeI = Graph->BegNI(); NodeI < Graph->EndNI(); NodeI++) {
NodeIdH.AddKey(NodeI.GetId()); }
for (TNGraph::TNodeI NodeI = Graph->BegNI(); NodeI < Graph->EndNI(); NodeI++) {
const int NodeId = NodeIdH.GetKeyId(NodeI.GetId()) + 1;
for (int e = 0; e < NodeI.GetOutDeg(); e++) {
const int DstNId = NodeIdH.GetKeyId(NodeI.GetOutNId(e)) + 1; // no self edges
if (NodeId != DstNId) AdjMtx.At(NodeId, DstNId) = 1;
}
}
try { // can fail to converge but results seem to be good
TSvd::Svd1Based(AdjMtx, LSingV, SngValV, RSingV); }
catch(...) {
printf("\n***No SVD convergence: G(%d, %d)\n", Nodes, Graph->GetEdges()); }
} else { // Lanczos
TNGraphMtx GraphMtx(Graph);
TSparseSVD::LanczosSVD(GraphMtx, 1, 8, ssotFull, SngValV, LSingV, RSingV);
}
TFlt MxSngVal = TFlt::Mn;
int ValN = 0;
for (int i = 0; i < SngValV.Len(); i++) {
if (MxSngVal < SngValV[i]) { MxSngVal = SngValV[i]; ValN = i; } }
LSingV.GetCol(ValN, LeftSV);
RSingV.GetCol(ValN, RightSV);
IsAllValVNeg(LeftSV, true);
IsAllValVNeg(RightSV, true);
}
void analyzeSimNetProps(TStr messGraph, TStr netGraph) {
// Make graphs
PNGraph eGraph = TSnap::LoadEdgeListStr<PNGraph>(netGraph, 0, 1);
PNGraph mGraph = TSnap::LoadEdgeListStr<PNGraph>(messGraph, 0, 1);
PNGraph randGraph = TSnap::GenRndGnm<PNGraph>(mGraph->GetNodes(), mGraph->GetEdges(), true, TInt::Rnd);
// Induce network graph from the entire network based on message graph
TIntV NIdV;
for (TNGraph::TNodeI NI = mGraph->BegNI(); NI < mGraph->EndNI(); NI++)
NIdV.AddUnique(NI.GetId());
PNGraph indGraph = TSnap::GetSubGraph<PNGraph>(eGraph, NIdV);
//printf("%s:: ", messGraph.CStr()); printf("nodes %d; ", indGraph->GetNodes());
//printf("induced edges %d, random edges %d\n", indGraph->GetEdges(), randGraph->GetEdges());
printf("%d\t%d\t%d\n", indGraph->GetNodes(), indGraph->GetEdges(), mGraph->GetEdges());
}
// Test graph conversion
TEST(subgraph, TestConvertGraphs) {
PNGraph NGraph;
PUNGraph UNGraph;
NGraph = GetTestTNGraph();
EXPECT_EQ(20,NGraph->GetNodes());
EXPECT_EQ(60,NGraph->GetEdges());
UNGraph = TSnap::ConvertGraph<PUNGraph>(NGraph);
EXPECT_EQ(20,UNGraph->GetNodes());
EXPECT_EQ(60,UNGraph->GetEdges());
NGraph = TSnap::ConvertGraph<PNGraph>(UNGraph);
EXPECT_EQ(20,NGraph->GetNodes());
EXPECT_EQ(120,NGraph->GetEdges());
}
void SaveAndPlot(const PNGraph& G, const TStr& name, bool isCum){
TFltPrV in, out;
TSnap::GetInDegCnt(G, in);
TSnap::GetOutDegCnt(G, out);
int nodes = G->GetNodes(), edges = G->GetEdges();
TSnap::PlotDegDistr(in, nodes, edges, name, name, isCum, false, true);
TSnap::PlotDegDistr(out, nodes, edges, name, name, isCum, false, false);
}
void PlotSngValRank(const PNGraph& Graph, const int& SngVals, const TStr& FNmPref, TStr DescStr) {
TFltV SngValV;
TSnap::GetSngVals(Graph, SngVals, SngValV);
SngValV.Sort(false);
if (DescStr.Empty()) { DescStr = FNmPref; }
TGnuPlot::PlotValV(SngValV, "sngVal."+FNmPref, TStr::Fmt("%s. G(%d, %d). Largest eig val = %f",
DescStr.CStr(), Graph->GetNodes(), Graph->GetEdges(), SngValV[0].Val), "Rank", "Singular value", gpsLog10XY, false, gpwLinesPoints);
}
void graphson(PNGraph Graph){
std::ofstream graph;
graph.open("graph.json");
graph << "{\n";
graph << " \"graph\": {\n";
graph << " " << "\"mode\": \"NORMAL\", \n";
graph << " " << "\"vertices\": [\n";
int i = 0;
for (TNGraph::TNodeI NI = Graph->BegNI(); NI < Graph->EndNI(); NI++) {
i++;
graph << " " << "{\n";
graph << " " << "\"_id\": \"" << NI.GetId() << "\",\n";
graph << " " << "\"_type\": \"vertex\"\n";
if (i == Graph->GetNodes()) {
graph << " " << "}\n";
} else {
graph << " " << "},\n";
}
}
graph << " " << "],\n";
graph << " " << "\"edges\": [\n";
i = 0;
printf("Edges: %d", Graph->GetEdges());
for (TNGraph::TEdgeI EI = Graph->BegEI(); EI < Graph->EndEI(); EI++) {
graph << " " << "{\n"
<< " " << "\"_id\": \"" << i++ << "\",\n"
<< " " << "\"_type\": \"edge\"\n"
<< " " << "\"_outV\": \"" << EI.GetSrcNId() << "\"\n"
<< " " << "\"_inV\": \""<< EI.GetDstNId() << "\"\n";
if (i == Graph->GetEdges())
graph << " " << "}\n";
else
graph << " " << "},\n";
}
graph << " " << "]\n";
graph << " " << "}\n";
graph << "}\n";
graph.close();
}
void PrintGraphStat(const PNGraph& G) {
PNGraph WCC = TSnap::GetMxWcc(G);
PNGraph SCC = TSnap::GetMxScc(G);
TFltPrV DegCCfV;
int ClosedTriads, OpenTriads, FullDiam;
double EffDiam;
printf("Nodes\t%d\n", G->GetNodes());
printf("Edges\t%d\n", G->GetEdges());
printf("Nodes in largest WCC\t%d (%.3f)\n", WCC->GetNodes(), WCC->GetNodes()/double(G->GetNodes()));
printf("Edges in largest WCC\t%d (%.3f)\n", WCC->GetEdges(), WCC->GetEdges()/double(G->GetEdges()));
printf("Nodes in largest SCC\t%d (%.3f)\n", SCC->GetNodes(), SCC->GetNodes()/double(G->GetNodes()));
printf("Edges in largest SCC\t%d (%.3f)\n", SCC->GetEdges(), SCC->GetEdges()/double(G->GetEdges()));
const double CCF = TSnap::GetClustCf(G, DegCCfV, ClosedTriads, OpenTriads);
printf("Average clustering coefficient\t%.4f\n", CCF);
printf("Number of triangles\t%d\n", ClosedTriads);
printf("Fraction of closed triangles\t%.4g\n", ClosedTriads/double(ClosedTriads+OpenTriads));
TSnap::GetBfsEffDiam(G, 1000, false, EffDiam, FullDiam);
printf("Diameter (longest shortest path)\t%d\n", FullDiam);
printf("90-percentile effective diameter\t%.2g\n", EffDiam);
}
int main(int argc, char* argv[]) {
Env = TEnv(argc, argv, TNotify::StdNotify);
Env.PrepArgs(TStr::Fmt("Motifs. build: %s, %s. Time: %s", __TIME__, __DATE__, TExeTm::GetCurTm()));
TExeTm ExeTm;
Try
const TStr InFNm = Env.GetIfArgPrefixStr("-i:", "../as20graph.txt", "Input directed graph file (single directed edge per line)");
const int MotifSz = Env.GetIfArgPrefixInt("-m:", 3, "Motif size (has to be 3 or 4)");
const bool DrawMotifs = Env.GetIfArgPrefixBool("-d:", true, "Draw motif shapes (requires GraphViz)");
TStr OutFNm = Env.GetIfArgPrefixStr("-o:", "", "Output file prefix");
if (OutFNm.Empty()) { OutFNm = InFNm.GetFMid(); }
EAssert(MotifSz==3 || MotifSz==4);
// load graph
PNGraph G;
if (InFNm.GetFExt().GetLc()==".ungraph") {
TFIn FIn(InFNm); G=TSnap::ConvertGraph<PNGraph>(TUNGraph::Load(FIn), true); }
else if (InFNm.GetFExt().GetLc()==".ngraph") {
TFIn FIn(InFNm); G=TNGraph::Load(FIn); }
else {
G = TSnap::LoadEdgeList<PNGraph>(InFNm, 0, 1); }
bool IsOk = true;
for (int nid = 0; nid < G->GetNodes(); nid++) {
if (! G->IsNode(nid)) { IsOk=false; break; } }
if (! IsOk) {
printf("Nodes of the input graph have to be numbered 0...N-1\nRenumbering nodes...\n");
PNGraph OG = G; G = TNGraph::New();
TGraphEnumUtils::GetNormalizedGraph(OG, G);
}
// G = TSnap::GenRndGnm<PNGraph>(100, Kilo(1));
// count frequency of connected subgraphs in G that have MotifSz nodes
TD34GraphCounter GraphCounter(MotifSz);
TSubGraphEnum<TD34GraphCounter> GraphEnum;
GraphEnum.GetSubGraphs(G, MotifSz, GraphCounter);
FILE *F = fopen(TStr::Fmt("%s-counts.tab", OutFNm.CStr()).CStr(), "wt");
fprintf(F, "MotifId\tNodes\tEdges\tCount\n");
for (int i = 0; i < GraphCounter.Len(); i++) {
const int gid = GraphCounter.GetId(i);
PNGraph SG = GraphCounter.GetGraph(gid);
if (DrawMotifs) {
TGraphViz::Plot(SG, gvlNeato, TStr::Fmt("%s-motif%03d.gif", OutFNm.CStr(), i),
TStr::Fmt("GId:%d Count: %llu", gid, GraphCounter.GetCnt(gid)));
}
fprintf(F, "%d\t%d\t%d\t%llu\n", gid, SG->GetNodes(), SG->GetEdges(), GraphCounter.GetCnt(gid));
}
printf("done.");
fclose(F);
Catch
printf("\nrun time: %s (%s)\n", ExeTm.GetTmStr(), TSecTm::GetCurTm().GetTmStr().CStr());
return 0;
}
// Test the default constructor
TEST(TNGraph, DefaultConstructor) {
PNGraph Graph;
Graph = TNGraph::New();
EXPECT_EQ(0,Graph->GetNodes());
EXPECT_EQ(0,Graph->GetEdges());
EXPECT_EQ(1,Graph->IsOk());
EXPECT_EQ(1,Graph->Empty());
EXPECT_EQ(1,Graph->HasFlag(gfDirected));
}
// Test small graph
TEST(TNGraph, GetSmallGraph) {
PNGraph Graph;
Graph = TNGraph::GetSmallGraph();
EXPECT_EQ(5,Graph->GetNodes());
EXPECT_EQ(6,Graph->GetEdges());
EXPECT_EQ(1,Graph->IsOk());
EXPECT_EQ(0,Graph->Empty());
EXPECT_EQ(1,Graph->HasFlag(gfDirected));
}
// Test edge subgraph conversion
void TestConvertESubGraphs() {
PNEGraph NEGraph;
PNGraph NGraph;
int N1, N2, N3;
int E1, E2, E3;
TIntV NIdV;
TIntV EIdV;
int i;
NGraph = GetTestTNGraph();
N1 = NGraph->GetNodes();
E1 = NGraph->GetEdges();
for (i = 0; i < 20; i += 2) {
NIdV.Add(i);
}
// TODO: fix TSnap::ConvertSubGraph<PUNGraph>(NGraph, NIdV, true), it fails
// UNGraph = TSnap::ConvertSubGraph<PUNGraph>(NGraph, NIdV, true);
NEGraph = TSnap::ConvertGraph<PNEGraph>(NGraph);
N2 = NEGraph->GetNodes();
E2 = NEGraph->GetEdges();
// select every second edge
i = 0;
for (TNEGraph::TEdgeI EI = NEGraph->BegEI(); EI < NEGraph->EndEI(); EI++) {
if (i == 0) {
EIdV.Add(EI.GetId());
}
i = (i + 1) % 2;
}
NGraph = TSnap::ConvertESubGraph<PNGraph>(NEGraph, EIdV);
N3 = NGraph->GetNodes();
E3 = NGraph->GetEdges();
printf("---- TestConvertESubGraphs -----\n");
printf("nodes: %d,%d,%d, edges: %d,%d,%d\n", N1, N2, N3, E1, E2, E3);
printf("\n");
}
// Test graph conversion
void TestConvertGraphs() {
PNGraph NGraph;
PUNGraph UNGraph;
int N1, N2, N3;
int E1, E2, E3;
NGraph = GetTestTNGraph();
N1 = NGraph->GetNodes();
E1 = NGraph->GetEdges();
UNGraph = TSnap::ConvertGraph<PUNGraph>(NGraph);
N2 = UNGraph->GetNodes();
E2 = UNGraph->GetEdges();
NGraph = TSnap::ConvertGraph<PNGraph>(UNGraph);
N3 = NGraph->GetNodes();
E3 = NGraph->GetEdges();
printf("---- TestConvertGraphs -----\n");
printf("nodes: %d,%d,%d, edges: %d,%d,%d\n", N1, N2, N3, E1, E2, E3);
printf("\n");
}
void GetSngVals(const PNGraph& Graph, const int& SngVals, TFltV& SngValV) {
const int Nodes = Graph->GetNodes();
IAssert(SngVals > 0);
if (Nodes < 100) {
// perform full SVD
TFltVV AdjMtx(Nodes+1, Nodes+1);
TFltVV LSingV, RSingV;
TIntH NodeIdH;
// create adjecency matrix
for (TNGraph::TNodeI NodeI = Graph->BegNI(); NodeI < Graph->EndNI(); NodeI++) {
NodeIdH.AddKey(NodeI.GetId()); }
for (TNGraph::TNodeI NodeI = Graph->BegNI(); NodeI < Graph->EndNI(); NodeI++) {
const int NodeId = NodeIdH.GetKeyId(NodeI.GetId()) + 1;
for (int e = 0; e < NodeI.GetOutDeg(); e++) {
const int DstNId = NodeIdH.GetKeyId(NodeI.GetOutNId(e)) + 1; // no self edges
if (NodeId != DstNId) AdjMtx.At(NodeId, DstNId) = 1;
}
}
try { // can fail to converge but results seem to be good
TSvd::Svd1Based(AdjMtx, LSingV, SngValV, RSingV); }
catch(...) {
printf("\n***No SVD convergence: G(%d, %d)\n", Nodes, Graph->GetEdges()); }
} else {
// Lanczos
TNGraphMtx GraphMtx(Graph);
int CalcVals = int(2*SngVals);
//if (CalcVals > Nodes) { CalcVals = int(2*Nodes); }
//if (CalcVals > Nodes) { CalcVals = Nodes; }
//while (SngValV.Len() < SngVals && CalcVals < 10*SngVals) {
try {
if (SngVals > 4) {
TSparseSVD::SimpleLanczosSVD(GraphMtx, 2*SngVals, SngValV, false); }
else { TFltVV LSingV, RSingV; // this is much more precise, but also much slower
TSparseSVD::LanczosSVD(GraphMtx, SngVals, 3*SngVals, ssotFull, SngValV, LSingV, RSingV); }
}
catch(...) {
printf("\n ***EXCEPTION: TRIED %d GOT %d values** \n", 2*SngVals, SngValV.Len()); }
if (SngValV.Len() < SngVals) {
printf(" ***TRIED %d GOT %d values** \n", CalcVals, SngValV.Len()); }
// CalcVals += SngVals;
//}
}
SngValV.Sort(false);
//if (SngValV.Len() > SngVals) {
// SngValV.Del(SngVals, SngValV.Len()-1); }
//else {
// while (SngValV.Len() < SngVals) SngValV.Add(1e-6); }
//IAssert(SngValV.Len() == SngVals);
}
// Test node subgraph conversion
TEST(subgraph, TestConvertSubGraphs) {
PNGraph NGraph;
PUNGraph UNGraph;
TIntV NIdV;
int i;
NGraph = GetTestTNGraph();
EXPECT_EQ(20,NGraph->GetNodes());
EXPECT_EQ(60,NGraph->GetEdges());
for (i = 0; i < 20; i += 2) {
NIdV.Add(i);
}
// TODO: fix TSnap::ConvertSubGraph<PUNGraph>(NGraph, NIdV, true), it fails
// UNGraph = TSnap::ConvertSubGraph<PUNGraph>(NGraph, NIdV, true);
UNGraph = TSnap::ConvertSubGraph<PUNGraph>(NGraph, NIdV);
EXPECT_EQ(10,UNGraph->GetNodes());
EXPECT_EQ(10,UNGraph->GetEdges());
NGraph = TSnap::ConvertSubGraph<PNGraph>(UNGraph, NIdV);
EXPECT_EQ(10,NGraph->GetNodes());
EXPECT_EQ(20,NGraph->GetEdges());
}
void GetSngVec(const PNGraph& Graph, const int& SngVecs, TFltV& SngValV, TVec<TFltV>& LeftSV, TVec<TFltV>& RightSV) {
const int Nodes = Graph->GetNodes();
SngValV.Clr();
LeftSV.Clr();
RightSV.Clr();
TFltVV LSingV, RSingV;
if (Nodes < 100) {
// perform full SVD
TFltVV AdjMtx(Nodes+1, Nodes+1);
TIntH NodeIdH;
// create adjecency matrix (1-based)
for (TNGraph::TNodeI NodeI = Graph->BegNI(); NodeI < Graph->EndNI(); NodeI++) {
NodeIdH.AddKey(NodeI.GetId()); }
for (TNGraph::TNodeI NodeI = Graph->BegNI(); NodeI < Graph->EndNI(); NodeI++) {
const int NodeId = NodeIdH.GetKeyId(NodeI.GetId())+1;
for (int e = 0; e < NodeI.GetOutDeg(); e++) {
const int DstNId = NodeIdH.GetKeyId(NodeI.GetOutNId(e))+1; // no self edges
if (NodeId != DstNId) AdjMtx.At(NodeId, DstNId) = 1;
}
}
try { // can fail to converge but results seem to be good
TSvd::Svd1Based(AdjMtx, LSingV, SngValV, RSingV);
} catch(...) {
printf("\n***No SVD convergence: G(%d, %d)\n", Nodes, Graph->GetEdges());
}
} else { // Lanczos
TNGraphMtx GraphMtx(Graph);
TSparseSVD::LanczosSVD(GraphMtx, SngVecs, 2*SngVecs, ssotFull, SngValV, LSingV, RSingV);
//TGAlg::SaveFullMtx(Graph, "adj_mtx.txt");
//TLAMisc::DumpTFltVVMjrSubMtrx(LSingV, LSingV.GetRows(), LSingV.GetCols(), "LSingV2.txt"); // save MTX
}
TFltIntPrV SngValIdV;
for (int i = 0; i < SngValV.Len(); i++) {
SngValIdV.Add(TFltIntPr(SngValV[i], i));
}
SngValIdV.Sort(false);
SngValV.Sort(false);
for (int v = 0; v < SngValIdV.Len(); v++) {
LeftSV.Add();
LSingV.GetCol(SngValIdV[v].Val2, LeftSV.Last());
RightSV.Add();
RSingV.GetCol(SngValIdV[v].Val2, RightSV.Last());
}
IsAllValVNeg(LeftSV[0], true);
IsAllValVNeg(RightSV[0], true);
}
// Tests sequential table to graph function.
TEST(TTable, ToGraph) {
TTableContext Context;
Schema LJS;
LJS.Add(TPair<TStr,TAttrType>("Src", atInt));
LJS.Add(TPair<TStr,TAttrType>("Dst", atInt));
TIntV RelevantCols; RelevantCols.Add(0); RelevantCols.Add(1);
PTable T1 = TTable::LoadSS(LJS, "table/soc-LiveJournal1_small.txt", &Context, RelevantCols);
EXPECT_EQ(499, T1->GetNumRows().Val);
EXPECT_EQ(499, T1->GetNumValidRows().Val);
TVec<TPair<TStr, TAttrType> > S = T1->GetSchema();
PNGraph Graph = TSnap::ToGraph<PNGraph>(T1, S[0].GetVal1(), S[1].GetVal1(), aaFirst);
EXPECT_EQ(689,Graph->GetNodes());
EXPECT_EQ(499,Graph->GetEdges());
EXPECT_EQ(1,Graph->IsOk());
}
/// Rewire the network. Keeps node degrees as is but randomly rewires the edges.
/// Use this function to generate a random graph with the same degree sequence
/// as the OrigGraph.
/// See: On the uniform generation of random graphs with prescribed degree
/// sequences by R. Milo, N. Kashtan, S. Itzkovitz, M. E. J. Newman, U. Alon.
/// URL: http://arxiv.org/abs/cond-mat/0312028
PNGraph GenRewire(const PNGraph& OrigGraph, const int& NSwitch, TRnd& Rnd) {
const int Nodes = OrigGraph->GetNodes();
const int Edges = OrigGraph->GetEdges();
PNGraph GraphPt = TNGraph::New();
TNGraph& Graph = *GraphPt;
Graph.Reserve(Nodes, -1);
TExeTm ExeTm;
// generate a graph that satisfies the constraints
printf("Randomizing edges (%d, %d)...\n", Nodes, Edges);
TIntPrSet EdgeSet(Edges);
for (TNGraph::TNodeI NI = OrigGraph->BegNI(); NI < OrigGraph->EndNI(); NI++) {
const int NId = NI.GetId();
for (int e = 0; e < NI.GetOutDeg(); e++) {
EdgeSet.AddKey(TIntPr(NId, NI.GetOutNId(e))); }
Graph.AddNode(NI);
}
// edge switching
uint skip=0;
for (uint swps = 0; swps < 2*uint(Edges)*uint(NSwitch); swps++) {
const int keyId1 = EdgeSet.GetRndKeyId(Rnd);
const int keyId2 = EdgeSet.GetRndKeyId(Rnd);
if (keyId1 == keyId2) { skip++; continue; }
const TIntPr& E1 = EdgeSet[keyId1];
const TIntPr& E2 = EdgeSet[keyId2];
TIntPr NewE1(E1.Val1, E2.Val1), NewE2(E1.Val2, E2.Val2);
if (NewE1.Val1!=NewE2.Val1 && NewE1.Val2!=NewE2.Val1 && NewE1.Val2!=NewE2.Val1 && NewE1.Val2!=NewE2.Val2 && ! EdgeSet.IsKey(NewE1) && ! EdgeSet.IsKey(NewE2)) {
EdgeSet.DelKeyId(keyId1); EdgeSet.DelKeyId(keyId2);
EdgeSet.AddKey(TIntPr(NewE1));
EdgeSet.AddKey(TIntPr(NewE2));
} else { skip++; }
if (swps % Edges == 0) {
printf("\r %uk/%uk: %uk skip [%s]", swps/1000u, 2*uint(Edges)*uint(NSwitch)/1000u, skip/1000u, ExeTm.GetStr());
if (ExeTm.GetSecs() > 2*3600) { printf(" *** Time limit!\n"); break; } // time limit 2 hours
}
}
printf("\r total %uk switchings attempted, %uk skiped [%s]\n", 2*uint(Edges)*uint(NSwitch)/1000u, skip/1000u, ExeTm.GetStr());
for (int e = 0; e < EdgeSet.Len(); e++) {
Graph.AddEdge(EdgeSet[e].Val1, EdgeSet[e].Val2); }
return GraphPt;
}
void PlotSngValDistr(const PNGraph& Graph, const int& SngVals, const TStr& FNmPref, TStr DescStr) {
const int NBuckets = 50;
TFltV SngValV;
for (int f = 1; SngValV.Empty() && f < 4; f++) {
TSnap::GetSngVals(Graph, f*SngVals, SngValV);
}
SngValV.Sort(true);
THash<TFlt, TFlt> BucketCntH;
double Step = (SngValV.Last()-SngValV[0]) / double(NBuckets-1);
for (int i = 0; i < NBuckets; i++) {
BucketCntH.AddDat(SngValV[0]+Step*(i+0.5), 0);
}
for (int i = 0; i < SngValV.Len(); i++) {
const int Bucket = (int) floor((SngValV[i]-SngValV[0]) / Step);
BucketCntH[Bucket] += 1;
}
TFltPrV EigCntV;
BucketCntH.GetKeyDatPrV(EigCntV);
if (DescStr.Empty()) { DescStr = FNmPref; }
TGnuPlot::PlotValV(EigCntV, "sngDistr."+FNmPref, TStr::Fmt("%s. G(%d, %d). Largest eig val = %f", DescStr.CStr(),
Graph->GetNodes(), Graph->GetEdges(), SngValV.Last().Val), "Singular value", "Count", gpsAuto, false, gpwLinesPoints);
}
// Test subgraphs
TEST(subgraph, TestSubTNGraphs) {
PNGraph Graph;
PNGraph Graph1;
PNGraph Graph2;
PNGraph Graph3;
PNGraph Graph4;
int i;
TIntV NIdV;
TIntV NIdV1;
Graph = GetTestTNGraph();
EXPECT_EQ(20,Graph->GetNodes());
EXPECT_EQ(60,Graph->GetEdges());
for (i = 10; i < 15; i++) {
NIdV.Add(i);
}
Graph1 = TSnap::GetSubGraph(Graph, NIdV);
EXPECT_EQ(5,Graph1->GetNodes());
EXPECT_EQ(9,Graph1->GetEdges());
Graph2 = TSnap::GetSubGraph(Graph, NIdV, true);
EXPECT_EQ(5,Graph2->GetNodes());
EXPECT_EQ(9,Graph2->GetEdges());
for (i = 0; i < 20; i += 2) {
NIdV1.Add(i);
}
Graph3 = TSnap::GetSubGraph(Graph, NIdV1, true);
EXPECT_EQ(10,Graph3->GetNodes());
EXPECT_EQ(10,Graph3->GetEdges());
Graph4 = TSnap::GetSubGraphRenumber(Graph, NIdV1);
EXPECT_EQ(10,Graph4->GetNodes());
EXPECT_EQ(10,Graph4->GetEdges());
}
// Print graph statistics
void PrintGStats(const char s[], PNGraph Graph) {
printf("graph %s, nodes %d, edges %d, empty %s\n",
s, Graph->GetNodes(), Graph->GetEdges(),
Graph->Empty() ? "yes" : "no");
}