C++ TIntSet类代码示例

/// Clique Percolation method communities
void TCliqueOverlap::GetCPMCommunities(const PUNGraph& G, int MinMaxCliqueSize, TVec<TIntV>& NIdCmtyVV) {
  printf("Clique Percolation Method\n");
  TExeTm ExeTm;
  TVec<TIntV> MaxCliques;
  TCliqueOverlap::GetMaxCliques(G, MinMaxCliqueSize, MaxCliques);
  // op RS 2012/05/15, commented out next line, a parameter is missing,
  //   creating a warning on OS X
  // printf("...%d cliques found\n");
  // get clique overlap matrix (graph)
  PUNGraph OverlapGraph = TCliqueOverlap::CalculateOverlapMtx(MaxCliques, MinMaxCliqueSize-1);
  printf("...overlap matrix (%d, %d)\n", G->GetNodes(), G->GetEdges());
  // connected components are communities
  TCnComV CnComV;
  TSnap::GetWccs(OverlapGraph, CnComV);
  NIdCmtyVV.Clr(false);
  TIntSet CmtySet;
  for (int c = 0; c < CnComV.Len(); c++) {
    CmtySet.Clr(false);
    for (int i = 0; i <CnComV[c].Len(); i++) {
      const TIntV& CliqueNIdV = MaxCliques[CnComV[c][i]];
      CmtySet.AddKeyV(CliqueNIdV);
    }
    NIdCmtyVV.Add();
    CmtySet.GetKeyV(NIdCmtyVV.Last());
    NIdCmtyVV.Last().Sort();
  }
  printf("done [%s].\n", ExeTm.GetStr());
}

double TAGMUtil::GetConductance(const PUNGraph& Graph, const TIntSet& CmtyS, const int Edges) {
    const int Edges2 = Edges >= 0 ? 2*Edges : Graph->GetEdges();
    int Vol = 0,  Cut = 0;
    double Phi = 0.0;
    for (int i = 0; i < CmtyS.Len(); i++) {
        if (! Graph->IsNode(CmtyS[i])) {
            continue;
        }
        TUNGraph::TNodeI NI = Graph->GetNI(CmtyS[i]);
        for (int e = 0; e < NI.GetOutDeg(); e++) {
            if (! CmtyS.IsKey(NI.GetOutNId(e))) {
                Cut += 1;
            }
        }
        Vol += NI.GetOutDeg();
    }
    // get conductance
    if (Vol != Edges2) {
        if (2 * Vol > Edges2) {
            Phi = Cut / double (Edges2 - Vol);
        }
        else if (Vol == 0) {
            Phi = 0.0;
        }
        else {
            Phi = Cut / double(Vol);
        }
    } else {
        if (Vol == Edges2) {
            Phi = 1.0;
        }
    }
    return Phi;
}

// Compute the empirical edge probability between a pair of nodes who share no community (epsilon), based on current community affiliations.
double TAGMFit::CalcPNoComByCmtyVV(const int& SamplePairs) {
  TIntV NIdV;
  G->GetNIdV(NIdV);
  uint64 PairNoCom = 0, EdgesNoCom = 0;
  for (int u = 0; u < NIdV.Len(); u++) {
    for (int v = u + 1; v < NIdV.Len(); v++) {
      int SrcNID = NIdV[u], DstNID = NIdV[v];
      TIntSet JointCom;
      TAGMUtil::GetIntersection(NIDComVH.GetDat(SrcNID),NIDComVH.GetDat(DstNID),JointCom);
      if(JointCom.Len() == 0) {
        PairNoCom++;
        if (G->IsEdge(SrcNID, DstNID)) { EdgesNoCom++; }
        if (SamplePairs > 0 && PairNoCom >= (uint64) SamplePairs) { break; }
      }
    }
    if (SamplePairs > 0 && PairNoCom >= (uint64) SamplePairs) { break; }
  }
  double DefaultVal = 1.0 / (double)G->GetNodes() / (double)G->GetNodes();
  if (EdgesNoCom > 0) {
    PNoCom = (double) EdgesNoCom / (double) PairNoCom;
  } else {
    PNoCom = DefaultVal;
  }
  printf("%s / %s edges without joint com detected (PNoCom = %f)\n", TUInt64::GetStr(EdgesNoCom).CStr(), TUInt64::GetStr(PairNoCom).CStr(), PNoCom.Val);
  return PNoCom;
}

void LSH::ElCheapoHashing(TQuoteBase *QuoteBase, TInt ShingleLen,
    THash<TMd5Sig, TIntSet>& ShingleToQuoteIds) {
  fprintf(stderr, "Hashing shingles the el cheapo way...\n");
  TIntV QuoteIds;
  QuoteBase->GetAllQuoteIds(QuoteIds);
  for (int qt = 0; qt < QuoteIds.Len(); qt++) {
    if (qt % 1000 == 0) {
      fprintf(stderr, "%d out of %d completed\n", qt, QuoteIds.Len());
    }
    TQuote Q;
    QuoteBase->GetQuote(QuoteIds[qt], Q);

    // Put x-character (or x-word) shingles into hash table; x is specified by ShingleLen parameter
    TStr QContentStr;
    Q.GetParsedContentString(QContentStr);
    TChA QContentChA = TChA(QContentStr);

    for (int i = 0; i < QContentChA.Len() - ShingleLen + 1; i++) {
      TChA ShingleChA = TChA();
      for (int j = 0; j < ShingleLen; j++) {
        ShingleChA.AddCh(QContentChA.GetCh(i + j));
      }
      TStr Shingle = TStr(ShingleChA);
      const TMd5Sig ShingleMd5(Shingle);
      TIntSet ShingleQuoteIds;
      if (ShingleToQuoteIds.IsKey(ShingleMd5)) {
        ShingleQuoteIds = ShingleToQuoteIds.GetDat(ShingleMd5);
      }

      ShingleQuoteIds.AddKey(QuoteIds[qt]);
      ShingleToQuoteIds.AddDat(ShingleMd5, ShingleQuoteIds);
    }
  }
  Err("Done with el cheapo hashing!\n");
}

///Generate graph using the AGM model. CProbV = vector of Pc
PUNGraph TAGM::GenAGM(TVec<TIntV>& CmtyVV, const TFltV& CProbV, TRnd& Rnd, const double PNoCom) {
    PUNGraph G = TUNGraph::New(100 * CmtyVV.Len(), -1);
    printf("AGM begins\n");
    for (int i = 0; i < CmtyVV.Len(); i++) {
        TIntV& CmtyV = CmtyVV[i];
        for (int u = 0; u < CmtyV.Len(); u++) {
            if ( G->IsNode(CmtyV[u])) {
                continue;
            }
            G->AddNode(CmtyV[u]);
        }
        double Prob = CProbV[i];
        RndConnectInsideCommunity(G, CmtyV, Prob, Rnd);
    }
    if (PNoCom > 0.0) { //if we want to connect nodes that do not share any community
        TIntSet NIDS;
        for (int c = 0; c < CmtyVV.Len(); c++) {
            for (int u = 0; u < CmtyVV[c].Len(); u++) {
                NIDS.AddKey(CmtyVV[c][u]);
            }
        }
        TIntV NIDV;
        NIDS.GetKeyV(NIDV);
        RndConnectInsideCommunity(G,NIDV,PNoCom,Rnd);
    }
    printf("AGM completed (%d nodes %d edges)\n",G->GetNodes(),G->GetEdges());
    G->Defrag();
    return G;
}

bool TBagOfWords::Update(const TStrV& TokenStrV) {    
    // Generate Ngrams if necessary
	TStrV NgramStrV;
    GenerateNgrams(TokenStrV, NgramStrV);

    // process tokens to update DF counts
    bool UpdateP = false;
    if (IsHashing()) {  
        // consolidate tokens and get their hashed IDs
        TIntSet TokenIdH;
        for (int TokenStrN = 0; TokenStrN < NgramStrV.Len(); TokenStrN++) {
            const TStr& TokenStr = NgramStrV[TokenStrN];
            TInt TokenId = TokenStr.GetHashTrick() % HashDim;
            TokenIdH.AddKey(TokenId);
            if (IsStoreHashWords()) { HashWordV[TokenId].AddKey(TokenStr); }
        }
        // update document counts
        int KeyId = TokenIdH.FFirstKeyId();
        while (TokenIdH.FNextKeyId(KeyId)) {
            const int TokenId = TokenIdH.GetKey(KeyId);
            // update DF
            DocFqV[TokenId]++;
        }
    } else {
        // consolidate tokens
        TStrH TokenStrH;
        for (int TokenStrN = 0; TokenStrN < NgramStrV.Len(); TokenStrN++) {
            const TStr& TokenStr = NgramStrV[TokenStrN];
            TokenStrH.AddKey(TokenStr);
        }
        // update document counts and update vocabulary with new tokens
        int KeyId = TokenStrH.FFirstKeyId();
        while (TokenStrH.FNextKeyId(KeyId)) {
            // get token
            const TStr& TokenStr = TokenStrH.GetKey(KeyId);
            // different processing for hashing
            int TokenId = TokenSet.GetKeyId(TokenStr);
            if (TokenId == -1) {
                // new token, remember the dimensionality change
                UpdateP = true;
                // remember the new token
                TokenId = TokenSet.AddKey(TokenStr);
                // increase document count table
                const int TokenDfId = DocFqV.Add(0);
                // increase also the old count table
                OldDocFqV.Add(0.0);
                // make sure we DF vector and TokenSet still in sync
                IAssert(TokenId == TokenDfId);
                IAssert(DocFqV.Len() == OldDocFqV.Len());
            }
            // document count update
            DocFqV[TokenId]++;
        }
    }
    // update document count
    Docs++;
    // tell if dimension changed
    return UpdateP;
}

double TAGMFit::SelectLambdaSum(const TFltV& NewLambdaV, const TIntSet& ComK) {
  double Result = 0.0;
  for (TIntSet::TIter SI = ComK.BegI(); SI < ComK.EndI(); SI++) {
    IAssert(NewLambdaV[SI.GetKey()] >= 0);
    Result += NewLambdaV[SI.GetKey()];
  }
  return Result;
}

void TAGMUtil::GetNbhCom(const PUNGraph& Graph, const int NID, TIntSet& NBCmtyS) {
    TUNGraph::TNodeI NI = Graph->GetNI(NID);
    NBCmtyS.Gen(NI.GetDeg());
    NBCmtyS.AddKey(NID);
    for (int e = 0; e < NI.GetDeg(); e++) {
        NBCmtyS.AddKey(NI.GetNbrNId(e));
    }
}

// YES I COPIED AND PASTED CODE my section leader would be so ashamed :D
void LSH::MinHash(THash<TMd5Sig, TIntSet>& ShingleToQuoteIds,
    TVec<THash<TIntV, TIntSet> >& SignatureBandBuckets) {
  TRnd RandomGenerator; // TODO: make this "more random" by incorporating time
  for (int i = 0; i < NumBands; ++i) {
    THash < TInt, TIntV > Inverted; // (QuoteID, QuoteSignatureForBand)
    THash < TIntV, TIntSet > BandBuckets; // (BandSignature, QuoteIDs)
    for (int j = 0; j < BandSize; ++j) {
      // Create new signature
      TVec < TMd5Sig > Signature;
      ShingleToQuoteIds.GetKeyV(Signature);
      Signature.Shuffle(RandomGenerator);

      // Place in bucket - not very efficient
      int SigLen = Signature.Len();
      for (int k = 0; k < SigLen; ++k) {
        TIntSet CurSet = ShingleToQuoteIds.GetDat(Signature[k]);
        for (TIntSet::TIter l = CurSet.BegI(); l < CurSet.EndI(); l++) {
          TInt Key = l.GetKey();
          if (Inverted.IsKey(Key)) {
            TIntV CurSignature = Inverted.GetDat(Key);
            if (CurSignature.Len() <= j) {
              CurSignature.Add(k);
              Inverted.AddDat(Key, CurSignature);
            }
          } else {
            TIntV NewSignature;
            NewSignature.Add(k);
            Inverted.AddDat(Key, NewSignature);
          }
        }
      }
    }

    TIntV InvertedKeys;
    Inverted.GetKeyV(InvertedKeys);
    TInt InvertedLen = InvertedKeys.Len();
    for (int k = 0; k < InvertedLen; ++k) {
      TIntSet Bucket;
      TIntV Signature = Inverted.GetDat(InvertedKeys[k]);
      if (BandBuckets.IsKey(Signature)) {
        Bucket = BandBuckets.GetDat(Signature);
      }
      Bucket.AddKey(InvertedKeys[k]);
      BandBuckets.AddDat(Signature, Bucket);
    }

    SignatureBandBuckets.Add(BandBuckets);
    Err("%d out of %d band signatures computed\n", i + 1, NumBands);
  }
  Err("Minhash step complete!\n");
}

// Helper: Return GroupSet based on NodeID
void GetGroupSet(int NId, TIntSet& GroupSet) {
  GroupSet.Clr();
  switch (NId) {
    case 0:
      GroupSet.AddKey(2);
      GroupSet.AddKey(4);
      GroupSet.AddKey(5);
      break;
    case 1:
      // Empty Set
      break;
    case 2:
      GroupSet.AddKey(0);
      GroupSet.AddKey(3);
      GroupSet.AddKey(5);
      break;
    case 3:
      GroupSet.AddKey(0);
      break;
    case 4:
      GroupSet.AddKey(0);
      break;
    case 5:
      GroupSet.AddKey(0);
      GroupSet.AddKey(1);
      break;
    default:
      ASSERT_FALSE(true); // NId Outside Graph Construction FAIL
      break;
  }
}

void TNetInfBs::SavePajek(const TStr& OutFNm) {
    TIntSet NIdSet;
    FILE *F = fopen(OutFNm.CStr(), "wt");
    fprintf(F, "*Vertices %d\r\n", NIdSet.Len());
    for (THash<TInt, TNodeInfo>::TIter NI = NodeNmH.BegI(); NI < NodeNmH.EndI(); NI++) {
      const TNodeInfo& I = NI.GetDat();
      fprintf(F, "%d \"%s\" ic Blue x_fact %f y_fact %f\r\n", NI.GetKey().Val,
        I.Name.CStr(), TMath::Mx<double>(log((double)I.Vol)-5,1), TMath::Mx<double>(log((double)I.Vol)-5,1));
    }
    fprintf(F, "*Arcs\r\n");
    for (TNGraph::TEdgeI EI = Graph->BegEI(); EI < Graph->EndEI(); EI++) {
      fprintf(F, "%d %d 1\r\n", EI.GetSrcNId(), EI.GetDstNId());
    }
    fclose(F);
}

void TGraphKey::TakeGraph(const PNGraph& Graph, TIntPrV& NodeMap) {
  TIntSet NodeIdH;
  int n = 0;
  NodeMap.Gen(Graph->GetNodes(), 0);
  for (TNGraph::TNodeI NI = Graph->BegNI(); NI < Graph->EndNI(); NI++, n++) {
    NodeIdH.AddKey(NI.GetId());
    NodeMap.Add(TIntPr(NI.GetId(), n));
  }
  Nodes = Graph->GetNodes();
  EdgeV.Gen(Nodes, 0);
  for (TNGraph::TNodeI NI = Graph->BegNI(); NI < Graph->EndNI(); NI++) {
    const int NewNId = NodeIdH.GetKeyId(NI.GetId());
    for (int i = 0; i < NI.GetOutDeg(); i++) {
      EdgeV.Add(TIntPr(NewNId, NodeIdH.GetKeyId(NI.GetOutNId(i))));
    }
  }
  EdgeV.Sort(true);
  EdgeV.Pack();
}

// For each (u, v) in edges, precompute C_uv (the set of communities u and v share).
void TAGMFit::GetEdgeJointCom() {
  ComEdgesV.Gen(CIDNSetV.Len());
  EdgeComVH.Gen(G->GetEdges());
  for (TUNGraph::TNodeI SrcNI = G->BegNI(); SrcNI < G->EndNI(); SrcNI++) {
    int SrcNID = SrcNI.GetId();
    for (int v = 0; v < SrcNI.GetDeg(); v++) {
      int DstNID = SrcNI.GetNbrNId(v);
      if (SrcNID >= DstNID) { continue; }
      TIntSet JointCom;
      IAssert(NIDComVH.IsKey(SrcNID));
      IAssert(NIDComVH.IsKey(DstNID));
      TAGMUtil::GetIntersection(NIDComVH.GetDat(SrcNID), NIDComVH.GetDat(DstNID), JointCom);
      EdgeComVH.AddDat(TIntPr(SrcNID,DstNID),JointCom);
      for (int k = 0; k < JointCom.Len(); k++) {
        ComEdgesV[JointCom[k]]++;
      }
    }
  }
  IAssert(EdgeComVH.Len() == G->GetEdges());
}

void QuoteGraph::CompareUsingShingles(THash<TMd5Sig, TIntSet>& Shingles) {
  int Count = 0;
  int RealCount = 0;
  TVec<TMd5Sig> ShingleKeys;
  Shingles.GetKeyV(ShingleKeys);
  THashSet<TIntPr> EdgeCache;

  for (int i = 0; i < ShingleKeys.Len(); i++) {
    if (i % 100 == 0) {
      Err("Processed %d out of %d shingles, count = %d\n", i, ShingleKeys.Len(), Count);
    }
    TIntSet Bucket;
    Shingles.IsKeyGetDat(ShingleKeys[i], Bucket);

    for (TIntSet::TIter Quote1 = Bucket.BegI(); Quote1 < Bucket.EndI(); Quote1++) {
      TIntSet::TIter Quote1Copy = Quote1;
      Quote1Copy++;
      for (TIntSet::TIter Quote2 = Quote1Copy; Quote2 < Bucket.EndI(); Quote2++) {
        if (!EdgeCache.IsKey(TIntPr(Quote1.GetKey(), Quote2.GetKey())) && !EdgeCache.IsKey(TIntPr(Quote2.GetKey(), Quote1.GetKey()))) {
          EdgeCache.AddKey(TIntPr(Quote1.GetKey(), Quote2.GetKey()));
          EdgeCache.AddKey(TIntPr(Quote2.GetKey(), Quote1.GetKey()));
          RealCount++;
          AddEdgeIfSimilar(Quote1.GetKey(), Quote2.GetKey());
        }
      }
    }
    int Len = Bucket.Len() * (Bucket.Len() - 1) / 2;
    Count += Len;
  }
  fprintf(stderr, "NUMBER OF COMPARES: %d\n", Count);
  fprintf(stderr, "NUMBER OF REAL COMPARES: %d\n", RealCount);
}

void TGStatVec::SaveTxt(const TStr& FNmPref, const TStr& Desc) const {
  FILE *F = fopen(TStr::Fmt("growth.%s.tab", FNmPref.CStr()).CStr(), "wt");
  fprintf(F, "# %s\n", Desc.CStr());
  fprintf(F, "# %s", TTmInfo::GetTmUnitStr(TmUnit).CStr());
  TIntSet StatValSet;
  for (int i = 0; i < Len(); i++) {
    for (int v = gsvNone; v < gsvMx; v++) {
      if (At(i)->HasVal(TGStatVal(v))) { StatValSet.AddKey(v); }
    }
  }
  TIntV StatValV;  StatValSet.GetKeyV(StatValV);  StatValV.Sort();
  for (int sv = 0; sv < StatValV.Len(); sv++) {
    fprintf(F, "\t%s", TGStat::GetValStr(TGStatVal(StatValV[sv].Val)).CStr()); }
  fprintf(F, "Time\n");
  for (int i = 0; i < Len(); i++) {
    const TGStat& G = *At(i);
    for (int sv = 0; sv < StatValV.Len(); sv++) {
      fprintf(F, "%g\t", G.GetVal(TGStatVal(StatValV[sv].Val))); }
    fprintf(F, "%s\n", G.GetTmStr().CStr());
  }
  fclose(F);
}