// -*- C++ -*- #ifndef HUGO_MAX_FLOW_H #define HUGO_MAX_FLOW_H #include #include #include #include #include #include #include #include /// \file /// \brief Maximum flow algorithms. /// \ingroup galgs namespace hugo { /// \addtogroup galgs /// @{ ///Maximum flow algorithms class. ///This class provides various algorithms for finding a flow of ///maximum value in a directed graph. The \e source node, the \e ///target node, the \e capacity of the edges and the \e starting \e ///flow value of the edges can be passed to the algorithm through the ///constructor. It is possible to change these quantities using the ///functions \ref resetSource, \ref resetTarget, \ref resetCap and ///\ref resetFlow. Before any subsequent runs of any algorithm of ///the class \ref resetFlow should be called, otherwise it will ///start from a maximum flow. ///After running an algorithm of the class, the maximum value of a ///value can be obtained by calling \ref flowValue(). The minimum ///value cut can be written into a \c node map of \c bools by ///calling \ref minCut. (\ref minMinCut and \ref maxMinCut writes ///the inclusionwise minimum and maximum of the minimum value ///cuts, resp.) ///\param Graph The undirected graph type the algorithm runs on. ///\param Num The number type of the capacities and the flow values. ///\param CapMap The type of the capacity map. ///\param FlowMap The type of the flow map. ///\author Marton Makai, Jacint Szabo template , typename FlowMap=typename Graph::template EdgeMap > class MaxFlow { protected: typedef typename Graph::Node Node; typedef typename Graph::NodeIt NodeIt; typedef typename Graph::EdgeIt EdgeIt; typedef typename Graph::OutEdgeIt OutEdgeIt; typedef typename Graph::InEdgeIt InEdgeIt; typedef typename std::vector > VecStack; typedef typename Graph::template NodeMap NNMap; typedef typename std::vector VecNode; const Graph* g; Node s; Node t; const CapMap* capacity; FlowMap* flow; int n; //the number of nodes of G typedef ResGraphWrapper ResGW; typedef typename ResGW::OutEdgeIt ResGWOutEdgeIt; typedef typename ResGW::Edge ResGWEdge; //typedef typename ResGW::template NodeMap ReachedMap; typedef typename Graph::template NodeMap ReachedMap; //level works as a bool map in augmenting path algorithms and is //used by bfs for storing reached information. In preflow, it //shows the levels of nodes. ReachedMap level; //excess is needed only in preflow typename Graph::template NodeMap excess; //fixme // protected: // MaxFlow() { } // void set(const Graph& _G, Node _s, Node _t, const CapMap& _capacity, // FlowMap& _flow) // { // g=&_G; // s=_s; // t=_t; // capacity=&_capacity; // flow=&_flow; // n=_G.nodeNum; // level.set (_G); //kellene vmi ilyesmi fv // excess(_G,0); //itt is // } // constants used for heuristics static const int H0=20; static const int H1=1; public: ///Indicates the property of the starting flow. ///Indicates the property of the starting flow. The meanings are as follows: ///- \c ZERO_FLOW: constant zero flow ///- \c GEN_FLOW: any flow, i.e. the sum of the in-flows equals to ///the sum of the out-flows in every node except the \e source and ///the \e target. ///- \c PRE_FLOW: any preflow, i.e. the sum of the in-flows is at ///least the sum of the out-flows in every node except the \e source. ///- \c NO_FLOW: indicates an unspecified edge map. \ref flow will be ///set to the constant zero flow in the beginning of the algorithm in this case. enum flowEnum{ ZERO_FLOW, GEN_FLOW, PRE_FLOW, NO_FLOW }; MaxFlow(const Graph& _G, Node _s, Node _t, const CapMap& _capacity, FlowMap& _flow) : g(&_G), s(_s), t(_t), capacity(&_capacity), flow(&_flow), n(_G.nodeNum()), level(_G), excess(_G,0) {} ///Runs a maximum flow algorithm. ///Runs a preflow algorithm, which is the fastest maximum flow ///algorithm up-to-date. The default for \c fe is ZERO_FLOW. ///\pre The starting flow must be /// - a constant zero flow if \c fe is \c ZERO_FLOW, /// - an arbitary flow if \c fe is \c GEN_FLOW, /// - an arbitary preflow if \c fe is \c PRE_FLOW, /// - any map if \c fe is NO_FLOW. void run(flowEnum fe=ZERO_FLOW) { preflow(fe); } ///Runs a preflow algorithm. ///Runs a preflow algorithm. The preflow algorithms provide the ///fastest way to compute a maximum flow in a directed graph. ///\pre The starting flow must be /// - a constant zero flow if \c fe is \c ZERO_FLOW, /// - an arbitary flow if \c fe is \c GEN_FLOW, /// - an arbitary preflow if \c fe is \c PRE_FLOW, /// - any map if \c fe is NO_FLOW. void preflow(flowEnum fe) { preflowPhase1(fe); preflowPhase2(); } // Heuristics: // 2 phase // gap // list 'level_list' on the nodes on level i implemented by hand // stack 'active' on the active nodes on level i // runs heuristic 'highest label' for H1*n relabels // runs heuristic 'bound decrease' for H0*n relabels, starts with 'highest label' // Parameters H0 and H1 are initialized to 20 and 1. ///Runs the first phase of the preflow algorithm. ///The preflow algorithm consists of two phases, this method runs the ///first phase. After the first phase the maximum flow value and a ///minimum value cut can already be computed, though a maximum flow ///is net yet obtained. So after calling this method \ref flowValue ///and \ref actMinCut gives proper results. ///\warning: \ref minCut, \ref minMinCut and \ref maxMinCut do not ///give minimum value cuts unless calling \ref preflowPhase2. ///\pre The starting flow must be /// - a constant zero flow if \c fe is \c ZERO_FLOW, /// - an arbitary flow if \c fe is \c GEN_FLOW, /// - an arbitary preflow if \c fe is \c PRE_FLOW, /// - any map if \c fe is NO_FLOW. void preflowPhase1( flowEnum fe ); ///Runs the second phase of the preflow algorithm. ///The preflow algorithm consists of two phases, this method runs ///the second phase. After calling \ref preflowPhase1 and then ///\ref preflowPhase2 the methods \ref flowValue, \ref minCut, ///\ref minMinCut and \ref maxMinCut give proper results. ///\pre \ref preflowPhase1 must be called before. void preflowPhase2(); /// Starting from a flow, this method searches for an augmenting path /// according to the Edmonds-Karp algorithm /// and augments the flow on if any. /// The return value shows if the augmentation was succesful. bool augmentOnShortestPath(); /// Starting from a flow, this method searches for an augmenting blocking /// flow according to Dinits' algorithm and augments the flow on if any. /// The blocking flow is computed in a physically constructed /// residual graph of type \c Mutablegraph. /// The return value show sif the augmentation was succesful. template bool augmentOnBlockingFlow(); /// The same as \c augmentOnBlockingFlow but the /// residual graph is not constructed physically. /// The return value shows if the augmentation was succesful. bool augmentOnBlockingFlow2(); /// Returns the maximum value of a flow. /// Returns the maximum value of a flow, by counting the /// over-flow of the target node \ref t. /// It can be called already after running \ref preflowPhase1. Num flowValue() { Num a=0; FOR_EACH_INC_LOC(InEdgeIt, e, *g, t) a+=(*flow)[e]; FOR_EACH_INC_LOC(OutEdgeIt, e, *g, t) a-=(*flow)[e]; return a; //marci figyu: excess[t] epp ezt adja preflow 1. fazisa utan } ///Returns a minimum value cut after calling \ref preflowPhase1. ///After the first phase of the preflow algorithm the maximum flow ///value and a minimum value cut can already be computed. This ///method can be called after running \ref preflowPhase1 for ///obtaining a minimum value cut. /// \warning Gives proper result only right after calling \ref /// preflowPhase1. /// \todo We have to make some status variable which shows the /// actual state /// of the class. This enables us to determine which methods are valid /// for MinCut computation template void actMinCut(_CutMap& M) { NodeIt v; for(g->first(v); g->valid(v); g->next(v)) { if ( level[v] < n ) { M.set(v,false); } else { M.set(v,true); } } } ///Returns the inclusionwise minimum of the minimum value cuts. ///Sets \c M to the characteristic vector of the minimum value cut ///which is inclusionwise minimum. It is computed by processing ///a bfs from the source node \c s in the residual graph. ///\pre M should be a node map of bools initialized to false. ///\pre \c flow must be a maximum flow. template void minMinCut(_CutMap& M) { std::queue queue; M.set(s,true); queue.push(s); while (!queue.empty()) { Node w=queue.front(); queue.pop(); OutEdgeIt e; for(g->first(e,w) ; g->valid(e); g->next(e)) { Node v=g->head(e); if (!M[v] && (*flow)[e] < (*capacity)[e] ) { queue.push(v); M.set(v, true); } } InEdgeIt f; for(g->first(f,w) ; g->valid(f); g->next(f)) { Node v=g->tail(f); if (!M[v] && (*flow)[f] > 0 ) { queue.push(v); M.set(v, true); } } } } ///Returns the inclusionwise maximum of the minimum value cuts. ///Sets \c M to the characteristic vector of the minimum value cut ///which is inclusionwise maximum. It is computed by processing a ///backward bfs from the target node \c t in the residual graph. ///\pre M should be a node map of bools initialized to false. ///\pre \c flow must be a maximum flow. template void maxMinCut(_CutMap& M) { NodeIt v; for(g->first(v) ; g->valid(v); g->next(v)) { M.set(v, true); } std::queue queue; M.set(t,false); queue.push(t); while (!queue.empty()) { Node w=queue.front(); queue.pop(); InEdgeIt e; for(g->first(e,w) ; g->valid(e); g->next(e)) { Node v=g->tail(e); if (M[v] && (*flow)[e] < (*capacity)[e] ) { queue.push(v); M.set(v, false); } } OutEdgeIt f; for(g->first(f,w) ; g->valid(f); g->next(f)) { Node v=g->head(f); if (M[v] && (*flow)[f] > 0 ) { queue.push(v); M.set(v, false); } } } } ///Returns a minimum value cut. ///Sets \c M to the characteristic vector of a minimum value cut. ///\pre M should be a node map of bools initialized to false. ///\pre \c flow must be a maximum flow. template void minCut(CutMap& M) { minMinCut(M); } ///Resets the source node to \c _s. ///Resets the source node to \c _s. /// void resetSource(Node _s) { s=_s; } ///Resets the target node to \c _t. ///Resets the target node to \c _t. /// void resetTarget(Node _t) { t=_t; } /// Resets the edge map of the capacities to _cap. /// Resets the edge map of the capacities to _cap. /// void resetCap(const CapMap& _cap) { capacity=&_cap; } /// Resets the edge map of the flows to _flow. /// Resets the edge map of the flows to _flow. /// void resetFlow(FlowMap& _flow) { flow=&_flow; } private: int push(Node w, VecStack& active) { int lev=level[w]; Num exc=excess[w]; int newlevel=n; //bound on the next level of w OutEdgeIt e; for(g->first(e,w); g->valid(e); g->next(e)) { if ( (*flow)[e] >= (*capacity)[e] ) continue; Node v=g->head(e); if( lev > level[v] ) { //Push is allowed now if ( excess[v]<=0 && v!=t && v!=s ) { int lev_v=level[v]; active[lev_v].push(v); } Num cap=(*capacity)[e]; Num flo=(*flow)[e]; Num remcap=cap-flo; if ( remcap >= exc ) { //A nonsaturating push. flow->set(e, flo+exc); excess.set(v, excess[v]+exc); exc=0; break; } else { //A saturating push. flow->set(e, cap); excess.set(v, excess[v]+remcap); exc-=remcap; } } else if ( newlevel > level[v] ) newlevel = level[v]; } //for out edges wv if ( exc > 0 ) { InEdgeIt e; for(g->first(e,w); g->valid(e); g->next(e)) { if( (*flow)[e] <= 0 ) continue; Node v=g->tail(e); if( lev > level[v] ) { //Push is allowed now if ( excess[v]<=0 && v!=t && v!=s ) { int lev_v=level[v]; active[lev_v].push(v); } Num flo=(*flow)[e]; if ( flo >= exc ) { //A nonsaturating push. flow->set(e, flo-exc); excess.set(v, excess[v]+exc); exc=0; break; } else { //A saturating push. excess.set(v, excess[v]+flo); exc-=flo; flow->set(e,0); } } else if ( newlevel > level[v] ) newlevel = level[v]; } //for in edges vw } // if w still has excess after the out edge for cycle excess.set(w, exc); return newlevel; } void preflowPreproc(flowEnum fe, VecStack& active, VecNode& level_list, NNMap& left, NNMap& right) { std::queue bfs_queue; switch (fe) { case NO_FLOW: //flow is already set to const zero in this case case ZERO_FLOW: { //Reverse_bfs from t, to find the starting level. level.set(t,0); bfs_queue.push(t); while (!bfs_queue.empty()) { Node v=bfs_queue.front(); bfs_queue.pop(); int l=level[v]+1; InEdgeIt e; for(g->first(e,v); g->valid(e); g->next(e)) { Node w=g->tail(e); if ( level[w] == n && w != s ) { bfs_queue.push(w); Node first=level_list[l]; if ( g->valid(first) ) left.set(first,w); right.set(w,first); level_list[l]=w; level.set(w, l); } } } //the starting flow OutEdgeIt e; for(g->first(e,s); g->valid(e); g->next(e)) { Num c=(*capacity)[e]; if ( c <= 0 ) continue; Node w=g->head(e); if ( level[w] < n ) { if ( excess[w] <= 0 && w!=t ) active[level[w]].push(w); flow->set(e, c); excess.set(w, excess[w]+c); } } break; } case GEN_FLOW: case PRE_FLOW: { //Reverse_bfs from t in the residual graph, //to find the starting level. level.set(t,0); bfs_queue.push(t); while (!bfs_queue.empty()) { Node v=bfs_queue.front(); bfs_queue.pop(); int l=level[v]+1; InEdgeIt e; for(g->first(e,v); g->valid(e); g->next(e)) { if ( (*capacity)[e] <= (*flow)[e] ) continue; Node w=g->tail(e); if ( level[w] == n && w != s ) { bfs_queue.push(w); Node first=level_list[l]; if ( g->valid(first) ) left.set(first,w); right.set(w,first); level_list[l]=w; level.set(w, l); } } OutEdgeIt f; for(g->first(f,v); g->valid(f); g->next(f)) { if ( 0 >= (*flow)[f] ) continue; Node w=g->head(f); if ( level[w] == n && w != s ) { bfs_queue.push(w); Node first=level_list[l]; if ( g->valid(first) ) left.set(first,w); right.set(w,first); level_list[l]=w; level.set(w, l); } } } //the starting flow OutEdgeIt e; for(g->first(e,s); g->valid(e); g->next(e)) { Num rem=(*capacity)[e]-(*flow)[e]; if ( rem <= 0 ) continue; Node w=g->head(e); if ( level[w] < n ) { if ( excess[w] <= 0 && w!=t ) active[level[w]].push(w); flow->set(e, (*capacity)[e]); excess.set(w, excess[w]+rem); } } InEdgeIt f; for(g->first(f,s); g->valid(f); g->next(f)) { if ( (*flow)[f] <= 0 ) continue; Node w=g->tail(f); if ( level[w] < n ) { if ( excess[w] <= 0 && w!=t ) active[level[w]].push(w); excess.set(w, excess[w]+(*flow)[f]); flow->set(f, 0); } } break; } //case PRE_FLOW } } //preflowPreproc void relabel(Node w, int newlevel, VecStack& active, VecNode& level_list, NNMap& left, NNMap& right, int& b, int& k, bool what_heur ) { Num lev=level[w]; Node right_n=right[w]; Node left_n=left[w]; //unlacing starts if ( g->valid(right_n) ) { if ( g->valid(left_n) ) { right.set(left_n, right_n); left.set(right_n, left_n); } else { level_list[lev]=right_n; left.set(right_n, INVALID); } } else { if ( g->valid(left_n) ) { right.set(left_n, INVALID); } else { level_list[lev]=INVALID; } } //unlacing ends if ( !g->valid(level_list[lev]) ) { //gapping starts for (int i=lev; i!=k ; ) { Node v=level_list[++i]; while ( g->valid(v) ) { level.set(v,n); v=right[v]; } level_list[i]=INVALID; if ( !what_heur ) { while ( !active[i].empty() ) { active[i].pop(); //FIXME: ezt szebben kene } } } level.set(w,n); b=lev-1; k=b; //gapping ends } else { if ( newlevel == n ) level.set(w,n); else { level.set(w,++newlevel); active[newlevel].push(w); if ( what_heur ) b=newlevel; if ( k < newlevel ) ++k; //now k=newlevel Node first=level_list[newlevel]; if ( g->valid(first) ) left.set(first,w); right.set(w,first); left.set(w,INVALID); level_list[newlevel]=w; } } } //relabel template class DistanceMap { protected: const MapGraphWrapper* g; typename MapGraphWrapper::template NodeMap dist; public: DistanceMap(MapGraphWrapper& _g) : g(&_g), dist(*g, g->nodeNum()) { } void set(const typename MapGraphWrapper::Node& n, int a) { dist.set(n, a); } int operator[](const typename MapGraphWrapper::Node& n) { return dist[n]; } // int get(const typename MapGraphWrapper::Node& n) const { // return dist[n]; } // bool get(const typename MapGraphWrapper::Edge& e) const { // return (dist.get(g->tail(e))head(e))); } bool operator[](const typename MapGraphWrapper::Edge& e) const { return (dist[g->tail(e)]head(e)]); } }; }; template void MaxFlow::preflowPhase1( flowEnum fe ) { int heur0=(int)(H0*n); //time while running 'bound decrease' int heur1=(int)(H1*n); //time while running 'highest label' int heur=heur1; //starting time interval (#of relabels) int numrelabel=0; bool what_heur=1; //It is 0 in case 'bound decrease' and 1 in case 'highest label' bool end=false; //Needed for 'bound decrease', true means no active nodes are above bound //b. int k=n-2; //bound on the highest level under n containing a node int b=k; //bound on the highest level under n of an active node VecStack active(n); NNMap left(*g, INVALID); NNMap right(*g, INVALID); VecNode level_list(n,INVALID); //List of the nodes in level ifirst(v); g->valid(v); g->next(v)) level.set(v,n); //setting each node to level n if ( fe == NO_FLOW ) { EdgeIt e; for(g->first(e); g->valid(e); g->next(e)) flow->set(e,0); } switch (fe) { //computing the excess case PRE_FLOW: { NodeIt v; for(g->first(v); g->valid(v); g->next(v)) { Num exc=0; InEdgeIt e; for(g->first(e,v); g->valid(e); g->next(e)) exc+=(*flow)[e]; OutEdgeIt f; for(g->first(f,v); g->valid(f); g->next(f)) exc-=(*flow)[f]; excess.set(v,exc); //putting the active nodes into the stack int lev=level[v]; if ( exc > 0 && lev < n && v != t ) active[lev].push(v); } break; } case GEN_FLOW: { NodeIt v; for(g->first(v); g->valid(v); g->next(v)) excess.set(v,0); Num exc=0; InEdgeIt e; for(g->first(e,t); g->valid(e); g->next(e)) exc+=(*flow)[e]; OutEdgeIt f; for(g->first(f,t); g->valid(f); g->next(f)) exc-=(*flow)[f]; excess.set(t,exc); break; } case ZERO_FLOW: case NO_FLOW: { NodeIt v; for(g->first(v); g->valid(v); g->next(v)) excess.set(v,0); break; } } preflowPreproc(fe, active, level_list, left, right); //End of preprocessing //Push/relabel on the highest level active nodes. while ( true ) { if ( b == 0 ) { if ( !what_heur && !end && k > 0 ) { b=k; end=true; } else break; } if ( active[b].empty() ) --b; else { end=false; Node w=active[b].top(); active[b].pop(); int newlevel=push(w,active); if ( excess[w] > 0 ) relabel(w, newlevel, active, level_list, left, right, b, k, what_heur); ++numrelabel; if ( numrelabel >= heur ) { numrelabel=0; if ( what_heur ) { what_heur=0; heur=heur0; end=false; } else { what_heur=1; heur=heur1; b=k; } } } } } template void MaxFlow::preflowPhase2() { int k=n-2; //bound on the highest level under n containing a node int b=k; //bound on the highest level under n of an active node VecStack active(n); level.set(s,0); std::queue bfs_queue; bfs_queue.push(s); while (!bfs_queue.empty()) { Node v=bfs_queue.front(); bfs_queue.pop(); int l=level[v]+1; InEdgeIt e; for(g->first(e,v); g->valid(e); g->next(e)) { if ( (*capacity)[e] <= (*flow)[e] ) continue; Node u=g->tail(e); if ( level[u] >= n ) { bfs_queue.push(u); level.set(u, l); if ( excess[u] > 0 ) active[l].push(u); } } OutEdgeIt f; for(g->first(f,v); g->valid(f); g->next(f)) { if ( 0 >= (*flow)[f] ) continue; Node u=g->head(f); if ( level[u] >= n ) { bfs_queue.push(u); level.set(u, l); if ( excess[u] > 0 ) active[l].push(u); } } } b=n-2; while ( true ) { if ( b == 0 ) break; if ( active[b].empty() ) --b; else { Node w=active[b].top(); active[b].pop(); int newlevel=push(w,active); //relabel if ( excess[w] > 0 ) { level.set(w,++newlevel); active[newlevel].push(w); b=newlevel; } } // if stack[b] is nonempty } // while(true) } template bool MaxFlow::augmentOnShortestPath() { ResGW res_graph(*g, *capacity, *flow); bool _augment=false; //ReachedMap level(res_graph); FOR_EACH_LOC(typename Graph::NodeIt, e, *g) level.set(e, 0); BfsIterator bfs(res_graph, level); bfs.pushAndSetReached(s); typename ResGW::template NodeMap pred(res_graph); pred.set(s, INVALID); typename ResGW::template NodeMap free(res_graph); //searching for augmenting path while ( !bfs.finished() ) { ResGWOutEdgeIt e=bfs; if (res_graph.valid(e) && bfs.isBNodeNewlyReached()) { Node v=res_graph.tail(e); Node w=res_graph.head(e); pred.set(w, e); if (res_graph.valid(pred[v])) { free.set(w, std::min(free[v], res_graph.resCap(e))); } else { free.set(w, res_graph.resCap(e)); } if (res_graph.head(e)==t) { _augment=true; break; } } ++bfs; } //end of searching augmenting path if (_augment) { Node n=t; Num augment_value=free[t]; while (res_graph.valid(pred[n])) { ResGWEdge e=pred[n]; res_graph.augment(e, augment_value); n=res_graph.tail(e); } } return _augment; } template template bool MaxFlow::augmentOnBlockingFlow() { typedef MutableGraph MG; bool _augment=false; ResGW res_graph(*g, *capacity, *flow); //bfs for distances on the residual graph //ReachedMap level(res_graph); FOR_EACH_LOC(typename Graph::NodeIt, e, *g) level.set(e, 0); BfsIterator bfs(res_graph, level); bfs.pushAndSetReached(s); typename ResGW::template NodeMap dist(res_graph); //filled up with 0's //F will contain the physical copy of the residual graph //with the set of edges which are on shortest paths MG F; typename ResGW::template NodeMap res_graph_to_F(res_graph); { typename ResGW::NodeIt n; for(res_graph.first(n); res_graph.valid(n); res_graph.next(n)) { res_graph_to_F.set(n, F.addNode()); } } typename MG::Node sF=res_graph_to_F[s]; typename MG::Node tF=res_graph_to_F[t]; typename MG::template EdgeMap original_edge(F); typename MG::template EdgeMap residual_capacity(F); while ( !bfs.finished() ) { ResGWOutEdgeIt e=bfs; if (res_graph.valid(e)) { if (bfs.isBNodeNewlyReached()) { dist.set(res_graph.head(e), dist[res_graph.tail(e)]+1); typename MG::Edge f=F.addEdge(res_graph_to_F[res_graph.tail(e)], res_graph_to_F[res_graph.head(e)]); original_edge.update(); original_edge.set(f, e); residual_capacity.update(); residual_capacity.set(f, res_graph.resCap(e)); } else { if (dist[res_graph.head(e)]==(dist[res_graph.tail(e)]+1)) { typename MG::Edge f=F.addEdge(res_graph_to_F[res_graph.tail(e)], res_graph_to_F[res_graph.head(e)]); original_edge.update(); original_edge.set(f, e); residual_capacity.update(); residual_capacity.set(f, res_graph.resCap(e)); } } } ++bfs; } //computing distances from s in the residual graph bool __augment=true; while (__augment) { __augment=false; //computing blocking flow with dfs DfsIterator< MG, typename MG::template NodeMap > dfs(F); typename MG::template NodeMap pred(F); pred.set(sF, INVALID); //invalid iterators for sources typename MG::template NodeMap free(F); dfs.pushAndSetReached(sF); while (!dfs.finished()) { ++dfs; if (F.valid(/*typename MG::OutEdgeIt*/(dfs))) { if (dfs.isBNodeNewlyReached()) { typename MG::Node v=F.aNode(dfs); typename MG::Node w=F.bNode(dfs); pred.set(w, dfs); if (F.valid(pred[v])) { free.set(w, std::min(free[v], residual_capacity[dfs])); } else { free.set(w, residual_capacity[dfs]); } if (w==tF) { __augment=true; _augment=true; break; } } else { F.erase(/*typename MG::OutEdgeIt*/(dfs)); } } } if (__augment) { typename MG::Node n=tF; Num augment_value=free[tF]; while (F.valid(pred[n])) { typename MG::Edge e=pred[n]; res_graph.augment(original_edge[e], augment_value); n=F.tail(e); if (residual_capacity[e]==augment_value) F.erase(e); else residual_capacity.set(e, residual_capacity[e]-augment_value); } } } return _augment; } template bool MaxFlow::augmentOnBlockingFlow2() { bool _augment=false; ResGW res_graph(*g, *capacity, *flow); //ReachedMap level(res_graph); FOR_EACH_LOC(typename Graph::NodeIt, e, *g) level.set(e, 0); BfsIterator bfs(res_graph, level); bfs.pushAndSetReached(s); DistanceMap dist(res_graph); while ( !bfs.finished() ) { ResGWOutEdgeIt e=bfs; if (res_graph.valid(e) && bfs.isBNodeNewlyReached()) { dist.set(res_graph.head(e), dist[res_graph.tail(e)]+1); } ++bfs; } //computing distances from s in the residual graph //Subgraph containing the edges on some shortest paths ConstMap true_map(true); typedef SubGraphWrapper, DistanceMap > FilterResGW; FilterResGW filter_res_graph(res_graph, true_map, dist); //Subgraph, which is able to delete edges which are already //met by the dfs typename FilterResGW::template NodeMap first_out_edges(filter_res_graph); typename FilterResGW::NodeIt v; for(filter_res_graph.first(v); filter_res_graph.valid(v); filter_res_graph.next(v)) { typename FilterResGW::OutEdgeIt e; filter_res_graph.first(e, v); first_out_edges.set(v, e); } typedef ErasingFirstGraphWrapper > ErasingResGW; ErasingResGW erasing_res_graph(filter_res_graph, first_out_edges); bool __augment=true; while (__augment) { __augment=false; //computing blocking flow with dfs DfsIterator< ErasingResGW, typename ErasingResGW::template NodeMap > dfs(erasing_res_graph); typename ErasingResGW:: template NodeMap pred(erasing_res_graph); pred.set(s, INVALID); //invalid iterators for sources typename ErasingResGW::template NodeMap free1(erasing_res_graph); dfs.pushAndSetReached ///\bug hugo 0.2 (typename ErasingResGW::Node (typename FilterResGW::Node (typename ResGW::Node(s) ) ) ); while (!dfs.finished()) { ++dfs; if (erasing_res_graph.valid(typename ErasingResGW::OutEdgeIt(dfs))) { if (dfs.isBNodeNewlyReached()) { typename ErasingResGW::Node v=erasing_res_graph.aNode(dfs); typename ErasingResGW::Node w=erasing_res_graph.bNode(dfs); pred.set(w, /*typename ErasingResGW::OutEdgeIt*/(dfs)); if (erasing_res_graph.valid(pred[v])) { free1.set (w, std::min(free1[v], res_graph.resCap (typename ErasingResGW::OutEdgeIt(dfs)))); } else { free1.set (w, res_graph.resCap (typename ErasingResGW::OutEdgeIt(dfs))); } if (w==t) { __augment=true; _augment=true; break; } } else { erasing_res_graph.erase(dfs); } } } if (__augment) { typename ErasingResGW::Node n=typename FilterResGW::Node(typename ResGW::Node(t)); // typename ResGW::NodeMap a(res_graph); // typename ResGW::Node b; // Num j=a[b]; // typename FilterResGW::NodeMap a1(filter_res_graph); // typename FilterResGW::Node b1; // Num j1=a1[b1]; // typename ErasingResGW::NodeMap a2(erasing_res_graph); // typename ErasingResGW::Node b2; // Num j2=a2[b2]; Num augment_value=free1[n]; while (erasing_res_graph.valid(pred[n])) { typename ErasingResGW::OutEdgeIt e=pred[n]; res_graph.augment(e, augment_value); n=erasing_res_graph.tail(e); if (res_graph.resCap(e)==0) erasing_res_graph.erase(e); } } } //while (__augment) return _augment; } } //namespace hugo #endif //HUGO_MAX_FLOW_H