gravatar
kpeter (Peter Kovacs)
kpeter@inf.elte.hu
Entirely rework CostScaling (#180) - Use the new interface similarly to NetworkSimplex. - Rework the implementation using an efficient internal structure for handling the residual network. This improvement made the code much faster. - Handle GEQ supply type (LEQ is not supported). - Handle infinite upper bounds. - Handle negative costs (for arcs of finite upper bound). - Traits class + named parameter for the LargeCost type used in internal computations. - Extend the documentation.
0 1 0
default
1 file changed with 860 insertions and 576 deletions:
↑ Collapse diff ↑
Ignore white space 6 line context
... ...
@@ -30,548 +30,912 @@
30 30
#include <lemon/core.h>
31 31
#include <lemon/maps.h>
32 32
#include <lemon/math.h>
33
#include <lemon/adaptors.h>
33
#include <lemon/static_graph.h>
34 34
#include <lemon/circulation.h>
35 35
#include <lemon/bellman_ford.h>
36 36

	
37 37
namespace lemon {
38 38

	
39
  /// \brief Default traits class of CostScaling algorithm.
40
  ///
41
  /// Default traits class of CostScaling algorithm.
42
  /// \tparam GR Digraph type.
43
  /// \tparam V The value type used for flow amounts, capacity bounds
44
  /// and supply values. By default it is \c int.
45
  /// \tparam C The value type used for costs and potentials.
46
  /// By default it is the same as \c V.
47
#ifdef DOXYGEN
48
  template <typename GR, typename V = int, typename C = V>
49
#else
50
  template < typename GR, typename V = int, typename C = V,
51
             bool integer = std::numeric_limits<C>::is_integer >
52
#endif
53
  struct CostScalingDefaultTraits
54
  {
55
    /// The type of the digraph
56
    typedef GR Digraph;
57
    /// The type of the flow amounts, capacity bounds and supply values
58
    typedef V Value;
59
    /// The type of the arc costs
60
    typedef C Cost;
61

	
62
    /// \brief The large cost type used for internal computations
63
    ///
64
    /// The large cost type used for internal computations.
65
    /// It is \c long \c long if the \c Cost type is integer,
66
    /// otherwise it is \c double.
67
    /// \c Cost must be convertible to \c LargeCost.
68
    typedef double LargeCost;
69
  };
70

	
71
  // Default traits class for integer cost types
72
  template <typename GR, typename V, typename C>
73
  struct CostScalingDefaultTraits<GR, V, C, true>
74
  {
75
    typedef GR Digraph;
76
    typedef V Value;
77
    typedef C Cost;
78
#ifdef LEMON_HAVE_LONG_LONG
79
    typedef long long LargeCost;
80
#else
81
    typedef long LargeCost;
82
#endif
83
  };
84

	
85

	
39 86
  /// \addtogroup min_cost_flow_algs
40 87
  /// @{
41 88

	
42
  /// \brief Implementation of the cost scaling algorithm for finding a
43
  /// minimum cost flow.
89
  /// \brief Implementation of the Cost Scaling algorithm for
90
  /// finding a \ref min_cost_flow "minimum cost flow".
44 91
  ///
45
  /// \ref CostScaling implements the cost scaling algorithm performing
46
  /// augment/push and relabel operations for finding a minimum cost
47
  /// flow.
92
  /// \ref CostScaling implements a cost scaling algorithm that performs
93
  /// push/augment and relabel operations for finding a minimum cost
94
  /// flow. It is an efficient primal-dual solution method, which
95
  /// can be viewed as the generalization of the \ref Preflow
96
  /// "preflow push-relabel" algorithm for the maximum flow problem.
48 97
  ///
49
  /// \tparam Digraph The digraph type the algorithm runs on.
50
  /// \tparam LowerMap The type of the lower bound map.
51
  /// \tparam CapacityMap The type of the capacity (upper bound) map.
52
  /// \tparam CostMap The type of the cost (length) map.
53
  /// \tparam SupplyMap The type of the supply map.
98
  /// Most of the parameters of the problem (except for the digraph)
99
  /// can be given using separate functions, and the algorithm can be
100
  /// executed using the \ref run() function. If some parameters are not
101
  /// specified, then default values will be used.
54 102
  ///
55
  /// \warning
56
  /// - Arc capacities and costs should be \e non-negative \e integers.
57
  /// - Supply values should be \e signed \e integers.
58
  /// - The value types of the maps should be convertible to each other.
59
  /// - \c CostMap::Value must be signed type.
103
  /// \tparam GR The digraph type the algorithm runs on.
104
  /// \tparam V The value type used for flow amounts, capacity bounds
105
  /// and supply values in the algorithm. By default it is \c int.
106
  /// \tparam C The value type used for costs and potentials in the
107
  /// algorithm. By default it is the same as \c V.
60 108
  ///
61
  /// \note Arc costs are multiplied with the number of nodes during
62
  /// the algorithm so overflow problems may arise more easily than with
63
  /// other minimum cost flow algorithms.
64
  /// If it is available, <tt>long long int</tt> type is used instead of
65
  /// <tt>long int</tt> in the inside computations.
66
  ///
67
  /// \author Peter Kovacs
68
  template < typename Digraph,
69
             typename LowerMap = typename Digraph::template ArcMap<int>,
70
             typename CapacityMap = typename Digraph::template ArcMap<int>,
71
             typename CostMap = typename Digraph::template ArcMap<int>,
72
             typename SupplyMap = typename Digraph::template NodeMap<int> >
109
  /// \warning Both value types must be signed and all input data must
110
  /// be integer.
111
  /// \warning This algorithm does not support negative costs for such
112
  /// arcs that have infinite upper bound.
113
#ifdef DOXYGEN
114
  template <typename GR, typename V, typename C, typename TR>
115
#else
116
  template < typename GR, typename V = int, typename C = V,
117
             typename TR = CostScalingDefaultTraits<GR, V, C> >
118
#endif
73 119
  class CostScaling
74 120
  {
75
    TEMPLATE_DIGRAPH_TYPEDEFS(Digraph);
121
  public:
76 122

	
77
    typedef typename CapacityMap::Value Capacity;
78
    typedef typename CostMap::Value Cost;
79
    typedef typename SupplyMap::Value Supply;
80
    typedef typename Digraph::template ArcMap<Capacity> CapacityArcMap;
81
    typedef typename Digraph::template NodeMap<Supply> SupplyNodeMap;
123
    /// The type of the digraph
124
    typedef typename TR::Digraph Digraph;
125
    /// The type of the flow amounts, capacity bounds and supply values
126
    typedef typename TR::Value Value;
127
    /// The type of the arc costs
128
    typedef typename TR::Cost Cost;
82 129

	
83
    typedef ResidualDigraph< const Digraph,
84
                             CapacityArcMap, CapacityArcMap > ResDigraph;
85
    typedef typename ResDigraph::Arc ResArc;
130
    /// \brief The large cost type
131
    ///
132
    /// The large cost type used for internal computations.
133
    /// Using the \ref CostScalingDefaultTraits "default traits class",
134
    /// it is \c long \c long if the \c Cost type is integer,
135
    /// otherwise it is \c double.
136
    typedef typename TR::LargeCost LargeCost;
86 137

	
87
#if defined __GNUC__ && !defined __STRICT_ANSI__
88
    typedef long long int LCost;
89
#else
90
    typedef long int LCost;
91
#endif
92
    typedef typename Digraph::template ArcMap<LCost> LargeCostMap;
138
    /// The \ref CostScalingDefaultTraits "traits class" of the algorithm
139
    typedef TR Traits;
93 140

	
94 141
  public:
95 142

	
96
    /// The type of the flow map.
97
    typedef typename Digraph::template ArcMap<Capacity> FlowMap;
98
    /// The type of the potential map.
99
    typedef typename Digraph::template NodeMap<LCost> PotentialMap;
143
    /// \brief Problem type constants for the \c run() function.
144
    ///
145
    /// Enum type containing the problem type constants that can be
146
    /// returned by the \ref run() function of the algorithm.
147
    enum ProblemType {
148
      /// The problem has no feasible solution (flow).
149
      INFEASIBLE,
150
      /// The problem has optimal solution (i.e. it is feasible and
151
      /// bounded), and the algorithm has found optimal flow and node
152
      /// potentials (primal and dual solutions).
153
      OPTIMAL,
154
      /// The digraph contains an arc of negative cost and infinite
155
      /// upper bound. It means that the objective function is unbounded
156
      /// on that arc, however note that it could actually be bounded
157
      /// over the feasible flows, but this algroithm cannot handle
158
      /// these cases.
159
      UNBOUNDED
160
    };
100 161

	
101 162
  private:
102 163

	
103
    /// \brief Map adaptor class for handling residual arc costs.
104
    ///
105
    /// Map adaptor class for handling residual arc costs.
106
    template <typename Map>
107
    class ResidualCostMap : public MapBase<ResArc, typename Map::Value>
108
    {
109
    private:
164
    TEMPLATE_DIGRAPH_TYPEDEFS(GR);
110 165

	
111
      const Map &_cost_map;
166
    typedef std::vector<int> IntVector;
167
    typedef std::vector<char> BoolVector;
168
    typedef std::vector<Value> ValueVector;
169
    typedef std::vector<Cost> CostVector;
170
    typedef std::vector<LargeCost> LargeCostVector;
112 171

	
172
  private:
173
  
174
    template <typename KT, typename VT>
175
    class VectorMap {
113 176
    public:
114

	
115
      ///\e
116
      ResidualCostMap(const Map &cost_map) :
117
        _cost_map(cost_map) {}
118

	
119
      ///\e
120
      inline typename Map::Value operator[](const ResArc &e) const {
121
        return ResDigraph::forward(e) ? _cost_map[e] : -_cost_map[e];
177
      typedef KT Key;
178
      typedef VT Value;
179
      
180
      VectorMap(std::vector<Value>& v) : _v(v) {}
181
      
182
      const Value& operator[](const Key& key) const {
183
        return _v[StaticDigraph::id(key)];
122 184
      }
123 185

	
124
    }; //class ResidualCostMap
125

	
126
    /// \brief Map adaptor class for handling reduced arc costs.
127
    ///
128
    /// Map adaptor class for handling reduced arc costs.
129
    class ReducedCostMap : public MapBase<Arc, LCost>
130
    {
131
    private:
132

	
133
      const Digraph &_gr;
134
      const LargeCostMap &_cost_map;
135
      const PotentialMap &_pot_map;
136

	
137
    public:
138

	
139
      ///\e
140
      ReducedCostMap( const Digraph &gr,
141
                      const LargeCostMap &cost_map,
142
                      const PotentialMap &pot_map ) :
143
        _gr(gr), _cost_map(cost_map), _pot_map(pot_map) {}
144

	
145
      ///\e
146
      inline LCost operator[](const Arc &e) const {
147
        return _cost_map[e] + _pot_map[_gr.source(e)]
148
                            - _pot_map[_gr.target(e)];
186
      Value& operator[](const Key& key) {
187
        return _v[StaticDigraph::id(key)];
188
      }
189
      
190
      void set(const Key& key, const Value& val) {
191
        _v[StaticDigraph::id(key)] = val;
149 192
      }
150 193

	
151
    }; //class ReducedCostMap
194
    private:
195
      std::vector<Value>& _v;
196
    };
197

	
198
    typedef VectorMap<StaticDigraph::Node, LargeCost> LargeCostNodeMap;
199
    typedef VectorMap<StaticDigraph::Arc, LargeCost> LargeCostArcMap;
152 200

	
153 201
  private:
154 202

	
155
    // The digraph the algorithm runs on
156
    const Digraph &_graph;
157
    // The original lower bound map
158
    const LowerMap *_lower;
159
    // The modified capacity map
160
    CapacityArcMap _capacity;
161
    // The original cost map
162
    const CostMap &_orig_cost;
163
    // The scaled cost map
164
    LargeCostMap _cost;
165
    // The modified supply map
166
    SupplyNodeMap _supply;
167
    bool _valid_supply;
203
    // Data related to the underlying digraph
204
    const GR &_graph;
205
    int _node_num;
206
    int _arc_num;
207
    int _res_node_num;
208
    int _res_arc_num;
209
    int _root;
168 210

	
169
    // Arc map of the current flow
170
    FlowMap *_flow;
171
    bool _local_flow;
172
    // Node map of the current potentials
173
    PotentialMap *_potential;
174
    bool _local_potential;
211
    // Parameters of the problem
212
    bool _have_lower;
213
    Value _sum_supply;
175 214

	
176
    // The residual cost map
177
    ResidualCostMap<LargeCostMap> _res_cost;
178
    // The residual digraph
179
    ResDigraph *_res_graph;
180
    // The reduced cost map
181
    ReducedCostMap *_red_cost;
182
    // The excess map
183
    SupplyNodeMap _excess;
184
    // The epsilon parameter used for cost scaling
185
    LCost _epsilon;
186
    // The scaling factor
215
    // Data structures for storing the digraph
216
    IntNodeMap _node_id;
217
    IntArcMap _arc_idf;
218
    IntArcMap _arc_idb;
219
    IntVector _first_out;
220
    BoolVector _forward;
221
    IntVector _source;
222
    IntVector _target;
223
    IntVector _reverse;
224

	
225
    // Node and arc data
226
    ValueVector _lower;
227
    ValueVector _upper;
228
    CostVector _scost;
229
    ValueVector _supply;
230

	
231
    ValueVector _res_cap;
232
    LargeCostVector _cost;
233
    LargeCostVector _pi;
234
    ValueVector _excess;
235
    IntVector _next_out;
236
    std::deque<int> _active_nodes;
237

	
238
    // Data for scaling
239
    LargeCost _epsilon;
187 240
    int _alpha;
188 241

	
242
    // Data for a StaticDigraph structure
243
    typedef std::pair<int, int> IntPair;
244
    StaticDigraph _sgr;
245
    std::vector<IntPair> _arc_vec;
246
    std::vector<LargeCost> _cost_vec;
247
    LargeCostArcMap _cost_map;
248
    LargeCostNodeMap _pi_map;
249
  
250
  public:
251
  
252
    /// \brief Constant for infinite upper bounds (capacities).
253
    ///
254
    /// Constant for infinite upper bounds (capacities).
255
    /// It is \c std::numeric_limits<Value>::infinity() if available,
256
    /// \c std::numeric_limits<Value>::max() otherwise.
257
    const Value INF;
258

	
189 259
  public:
190 260

	
191
    /// \brief General constructor (with lower bounds).
261
    /// \name Named Template Parameters
262
    /// @{
263

	
264
    template <typename T>
265
    struct SetLargeCostTraits : public Traits {
266
      typedef T LargeCost;
267
    };
268

	
269
    /// \brief \ref named-templ-param "Named parameter" for setting
270
    /// \c LargeCost type.
192 271
    ///
193
    /// General constructor (with lower bounds).
272
    /// \ref named-templ-param "Named parameter" for setting \c LargeCost
273
    /// type, which is used for internal computations in the algorithm.
274
    /// \c Cost must be convertible to \c LargeCost.
275
    template <typename T>
276
    struct SetLargeCost
277
      : public CostScaling<GR, V, C, SetLargeCostTraits<T> > {
278
      typedef  CostScaling<GR, V, C, SetLargeCostTraits<T> > Create;
279
    };
280

	
281
    /// @}
282

	
283
  public:
284

	
285
    /// \brief Constructor.
194 286
    ///
195
    /// \param digraph The digraph the algorithm runs on.
196
    /// \param lower The lower bounds of the arcs.
197
    /// \param capacity The capacities (upper bounds) of the arcs.
198
    /// \param cost The cost (length) values of the arcs.
199
    /// \param supply The supply values of the nodes (signed).
200
    CostScaling( const Digraph &digraph,
201
                 const LowerMap &lower,
202
                 const CapacityMap &capacity,
203
                 const CostMap &cost,
204
                 const SupplyMap &supply ) :
205
      _graph(digraph), _lower(&lower), _capacity(digraph), _orig_cost(cost),
206
      _cost(digraph), _supply(digraph), _flow(NULL), _local_flow(false),
207
      _potential(NULL), _local_potential(false), _res_cost(_cost),
208
      _res_graph(NULL), _red_cost(NULL), _excess(digraph, 0)
287
    /// The constructor of the class.
288
    ///
289
    /// \param graph The digraph the algorithm runs on.
290
    CostScaling(const GR& graph) :
291
      _graph(graph), _node_id(graph), _arc_idf(graph), _arc_idb(graph),
292
      _cost_map(_cost_vec), _pi_map(_pi),
293
      INF(std::numeric_limits<Value>::has_infinity ?
294
          std::numeric_limits<Value>::infinity() :
295
          std::numeric_limits<Value>::max())
209 296
    {
210
      // Check the sum of supply values
211
      Supply sum = 0;
212
      for (NodeIt n(_graph); n != INVALID; ++n) sum += _supply[n];
213
      _valid_supply = sum == 0;
297
      // Check the value types
298
      LEMON_ASSERT(std::numeric_limits<Value>::is_signed,
299
        "The flow type of CostScaling must be signed");
300
      LEMON_ASSERT(std::numeric_limits<Cost>::is_signed,
301
        "The cost type of CostScaling must be signed");
302

	
303
      // Resize vectors
304
      _node_num = countNodes(_graph);
305
      _arc_num = countArcs(_graph);
306
      _res_node_num = _node_num + 1;
307
      _res_arc_num = 2 * (_arc_num + _node_num);
308
      _root = _node_num;
309

	
310
      _first_out.resize(_res_node_num + 1);
311
      _forward.resize(_res_arc_num);
312
      _source.resize(_res_arc_num);
313
      _target.resize(_res_arc_num);
314
      _reverse.resize(_res_arc_num);
315

	
316
      _lower.resize(_res_arc_num);
317
      _upper.resize(_res_arc_num);
318
      _scost.resize(_res_arc_num);
319
      _supply.resize(_res_node_num);
214 320
      
215
      for (ArcIt e(_graph); e != INVALID; ++e) _capacity[e] = capacity[e];
216
      for (NodeIt n(_graph); n != INVALID; ++n) _supply[n] = supply[n];
321
      _res_cap.resize(_res_arc_num);
322
      _cost.resize(_res_arc_num);
323
      _pi.resize(_res_node_num);
324
      _excess.resize(_res_node_num);
325
      _next_out.resize(_res_node_num);
217 326

	
218
      // Remove non-zero lower bounds
219
      for (ArcIt e(_graph); e != INVALID; ++e) {
220
        if (lower[e] != 0) {
221
          _capacity[e] -= lower[e];
222
          _supply[_graph.source(e)] -= lower[e];
223
          _supply[_graph.target(e)] += lower[e];
327
      _arc_vec.reserve(_res_arc_num);
328
      _cost_vec.reserve(_res_arc_num);
329

	
330
      // Copy the graph
331
      int i = 0, j = 0, k = 2 * _arc_num + _node_num;
332
      for (NodeIt n(_graph); n != INVALID; ++n, ++i) {
333
        _node_id[n] = i;
334
      }
335
      i = 0;
336
      for (NodeIt n(_graph); n != INVALID; ++n, ++i) {
337
        _first_out[i] = j;
338
        for (OutArcIt a(_graph, n); a != INVALID; ++a, ++j) {
339
          _arc_idf[a] = j;
340
          _forward[j] = true;
341
          _source[j] = i;
342
          _target[j] = _node_id[_graph.runningNode(a)];
224 343
        }
344
        for (InArcIt a(_graph, n); a != INVALID; ++a, ++j) {
345
          _arc_idb[a] = j;
346
          _forward[j] = false;
347
          _source[j] = i;
348
          _target[j] = _node_id[_graph.runningNode(a)];
349
        }
350
        _forward[j] = false;
351
        _source[j] = i;
352
        _target[j] = _root;
353
        _reverse[j] = k;
354
        _forward[k] = true;
355
        _source[k] = _root;
356
        _target[k] = i;
357
        _reverse[k] = j;
358
        ++j; ++k;
225 359
      }
226
    }
227
/*
228
    /// \brief General constructor (without lower bounds).
229
    ///
230
    /// General constructor (without lower bounds).
231
    ///
232
    /// \param digraph The digraph the algorithm runs on.
233
    /// \param capacity The capacities (upper bounds) of the arcs.
234
    /// \param cost The cost (length) values of the arcs.
235
    /// \param supply The supply values of the nodes (signed).
236
    CostScaling( const Digraph &digraph,
237
                 const CapacityMap &capacity,
238
                 const CostMap &cost,
239
                 const SupplyMap &supply ) :
240
      _graph(digraph), _lower(NULL), _capacity(capacity), _orig_cost(cost),
241
      _cost(digraph), _supply(supply), _flow(NULL), _local_flow(false),
242
      _potential(NULL), _local_potential(false), _res_cost(_cost),
243
      _res_graph(NULL), _red_cost(NULL), _excess(digraph, 0)
244
    {
245
      // Check the sum of supply values
246
      Supply sum = 0;
247
      for (NodeIt n(_graph); n != INVALID; ++n) sum += _supply[n];
248
      _valid_supply = sum == 0;
360
      _first_out[i] = j;
361
      _first_out[_res_node_num] = k;
362
      for (ArcIt a(_graph); a != INVALID; ++a) {
363
        int fi = _arc_idf[a];
364
        int bi = _arc_idb[a];
365
        _reverse[fi] = bi;
366
        _reverse[bi] = fi;
367
      }
368
      
369
      // Reset parameters
370
      reset();
249 371
    }
250 372

	
251
    /// \brief Simple constructor (with lower bounds).
373
    /// \name Parameters
374
    /// The parameters of the algorithm can be specified using these
375
    /// functions.
376

	
377
    /// @{
378

	
379
    /// \brief Set the lower bounds on the arcs.
252 380
    ///
253
    /// Simple constructor (with lower bounds).
381
    /// This function sets the lower bounds on the arcs.
382
    /// If it is not used before calling \ref run(), the lower bounds
383
    /// will be set to zero on all arcs.
254 384
    ///
255
    /// \param digraph The digraph the algorithm runs on.
256
    /// \param lower The lower bounds of the arcs.
257
    /// \param capacity The capacities (upper bounds) of the arcs.
258
    /// \param cost The cost (length) values of the arcs.
259
    /// \param s The source node.
260
    /// \param t The target node.
261
    /// \param flow_value The required amount of flow from node \c s
262
    /// to node \c t (i.e. the supply of \c s and the demand of \c t).
263
    CostScaling( const Digraph &digraph,
264
                 const LowerMap &lower,
265
                 const CapacityMap &capacity,
266
                 const CostMap &cost,
267
                 Node s, Node t,
268
                 Supply flow_value ) :
269
      _graph(digraph), _lower(&lower), _capacity(capacity), _orig_cost(cost),
270
      _cost(digraph), _supply(digraph, 0), _flow(NULL), _local_flow(false),
271
      _potential(NULL), _local_potential(false), _res_cost(_cost),
272
      _res_graph(NULL), _red_cost(NULL), _excess(digraph, 0)
273
    {
274
      // Remove non-zero lower bounds
275
      _supply[s] =  flow_value;
276
      _supply[t] = -flow_value;
277
      for (ArcIt e(_graph); e != INVALID; ++e) {
278
        if (lower[e] != 0) {
279
          _capacity[e] -= lower[e];
280
          _supply[_graph.source(e)] -= lower[e];
281
          _supply[_graph.target(e)] += lower[e];
282
        }
385
    /// \param map An arc map storing the lower bounds.
386
    /// Its \c Value type must be convertible to the \c Value type
387
    /// of the algorithm.
388
    ///
389
    /// \return <tt>(*this)</tt>
390
    template <typename LowerMap>
391
    CostScaling& lowerMap(const LowerMap& map) {
392
      _have_lower = true;
393
      for (ArcIt a(_graph); a != INVALID; ++a) {
394
        _lower[_arc_idf[a]] = map[a];
395
        _lower[_arc_idb[a]] = map[a];
283 396
      }
284
      _valid_supply = true;
285
    }
286

	
287
    /// \brief Simple constructor (without lower bounds).
288
    ///
289
    /// Simple constructor (without lower bounds).
290
    ///
291
    /// \param digraph The digraph the algorithm runs on.
292
    /// \param capacity The capacities (upper bounds) of the arcs.
293
    /// \param cost The cost (length) values of the arcs.
294
    /// \param s The source node.
295
    /// \param t The target node.
296
    /// \param flow_value The required amount of flow from node \c s
297
    /// to node \c t (i.e. the supply of \c s and the demand of \c t).
298
    CostScaling( const Digraph &digraph,
299
                 const CapacityMap &capacity,
300
                 const CostMap &cost,
301
                 Node s, Node t,
302
                 Supply flow_value ) :
303
      _graph(digraph), _lower(NULL), _capacity(capacity), _orig_cost(cost),
304
      _cost(digraph), _supply(digraph, 0), _flow(NULL), _local_flow(false),
305
      _potential(NULL), _local_potential(false), _res_cost(_cost),
306
      _res_graph(NULL), _red_cost(NULL), _excess(digraph, 0)
307
    {
308
      _supply[s] =  flow_value;
309
      _supply[t] = -flow_value;
310
      _valid_supply = true;
311
    }
312
*/
313
    /// Destructor.
314
    ~CostScaling() {
315
      if (_local_flow) delete _flow;
316
      if (_local_potential) delete _potential;
317
      delete _res_graph;
318
      delete _red_cost;
319
    }
320

	
321
    /// \brief Set the flow map.
322
    ///
323
    /// Set the flow map.
324
    ///
325
    /// \return \c (*this)
326
    CostScaling& flowMap(FlowMap &map) {
327
      if (_local_flow) {
328
        delete _flow;
329
        _local_flow = false;
330
      }
331
      _flow = &map;
332 397
      return *this;
333 398
    }
334 399

	
335
    /// \brief Set the potential map.
400
    /// \brief Set the upper bounds (capacities) on the arcs.
336 401
    ///
337
    /// Set the potential map.
402
    /// This function sets the upper bounds (capacities) on the arcs.
403
    /// If it is not used before calling \ref run(), the upper bounds
404
    /// will be set to \ref INF on all arcs (i.e. the flow value will be
405
    /// unbounded from above on each arc).
338 406
    ///
339
    /// \return \c (*this)
340
    CostScaling& potentialMap(PotentialMap &map) {
341
      if (_local_potential) {
342
        delete _potential;
343
        _local_potential = false;
407
    /// \param map An arc map storing the upper bounds.
408
    /// Its \c Value type must be convertible to the \c Value type
409
    /// of the algorithm.
410
    ///
411
    /// \return <tt>(*this)</tt>
412
    template<typename UpperMap>
413
    CostScaling& upperMap(const UpperMap& map) {
414
      for (ArcIt a(_graph); a != INVALID; ++a) {
415
        _upper[_arc_idf[a]] = map[a];
344 416
      }
345
      _potential = &map;
346 417
      return *this;
347 418
    }
348 419

	
420
    /// \brief Set the costs of the arcs.
421
    ///
422
    /// This function sets the costs of the arcs.
423
    /// If it is not used before calling \ref run(), the costs
424
    /// will be set to \c 1 on all arcs.
425
    ///
426
    /// \param map An arc map storing the costs.
427
    /// Its \c Value type must be convertible to the \c Cost type
428
    /// of the algorithm.
429
    ///
430
    /// \return <tt>(*this)</tt>
431
    template<typename CostMap>
432
    CostScaling& costMap(const CostMap& map) {
433
      for (ArcIt a(_graph); a != INVALID; ++a) {
434
        _scost[_arc_idf[a]] =  map[a];
435
        _scost[_arc_idb[a]] = -map[a];
436
      }
437
      return *this;
438
    }
439

	
440
    /// \brief Set the supply values of the nodes.
441
    ///
442
    /// This function sets the supply values of the nodes.
443
    /// If neither this function nor \ref stSupply() is used before
444
    /// calling \ref run(), the supply of each node will be set to zero.
445
    ///
446
    /// \param map A node map storing the supply values.
447
    /// Its \c Value type must be convertible to the \c Value type
448
    /// of the algorithm.
449
    ///
450
    /// \return <tt>(*this)</tt>
451
    template<typename SupplyMap>
452
    CostScaling& supplyMap(const SupplyMap& map) {
453
      for (NodeIt n(_graph); n != INVALID; ++n) {
454
        _supply[_node_id[n]] = map[n];
455
      }
456
      return *this;
457
    }
458

	
459
    /// \brief Set single source and target nodes and a supply value.
460
    ///
461
    /// This function sets a single source node and a single target node
462
    /// and the required flow value.
463
    /// If neither this function nor \ref supplyMap() is used before
464
    /// calling \ref run(), the supply of each node will be set to zero.
465
    ///
466
    /// Using this function has the same effect as using \ref supplyMap()
467
    /// with such a map in which \c k is assigned to \c s, \c -k is
468
    /// assigned to \c t and all other nodes have zero supply value.
469
    ///
470
    /// \param s The source node.
471
    /// \param t The target node.
472
    /// \param k The required amount of flow from node \c s to node \c t
473
    /// (i.e. the supply of \c s and the demand of \c t).
474
    ///
475
    /// \return <tt>(*this)</tt>
476
    CostScaling& stSupply(const Node& s, const Node& t, Value k) {
477
      for (int i = 0; i != _res_node_num; ++i) {
478
        _supply[i] = 0;
479
      }
480
      _supply[_node_id[s]] =  k;
481
      _supply[_node_id[t]] = -k;
482
      return *this;
483
    }
484
    
485
    /// @}
486

	
349 487
    /// \name Execution control
488
    /// The algorithm can be executed using \ref run().
350 489

	
351 490
    /// @{
352 491

	
353 492
    /// \brief Run the algorithm.
354 493
    ///
355
    /// Run the algorithm.
494
    /// This function runs the algorithm.
495
    /// The paramters can be specified using functions \ref lowerMap(),
496
    /// \ref upperMap(), \ref costMap(), \ref supplyMap(), \ref stSupply().
497
    /// For example,
498
    /// \code
499
    ///   CostScaling<ListDigraph> cs(graph);
500
    ///   cs.lowerMap(lower).upperMap(upper).costMap(cost)
501
    ///     .supplyMap(sup).run();
502
    /// \endcode
503
    ///
504
    /// This function can be called more than once. All the parameters
505
    /// that have been given are kept for the next call, unless
506
    /// \ref reset() is called, thus only the modified parameters
507
    /// have to be set again. See \ref reset() for examples.
508
    /// However the underlying digraph must not be modified after this
509
    /// class have been constructed, since it copies the digraph.
356 510
    ///
357 511
    /// \param partial_augment By default the algorithm performs
358 512
    /// partial augment and relabel operations in the cost scaling
359 513
    /// phases. Set this parameter to \c false for using local push and
360 514
    /// relabel operations instead.
361 515
    ///
362
    /// \return \c true if a feasible flow can be found.
363
    bool run(bool partial_augment = true) {
364
      if (partial_augment) {
365
        return init() && startPartialAugment();
366
      } else {
367
        return init() && startPushRelabel();
516
    /// \return \c INFEASIBLE if no feasible flow exists,
517
    /// \n \c OPTIMAL if the problem has optimal solution
518
    /// (i.e. it is feasible and bounded), and the algorithm has found
519
    /// optimal flow and node potentials (primal and dual solutions),
520
    /// \n \c UNBOUNDED if the digraph contains an arc of negative cost
521
    /// and infinite upper bound. It means that the objective function
522
    /// is unbounded on that arc, however note that it could actually be
523
    /// bounded over the feasible flows, but this algroithm cannot handle
524
    /// these cases.
525
    ///
526
    /// \see ProblemType
527
    ProblemType run(bool partial_augment = true) {
528
      ProblemType pt = init();
529
      if (pt != OPTIMAL) return pt;
530
      start(partial_augment);
531
      return OPTIMAL;
532
    }
533

	
534
    /// \brief Reset all the parameters that have been given before.
535
    ///
536
    /// This function resets all the paramaters that have been given
537
    /// before using functions \ref lowerMap(), \ref upperMap(),
538
    /// \ref costMap(), \ref supplyMap(), \ref stSupply().
539
    ///
540
    /// It is useful for multiple run() calls. If this function is not
541
    /// used, all the parameters given before are kept for the next
542
    /// \ref run() call.
543
    /// However the underlying digraph must not be modified after this
544
    /// class have been constructed, since it copies and extends the graph.
545
    ///
546
    /// For example,
547
    /// \code
548
    ///   CostScaling<ListDigraph> cs(graph);
549
    ///
550
    ///   // First run
551
    ///   cs.lowerMap(lower).upperMap(upper).costMap(cost)
552
    ///     .supplyMap(sup).run();
553
    ///
554
    ///   // Run again with modified cost map (reset() is not called,
555
    ///   // so only the cost map have to be set again)
556
    ///   cost[e] += 100;
557
    ///   cs.costMap(cost).run();
558
    ///
559
    ///   // Run again from scratch using reset()
560
    ///   // (the lower bounds will be set to zero on all arcs)
561
    ///   cs.reset();
562
    ///   cs.upperMap(capacity).costMap(cost)
563
    ///     .supplyMap(sup).run();
564
    /// \endcode
565
    ///
566
    /// \return <tt>(*this)</tt>
567
    CostScaling& reset() {
568
      for (int i = 0; i != _res_node_num; ++i) {
569
        _supply[i] = 0;
368 570
      }
571
      int limit = _first_out[_root];
572
      for (int j = 0; j != limit; ++j) {
573
        _lower[j] = 0;
574
        _upper[j] = INF;
575
        _scost[j] = _forward[j] ? 1 : -1;
576
      }
577
      for (int j = limit; j != _res_arc_num; ++j) {
578
        _lower[j] = 0;
579
        _upper[j] = INF;
580
        _scost[j] = 0;
581
        _scost[_reverse[j]] = 0;
582
      }      
583
      _have_lower = false;
584
      return *this;
369 585
    }
370 586

	
371 587
    /// @}
372 588

	
373 589
    /// \name Query Functions
374
    /// The result of the algorithm can be obtained using these
590
    /// The results of the algorithm can be obtained using these
375 591
    /// functions.\n
376
    /// \ref lemon::CostScaling::run() "run()" must be called before
377
    /// using them.
592
    /// The \ref run() function must be called before using them.
378 593

	
379 594
    /// @{
380 595

	
381
    /// \brief Return a const reference to the arc map storing the
382
    /// found flow.
596
    /// \brief Return the total cost of the found flow.
383 597
    ///
384
    /// Return a const reference to the arc map storing the found flow.
598
    /// This function returns the total cost of the found flow.
599
    /// Its complexity is O(e).
600
    ///
601
    /// \note The return type of the function can be specified as a
602
    /// template parameter. For example,
603
    /// \code
604
    ///   cs.totalCost<double>();
605
    /// \endcode
606
    /// It is useful if the total cost cannot be stored in the \c Cost
607
    /// type of the algorithm, which is the default return type of the
608
    /// function.
385 609
    ///
386 610
    /// \pre \ref run() must be called before using this function.
387
    const FlowMap& flowMap() const {
388
      return *_flow;
611
    template <typename Number>
612
    Number totalCost() const {
613
      Number c = 0;
614
      for (ArcIt a(_graph); a != INVALID; ++a) {
615
        int i = _arc_idb[a];
616
        c += static_cast<Number>(_res_cap[i]) *
617
             (-static_cast<Number>(_scost[i]));
618
      }
619
      return c;
389 620
    }
390 621

	
391
    /// \brief Return a const reference to the node map storing the
392
    /// found potentials (the dual solution).
393
    ///
394
    /// Return a const reference to the node map storing the found
395
    /// potentials (the dual solution).
396
    ///
397
    /// \pre \ref run() must be called before using this function.
398
    const PotentialMap& potentialMap() const {
399
      return *_potential;
622
#ifndef DOXYGEN
623
    Cost totalCost() const {
624
      return totalCost<Cost>();
400 625
    }
626
#endif
401 627

	
402 628
    /// \brief Return the flow on the given arc.
403 629
    ///
404
    /// Return the flow on the given arc.
630
    /// This function returns the flow on the given arc.
405 631
    ///
406 632
    /// \pre \ref run() must be called before using this function.
407
    Capacity flow(const Arc& arc) const {
408
      return (*_flow)[arc];
633
    Value flow(const Arc& a) const {
634
      return _res_cap[_arc_idb[a]];
409 635
    }
410 636

	
411
    /// \brief Return the potential of the given node.
637
    /// \brief Return the flow map (the primal solution).
412 638
    ///
413
    /// Return the potential of the given node.
639
    /// This function copies the flow value on each arc into the given
640
    /// map. The \c Value type of the algorithm must be convertible to
641
    /// the \c Value type of the map.
414 642
    ///
415 643
    /// \pre \ref run() must be called before using this function.
416
    Cost potential(const Node& node) const {
417
      return (*_potential)[node];
644
    template <typename FlowMap>
645
    void flowMap(FlowMap &map) const {
646
      for (ArcIt a(_graph); a != INVALID; ++a) {
647
        map.set(a, _res_cap[_arc_idb[a]]);
648
      }
418 649
    }
419 650

	
420
    /// \brief Return the total cost of the found flow.
651
    /// \brief Return the potential (dual value) of the given node.
421 652
    ///
422
    /// Return the total cost of the found flow. The complexity of the
423
    /// function is \f$ O(e) \f$.
653
    /// This function returns the potential (dual value) of the
654
    /// given node.
424 655
    ///
425 656
    /// \pre \ref run() must be called before using this function.
426
    Cost totalCost() const {
427
      Cost c = 0;
428
      for (ArcIt e(_graph); e != INVALID; ++e)
429
        c += (*_flow)[e] * _orig_cost[e];
430
      return c;
657
    Cost potential(const Node& n) const {
658
      return static_cast<Cost>(_pi[_node_id[n]]);
659
    }
660

	
661
    /// \brief Return the potential map (the dual solution).
662
    ///
663
    /// This function copies the potential (dual value) of each node
664
    /// into the given map.
665
    /// The \c Cost type of the algorithm must be convertible to the
666
    /// \c Value type of the map.
667
    ///
668
    /// \pre \ref run() must be called before using this function.
669
    template <typename PotentialMap>
670
    void potentialMap(PotentialMap &map) const {
671
      for (NodeIt n(_graph); n != INVALID; ++n) {
672
        map.set(n, static_cast<Cost>(_pi[_node_id[n]]));
673
      }
431 674
    }
432 675

	
433 676
    /// @}
434 677

	
435 678
  private:
436 679

	
437
    /// Initialize the algorithm.
438
    bool init() {
439
      if (!_valid_supply) return false;
440
      // The scaling factor
680
    // Initialize the algorithm
681
    ProblemType init() {
682
      if (_res_node_num == 0) return INFEASIBLE;
683

	
684
      // Scaling factor
441 685
      _alpha = 8;
442 686

	
443
      // Initialize flow and potential maps
444
      if (!_flow) {
445
        _flow = new FlowMap(_graph);
446
        _local_flow = true;
687
      // Check the sum of supply values
688
      _sum_supply = 0;
689
      for (int i = 0; i != _root; ++i) {
690
        _sum_supply += _supply[i];
447 691
      }
448
      if (!_potential) {
449
        _potential = new PotentialMap(_graph);
450
        _local_potential = true;
692
      if (_sum_supply > 0) return INFEASIBLE;
693
      
694

	
695
      // Initialize vectors
696
      for (int i = 0; i != _res_node_num; ++i) {
697
        _pi[i] = 0;
698
        _excess[i] = _supply[i];
699
      }
700
      
701
      // Remove infinite upper bounds and check negative arcs
702
      const Value MAX = std::numeric_limits<Value>::max();
703
      int last_out;
704
      if (_have_lower) {
705
        for (int i = 0; i != _root; ++i) {
706
          last_out = _first_out[i+1];
707
          for (int j = _first_out[i]; j != last_out; ++j) {
708
            if (_forward[j]) {
709
              Value c = _scost[j] < 0 ? _upper[j] : _lower[j];
710
              if (c >= MAX) return UNBOUNDED;
711
              _excess[i] -= c;
712
              _excess[_target[j]] += c;
713
            }
714
          }
715
        }
716
      } else {
717
        for (int i = 0; i != _root; ++i) {
718
          last_out = _first_out[i+1];
719
          for (int j = _first_out[i]; j != last_out; ++j) {
720
            if (_forward[j] && _scost[j] < 0) {
721
              Value c = _upper[j];
722
              if (c >= MAX) return UNBOUNDED;
723
              _excess[i] -= c;
724
              _excess[_target[j]] += c;
725
            }
726
          }
727
        }
728
      }
729
      Value ex, max_cap = 0;
730
      for (int i = 0; i != _res_node_num; ++i) {
731
        ex = _excess[i];
732
        _excess[i] = 0;
733
        if (ex < 0) max_cap -= ex;
734
      }
735
      for (int j = 0; j != _res_arc_num; ++j) {
736
        if (_upper[j] >= MAX) _upper[j] = max_cap;
451 737
      }
452 738

	
453
      _red_cost = new ReducedCostMap(_graph, _cost, *_potential);
454
      _res_graph = new ResDigraph(_graph, _capacity, *_flow);
739
      // Initialize the large cost vector and the epsilon parameter
740
      _epsilon = 0;
741
      LargeCost lc;
742
      for (int i = 0; i != _root; ++i) {
743
        last_out = _first_out[i+1];
744
        for (int j = _first_out[i]; j != last_out; ++j) {
745
          lc = static_cast<LargeCost>(_scost[j]) * _res_node_num * _alpha;
746
          _cost[j] = lc;
747
          if (lc > _epsilon) _epsilon = lc;
748
        }
749
      }
750
      _epsilon /= _alpha;
455 751

	
456
      // Initialize the scaled cost map and the epsilon parameter
457
      Cost max_cost = 0;
458
      int node_num = countNodes(_graph);
459
      for (ArcIt e(_graph); e != INVALID; ++e) {
460
        _cost[e] = LCost(_orig_cost[e]) * node_num * _alpha;
461
        if (_orig_cost[e] > max_cost) max_cost = _orig_cost[e];
752
      // Initialize maps for Circulation and remove non-zero lower bounds
753
      ConstMap<Arc, Value> low(0);
754
      typedef typename Digraph::template ArcMap<Value> ValueArcMap;
755
      typedef typename Digraph::template NodeMap<Value> ValueNodeMap;
756
      ValueArcMap cap(_graph), flow(_graph);
757
      ValueNodeMap sup(_graph);
758
      for (NodeIt n(_graph); n != INVALID; ++n) {
759
        sup[n] = _supply[_node_id[n]];
462 760
      }
463
      _epsilon = max_cost * node_num;
761
      if (_have_lower) {
762
        for (ArcIt a(_graph); a != INVALID; ++a) {
763
          int j = _arc_idf[a];
764
          Value c = _lower[j];
765
          cap[a] = _upper[j] - c;
766
          sup[_graph.source(a)] -= c;
767
          sup[_graph.target(a)] += c;
768
        }
769
      } else {
770
        for (ArcIt a(_graph); a != INVALID; ++a) {
771
          cap[a] = _upper[_arc_idf[a]];
772
        }
773
      }
464 774

	
465 775
      // Find a feasible flow using Circulation
466
      Circulation< Digraph, ConstMap<Arc, Capacity>, CapacityArcMap,
467
                   SupplyMap >
468
        circulation( _graph, constMap<Arc>(Capacity(0)), _capacity,
469
                     _supply );
470
      return circulation.flowMap(*_flow).run();
776
      Circulation<Digraph, ConstMap<Arc, Value>, ValueArcMap, ValueNodeMap>
777
        circ(_graph, low, cap, sup);
778
      if (!circ.flowMap(flow).run()) return INFEASIBLE;
779

	
780
      // Set residual capacities and handle GEQ supply type
781
      if (_sum_supply < 0) {
782
        for (ArcIt a(_graph); a != INVALID; ++a) {
783
          Value fa = flow[a];
784
          _res_cap[_arc_idf[a]] = cap[a] - fa;
785
          _res_cap[_arc_idb[a]] = fa;
786
          sup[_graph.source(a)] -= fa;
787
          sup[_graph.target(a)] += fa;
788
        }
789
        for (NodeIt n(_graph); n != INVALID; ++n) {
790
          _excess[_node_id[n]] = sup[n];
791
        }
792
        for (int a = _first_out[_root]; a != _res_arc_num; ++a) {
793
          int u = _target[a];
794
          int ra = _reverse[a];
795
          _res_cap[a] = -_sum_supply + 1;
796
          _res_cap[ra] = -_excess[u];
797
          _cost[a] = 0;
798
          _cost[ra] = 0;
799
          _excess[u] = 0;
800
        }
801
      } else {
802
        for (ArcIt a(_graph); a != INVALID; ++a) {
803
          Value fa = flow[a];
804
          _res_cap[_arc_idf[a]] = cap[a] - fa;
805
          _res_cap[_arc_idb[a]] = fa;
806
        }
807
        for (int a = _first_out[_root]; a != _res_arc_num; ++a) {
808
          int ra = _reverse[a];
809
          _res_cap[a] = 1;
810
          _res_cap[ra] = 0;
811
          _cost[a] = 0;
812
          _cost[ra] = 0;
813
        }
814
      }
815
      
816
      return OPTIMAL;
817
    }
818

	
819
    // Execute the algorithm and transform the results
820
    void start(bool partial_augment) {
821
      // Execute the algorithm
822
      if (partial_augment) {
823
        startPartialAugment();
824
      } else {
825
        startPushRelabel();
826
      }
827

	
828
      // Compute node potentials for the original costs
829
      _arc_vec.clear();
830
      _cost_vec.clear();
831
      for (int j = 0; j != _res_arc_num; ++j) {
832
        if (_res_cap[j] > 0) {
833
          _arc_vec.push_back(IntPair(_source[j], _target[j]));
834
          _cost_vec.push_back(_scost[j]);
835
        }
836
      }
837
      _sgr.build(_res_node_num, _arc_vec.begin(), _arc_vec.end());
838

	
839
      typename BellmanFord<StaticDigraph, LargeCostArcMap>
840
        ::template SetDistMap<LargeCostNodeMap>::Create bf(_sgr, _cost_map);
841
      bf.distMap(_pi_map);
842
      bf.init(0);
843
      bf.start();
844

	
845
      // Handle non-zero lower bounds
846
      if (_have_lower) {
847
        int limit = _first_out[_root];
848
        for (int j = 0; j != limit; ++j) {
849
          if (!_forward[j]) _res_cap[j] += _lower[j];
850
        }
851
      }
471 852
    }
472 853

	
473 854
    /// Execute the algorithm performing partial augmentation and
474
    /// relabel operations.
475
    bool startPartialAugment() {
855
    /// relabel operations
856
    void startPartialAugment() {
476 857
      // Paramters for heuristics
477
//      const int BF_HEURISTIC_EPSILON_BOUND = 1000;
478
//      const int BF_HEURISTIC_BOUND_FACTOR  = 3;
858
      const int BF_HEURISTIC_EPSILON_BOUND = 1000;
859
      const int BF_HEURISTIC_BOUND_FACTOR  = 3;
479 860
      // Maximum augment path length
480 861
      const int MAX_PATH_LENGTH = 4;
481 862

	
482
      // Variables
483
      typename Digraph::template NodeMap<Arc> pred_arc(_graph);
484
      typename Digraph::template NodeMap<bool> forward(_graph);
485
      typename Digraph::template NodeMap<OutArcIt> next_out(_graph);
486
      typename Digraph::template NodeMap<InArcIt> next_in(_graph);
487
      typename Digraph::template NodeMap<bool> next_dir(_graph);
488
      std::deque<Node> active_nodes;
489
      std::vector<Node> path_nodes;
490

	
491
//      int node_num = countNodes(_graph);
863
      // Perform cost scaling phases
864
      IntVector pred_arc(_res_node_num);
865
      std::vector<int> path_nodes;
492 866
      for ( ; _epsilon >= 1; _epsilon = _epsilon < _alpha && _epsilon > 1 ?
493 867
                                        1 : _epsilon / _alpha )
494 868
      {
495
/*
496 869
        // "Early Termination" heuristic: use Bellman-Ford algorithm
497 870
        // to check if the current flow is optimal
498 871
        if (_epsilon <= BF_HEURISTIC_EPSILON_BOUND) {
499
          typedef ShiftMap< ResidualCostMap<LargeCostMap> > ShiftCostMap;
500
          ShiftCostMap shift_cost(_res_cost, 1);
501
          BellmanFord<ResDigraph, ShiftCostMap> bf(*_res_graph, shift_cost);
872
          _arc_vec.clear();
873
          _cost_vec.clear();
874
          for (int j = 0; j != _res_arc_num; ++j) {
875
            if (_res_cap[j] > 0) {
876
              _arc_vec.push_back(IntPair(_source[j], _target[j]));
877
              _cost_vec.push_back(_cost[j] + 1);
878
            }
879
          }
880
          _sgr.build(_res_node_num, _arc_vec.begin(), _arc_vec.end());
881

	
882
          BellmanFord<StaticDigraph, LargeCostArcMap> bf(_sgr, _cost_map);
502 883
          bf.init(0);
503 884
          bool done = false;
504
          int K = int(BF_HEURISTIC_BOUND_FACTOR * sqrt(node_num));
885
          int K = int(BF_HEURISTIC_BOUND_FACTOR * sqrt(_res_node_num));
505 886
          for (int i = 0; i < K && !done; ++i)
506 887
            done = bf.processNextWeakRound();
507 888
          if (done) break;
508 889
        }
509
*/
890

	
510 891
        // Saturate arcs not satisfying the optimality condition
511
        Capacity delta;
512
        for (ArcIt e(_graph); e != INVALID; ++e) {
513
          if (_capacity[e] - (*_flow)[e] > 0 && (*_red_cost)[e] < 0) {
514
            delta = _capacity[e] - (*_flow)[e];
515
            _excess[_graph.source(e)] -= delta;
516
            _excess[_graph.target(e)] += delta;
517
            (*_flow)[e] = _capacity[e];
518
          }
519
          if ((*_flow)[e] > 0 && -(*_red_cost)[e] < 0) {
520
            _excess[_graph.target(e)] -= (*_flow)[e];
521
            _excess[_graph.source(e)] += (*_flow)[e];
522
            (*_flow)[e] = 0;
892
        for (int a = 0; a != _res_arc_num; ++a) {
893
          if (_res_cap[a] > 0 &&
894
              _cost[a] + _pi[_source[a]] - _pi[_target[a]] < 0) {
895
            Value delta = _res_cap[a];
896
            _excess[_source[a]] -= delta;
897
            _excess[_target[a]] += delta;
898
            _res_cap[a] = 0;
899
            _res_cap[_reverse[a]] += delta;
523 900
          }
524 901
        }
525

	
902
        
526 903
        // Find active nodes (i.e. nodes with positive excess)
527
        for (NodeIt n(_graph); n != INVALID; ++n) {
528
          if (_excess[n] > 0) active_nodes.push_back(n);
904
        for (int u = 0; u != _res_node_num; ++u) {
905
          if (_excess[u] > 0) _active_nodes.push_back(u);
529 906
        }
530 907

	
531
        // Initialize the next arc maps
532
        for (NodeIt n(_graph); n != INVALID; ++n) {
533
          next_out[n] = OutArcIt(_graph, n);
534
          next_in[n] = InArcIt(_graph, n);
535
          next_dir[n] = true;
908
        // Initialize the next arcs
909
        for (int u = 0; u != _res_node_num; ++u) {
910
          _next_out[u] = _first_out[u];
536 911
        }
537 912

	
538 913
        // Perform partial augment and relabel operations
539
        while (active_nodes.size() > 0) {
914
        while (true) {
540 915
          // Select an active node (FIFO selection)
541
          if (_excess[active_nodes[0]] <= 0) {
542
            active_nodes.pop_front();
543
            continue;
916
          while (_active_nodes.size() > 0 &&
917
                 _excess[_active_nodes.front()] <= 0) {
918
            _active_nodes.pop_front();
544 919
          }
545
          Node start = active_nodes[0];
920
          if (_active_nodes.size() == 0) break;
921
          int start = _active_nodes.front();
546 922
          path_nodes.clear();
547 923
          path_nodes.push_back(start);
548 924

	
549 925
          // Find an augmenting path from the start node
550
          Node u, tip = start;
551
          LCost min_red_cost;
552
          while ( _excess[tip] >= 0 &&
553
                  int(path_nodes.size()) <= MAX_PATH_LENGTH )
554
          {
555
            if (next_dir[tip]) {
556
              for (OutArcIt e = next_out[tip]; e != INVALID; ++e) {
557
                if (_capacity[e] - (*_flow)[e] > 0 && (*_red_cost)[e] < 0) {
558
                  u = _graph.target(e);
559
                  pred_arc[u] = e;
560
                  forward[u] = true;
561
                  next_out[tip] = e;
562
                  tip = u;
563
                  path_nodes.push_back(tip);
564
                  goto next_step;
565
                }
566
              }
567
              next_dir[tip] = false;
568
            }
569
            for (InArcIt e = next_in[tip]; e != INVALID; ++e) {
570
              if ((*_flow)[e] > 0 && -(*_red_cost)[e] < 0) {
571
                u = _graph.source(e);
572
                pred_arc[u] = e;
573
                forward[u] = false;
574
                next_in[tip] = e;
926
          int tip = start;
927
          while (_excess[tip] >= 0 &&
928
                 int(path_nodes.size()) <= MAX_PATH_LENGTH) {
929
            int u;
930
            LargeCost min_red_cost, rc;
931
            int last_out = _sum_supply < 0 ?
932
              _first_out[tip+1] : _first_out[tip+1] - 1;
933
            for (int a = _next_out[tip]; a != last_out; ++a) {
934
              if (_res_cap[a] > 0 &&
935
                  _cost[a] + _pi[_source[a]] - _pi[_target[a]] < 0) {
936
                u = _target[a];
937
                pred_arc[u] = a;
938
                _next_out[tip] = a;
575 939
                tip = u;
576 940
                path_nodes.push_back(tip);
577 941
                goto next_step;
... ...
@@ -579,266 +943,186 @@
579 943
            }
580 944

	
581 945
            // Relabel tip node
582
            min_red_cost = std::numeric_limits<LCost>::max() / 2;
583
            for (OutArcIt oe(_graph, tip); oe != INVALID; ++oe) {
584
              if ( _capacity[oe] - (*_flow)[oe] > 0 &&
585
                   (*_red_cost)[oe] < min_red_cost )
586
                min_red_cost = (*_red_cost)[oe];
946
            min_red_cost = std::numeric_limits<LargeCost>::max() / 2;
947
            for (int a = _first_out[tip]; a != last_out; ++a) {
948
              rc = _cost[a] + _pi[_source[a]] - _pi[_target[a]];
949
              if (_res_cap[a] > 0 && rc < min_red_cost) {
950
                min_red_cost = rc;
951
              }
587 952
            }
588
            for (InArcIt ie(_graph, tip); ie != INVALID; ++ie) {
589
              if ((*_flow)[ie] > 0 && -(*_red_cost)[ie] < min_red_cost)
590
                min_red_cost = -(*_red_cost)[ie];
591
            }
592
            (*_potential)[tip] -= min_red_cost + _epsilon;
953
            _pi[tip] -= min_red_cost + _epsilon;
593 954

	
594
            // Reset the next arc maps
595
            next_out[tip] = OutArcIt(_graph, tip);
596
            next_in[tip] = InArcIt(_graph, tip);
597
            next_dir[tip] = true;
955
            // Reset the next arc of tip
956
            _next_out[tip] = _first_out[tip];
598 957

	
599 958
            // Step back
600 959
            if (tip != start) {
601 960
              path_nodes.pop_back();
602
              tip = path_nodes[path_nodes.size()-1];
961
              tip = path_nodes.back();
603 962
            }
604 963

	
605
          next_step:
606
            continue;
964
          next_step: ;
607 965
          }
608 966

	
609 967
          // Augment along the found path (as much flow as possible)
610
          Capacity delta;
968
          Value delta;
969
          int u, v = path_nodes.front(), pa;
611 970
          for (int i = 1; i < int(path_nodes.size()); ++i) {
612
            u = path_nodes[i];
613
            delta = forward[u] ?
614
              _capacity[pred_arc[u]] - (*_flow)[pred_arc[u]] :
615
              (*_flow)[pred_arc[u]];
616
            delta = std::min(delta, _excess[path_nodes[i-1]]);
617
            (*_flow)[pred_arc[u]] += forward[u] ? delta : -delta;
618
            _excess[path_nodes[i-1]] -= delta;
619
            _excess[u] += delta;
620
            if (_excess[u] > 0 && _excess[u] <= delta) active_nodes.push_back(u);
971
            u = v;
972
            v = path_nodes[i];
973
            pa = pred_arc[v];
974
            delta = std::min(_res_cap[pa], _excess[u]);
975
            _res_cap[pa] -= delta;
976
            _res_cap[_reverse[pa]] += delta;
977
            _excess[u] -= delta;
978
            _excess[v] += delta;
979
            if (_excess[v] > 0 && _excess[v] <= delta)
980
              _active_nodes.push_back(v);
621 981
          }
622 982
        }
623 983
      }
624

	
625
      // Compute node potentials for the original costs
626
      ResidualCostMap<CostMap> res_cost(_orig_cost);
627
      BellmanFord< ResDigraph, ResidualCostMap<CostMap> >
628
        bf(*_res_graph, res_cost);
629
      bf.init(0); bf.start();
630
      for (NodeIt n(_graph); n != INVALID; ++n)
631
        (*_potential)[n] = bf.dist(n);
632

	
633
      // Handle non-zero lower bounds
634
      if (_lower) {
635
        for (ArcIt e(_graph); e != INVALID; ++e)
636
          (*_flow)[e] += (*_lower)[e];
637
      }
638
      return true;
639 984
    }
640 985

	
641
    /// Execute the algorithm performing push and relabel operations.
642
    bool startPushRelabel() {
986
    /// Execute the algorithm performing push and relabel operations
987
    void startPushRelabel() {
643 988
      // Paramters for heuristics
644
//      const int BF_HEURISTIC_EPSILON_BOUND = 1000;
645
//      const int BF_HEURISTIC_BOUND_FACTOR  = 3;
989
      const int BF_HEURISTIC_EPSILON_BOUND = 1000;
990
      const int BF_HEURISTIC_BOUND_FACTOR  = 3;
646 991

	
647
      typename Digraph::template NodeMap<bool> hyper(_graph, false);
648
      typename Digraph::template NodeMap<Arc> pred_arc(_graph);
649
      typename Digraph::template NodeMap<bool> forward(_graph);
650
      typename Digraph::template NodeMap<OutArcIt> next_out(_graph);
651
      typename Digraph::template NodeMap<InArcIt> next_in(_graph);
652
      typename Digraph::template NodeMap<bool> next_dir(_graph);
653
      std::deque<Node> active_nodes;
654

	
655
//      int node_num = countNodes(_graph);
992
      // Perform cost scaling phases
993
      BoolVector hyper(_res_node_num, false);
656 994
      for ( ; _epsilon >= 1; _epsilon = _epsilon < _alpha && _epsilon > 1 ?
657 995
                                        1 : _epsilon / _alpha )
658 996
      {
659
/*
660 997
        // "Early Termination" heuristic: use Bellman-Ford algorithm
661 998
        // to check if the current flow is optimal
662 999
        if (_epsilon <= BF_HEURISTIC_EPSILON_BOUND) {
663
          typedef ShiftMap< ResidualCostMap<LargeCostMap> > ShiftCostMap;
664
          ShiftCostMap shift_cost(_res_cost, 1);
665
          BellmanFord<ResDigraph, ShiftCostMap> bf(*_res_graph, shift_cost);
1000
          _arc_vec.clear();
1001
          _cost_vec.clear();
1002
          for (int j = 0; j != _res_arc_num; ++j) {
1003
            if (_res_cap[j] > 0) {
1004
              _arc_vec.push_back(IntPair(_source[j], _target[j]));
1005
              _cost_vec.push_back(_cost[j] + 1);
1006
            }
1007
          }
1008
          _sgr.build(_res_node_num, _arc_vec.begin(), _arc_vec.end());
1009

	
1010
          BellmanFord<StaticDigraph, LargeCostArcMap> bf(_sgr, _cost_map);
666 1011
          bf.init(0);
667 1012
          bool done = false;
668
          int K = int(BF_HEURISTIC_BOUND_FACTOR * sqrt(node_num));
1013
          int K = int(BF_HEURISTIC_BOUND_FACTOR * sqrt(_res_node_num));
669 1014
          for (int i = 0; i < K && !done; ++i)
670 1015
            done = bf.processNextWeakRound();
671 1016
          if (done) break;
672 1017
        }
673
*/
674 1018

	
675 1019
        // Saturate arcs not satisfying the optimality condition
676
        Capacity delta;
677
        for (ArcIt e(_graph); e != INVALID; ++e) {
678
          if (_capacity[e] - (*_flow)[e] > 0 && (*_red_cost)[e] < 0) {
679
            delta = _capacity[e] - (*_flow)[e];
680
            _excess[_graph.source(e)] -= delta;
681
            _excess[_graph.target(e)] += delta;
682
            (*_flow)[e] = _capacity[e];
683
          }
684
          if ((*_flow)[e] > 0 && -(*_red_cost)[e] < 0) {
685
            _excess[_graph.target(e)] -= (*_flow)[e];
686
            _excess[_graph.source(e)] += (*_flow)[e];
687
            (*_flow)[e] = 0;
1020
        for (int a = 0; a != _res_arc_num; ++a) {
1021
          if (_res_cap[a] > 0 &&
1022
              _cost[a] + _pi[_source[a]] - _pi[_target[a]] < 0) {
1023
            Value delta = _res_cap[a];
1024
            _excess[_source[a]] -= delta;
1025
            _excess[_target[a]] += delta;
1026
            _res_cap[a] = 0;
1027
            _res_cap[_reverse[a]] += delta;
688 1028
          }
689 1029
        }
690 1030

	
691 1031
        // Find active nodes (i.e. nodes with positive excess)
692
        for (NodeIt n(_graph); n != INVALID; ++n) {
693
          if (_excess[n] > 0) active_nodes.push_back(n);
1032
        for (int u = 0; u != _res_node_num; ++u) {
1033
          if (_excess[u] > 0) _active_nodes.push_back(u);
694 1034
        }
695 1035

	
696
        // Initialize the next arc maps
697
        for (NodeIt n(_graph); n != INVALID; ++n) {
698
          next_out[n] = OutArcIt(_graph, n);
699
          next_in[n] = InArcIt(_graph, n);
700
          next_dir[n] = true;
1036
        // Initialize the next arcs
1037
        for (int u = 0; u != _res_node_num; ++u) {
1038
          _next_out[u] = _first_out[u];
701 1039
        }
702 1040

	
703 1041
        // Perform push and relabel operations
704
        while (active_nodes.size() > 0) {
1042
        while (_active_nodes.size() > 0) {
1043
          LargeCost min_red_cost, rc;
1044
          Value delta;
1045
          int n, t, a, last_out = _res_arc_num;
1046

	
705 1047
          // Select an active node (FIFO selection)
706
          Node n = active_nodes[0], t;
707
          bool relabel_enabled = true;
1048
        next_node:
1049
          n = _active_nodes.front();
1050
          last_out = _sum_supply < 0 ?
1051
            _first_out[n+1] : _first_out[n+1] - 1;
708 1052

	
709 1053
          // Perform push operations if there are admissible arcs
710
          if (_excess[n] > 0 && next_dir[n]) {
711
            OutArcIt e = next_out[n];
712
            for ( ; e != INVALID; ++e) {
713
              if (_capacity[e] - (*_flow)[e] > 0 && (*_red_cost)[e] < 0) {
714
                delta = std::min(_capacity[e] - (*_flow)[e], _excess[n]);
715
                t = _graph.target(e);
1054
          if (_excess[n] > 0) {
1055
            for (a = _next_out[n]; a != last_out; ++a) {
1056
              if (_res_cap[a] > 0 &&
1057
                  _cost[a] + _pi[_source[a]] - _pi[_target[a]] < 0) {
1058
                delta = std::min(_res_cap[a], _excess[n]);
1059
                t = _target[a];
716 1060

	
717 1061
                // Push-look-ahead heuristic
718
                Capacity ahead = -_excess[t];
719
                for (OutArcIt oe(_graph, t); oe != INVALID; ++oe) {
720
                  if (_capacity[oe] - (*_flow)[oe] > 0 && (*_red_cost)[oe] < 0)
721
                    ahead += _capacity[oe] - (*_flow)[oe];
722
                }
723
                for (InArcIt ie(_graph, t); ie != INVALID; ++ie) {
724
                  if ((*_flow)[ie] > 0 && -(*_red_cost)[ie] < 0)
725
                    ahead += (*_flow)[ie];
1062
                Value ahead = -_excess[t];
1063
                int last_out_t = _sum_supply < 0 ?
1064
                  _first_out[t+1] : _first_out[t+1] - 1;
1065
                for (int ta = _next_out[t]; ta != last_out_t; ++ta) {
1066
                  if (_res_cap[ta] > 0 && 
1067
                      _cost[ta] + _pi[_source[ta]] - _pi[_target[ta]] < 0)
1068
                    ahead += _res_cap[ta];
1069
                  if (ahead >= delta) break;
726 1070
                }
727 1071
                if (ahead < 0) ahead = 0;
728 1072

	
729 1073
                // Push flow along the arc
730 1074
                if (ahead < delta) {
731
                  (*_flow)[e] += ahead;
1075
                  _res_cap[a] -= ahead;
1076
                  _res_cap[_reverse[a]] += ahead;
732 1077
                  _excess[n] -= ahead;
733 1078
                  _excess[t] += ahead;
734
                  active_nodes.push_front(t);
1079
                  _active_nodes.push_front(t);
735 1080
                  hyper[t] = true;
736
                  relabel_enabled = false;
737
                  break;
1081
                  _next_out[n] = a;
1082
                  goto next_node;
738 1083
                } else {
739
                  (*_flow)[e] += delta;
1084
                  _res_cap[a] -= delta;
1085
                  _res_cap[_reverse[a]] += delta;
740 1086
                  _excess[n] -= delta;
741 1087
                  _excess[t] += delta;
742 1088
                  if (_excess[t] > 0 && _excess[t] <= delta)
743
                    active_nodes.push_back(t);
1089
                    _active_nodes.push_back(t);
744 1090
                }
745 1091

	
746
                if (_excess[n] == 0) break;
1092
                if (_excess[n] == 0) {
1093
                  _next_out[n] = a;
1094
                  goto remove_nodes;
1095
                }
747 1096
              }
748 1097
            }
749
            if (e != INVALID) {
750
              next_out[n] = e;
751
            } else {
752
              next_dir[n] = false;
753
            }
754
          }
755

	
756
          if (_excess[n] > 0 && !next_dir[n]) {
757
            InArcIt e = next_in[n];
758
            for ( ; e != INVALID; ++e) {
759
              if ((*_flow)[e] > 0 && -(*_red_cost)[e] < 0) {
760
                delta = std::min((*_flow)[e], _excess[n]);
761
                t = _graph.source(e);
762

	
763
                // Push-look-ahead heuristic
764
                Capacity ahead = -_excess[t];
765
                for (OutArcIt oe(_graph, t); oe != INVALID; ++oe) {
766
                  if (_capacity[oe] - (*_flow)[oe] > 0 && (*_red_cost)[oe] < 0)
767
                    ahead += _capacity[oe] - (*_flow)[oe];
768
                }
769
                for (InArcIt ie(_graph, t); ie != INVALID; ++ie) {
770
                  if ((*_flow)[ie] > 0 && -(*_red_cost)[ie] < 0)
771
                    ahead += (*_flow)[ie];
772
                }
773
                if (ahead < 0) ahead = 0;
774

	
775
                // Push flow along the arc
776
                if (ahead < delta) {
777
                  (*_flow)[e] -= ahead;
778
                  _excess[n] -= ahead;
779
                  _excess[t] += ahead;
780
                  active_nodes.push_front(t);
781
                  hyper[t] = true;
782
                  relabel_enabled = false;
783
                  break;
784
                } else {
785
                  (*_flow)[e] -= delta;
786
                  _excess[n] -= delta;
787
                  _excess[t] += delta;
788
                  if (_excess[t] > 0 && _excess[t] <= delta)
789
                    active_nodes.push_back(t);
790
                }
791

	
792
                if (_excess[n] == 0) break;
793
              }
794
            }
795
            next_in[n] = e;
1098
            _next_out[n] = a;
796 1099
          }
797 1100

	
798 1101
          // Relabel the node if it is still active (or hyper)
799
          if (relabel_enabled && (_excess[n] > 0 || hyper[n])) {
800
            LCost min_red_cost = std::numeric_limits<LCost>::max() / 2;
801
            for (OutArcIt oe(_graph, n); oe != INVALID; ++oe) {
802
              if ( _capacity[oe] - (*_flow)[oe] > 0 &&
803
                   (*_red_cost)[oe] < min_red_cost )
804
                min_red_cost = (*_red_cost)[oe];
1102
          if (_excess[n] > 0 || hyper[n]) {
1103
            min_red_cost = std::numeric_limits<LargeCost>::max() / 2;
1104
            for (int a = _first_out[n]; a != last_out; ++a) {
1105
              rc = _cost[a] + _pi[_source[a]] - _pi[_target[a]];
1106
              if (_res_cap[a] > 0 && rc < min_red_cost) {
1107
                min_red_cost = rc;
1108
              }
805 1109
            }
806
            for (InArcIt ie(_graph, n); ie != INVALID; ++ie) {
807
              if ((*_flow)[ie] > 0 && -(*_red_cost)[ie] < min_red_cost)
808
                min_red_cost = -(*_red_cost)[ie];
809
            }
810
            (*_potential)[n] -= min_red_cost + _epsilon;
1110
            _pi[n] -= min_red_cost + _epsilon;
811 1111
            hyper[n] = false;
812 1112

	
813
            // Reset the next arc maps
814
            next_out[n] = OutArcIt(_graph, n);
815
            next_in[n] = InArcIt(_graph, n);
816
            next_dir[n] = true;
1113
            // Reset the next arc
1114
            _next_out[n] = _first_out[n];
817 1115
          }
818

	
1116
        
819 1117
          // Remove nodes that are not active nor hyper
820
          while ( active_nodes.size() > 0 &&
821
                  _excess[active_nodes[0]] <= 0 &&
822
                  !hyper[active_nodes[0]] ) {
823
            active_nodes.pop_front();
1118
        remove_nodes:
1119
          while ( _active_nodes.size() > 0 &&
1120
                  _excess[_active_nodes.front()] <= 0 &&
1121
                  !hyper[_active_nodes.front()] ) {
1122
            _active_nodes.pop_front();
824 1123
          }
825 1124
        }
826 1125
      }
827

	
828
      // Compute node potentials for the original costs
829
      ResidualCostMap<CostMap> res_cost(_orig_cost);
830
      BellmanFord< ResDigraph, ResidualCostMap<CostMap> >
831
        bf(*_res_graph, res_cost);
832
      bf.init(0); bf.start();
833
      for (NodeIt n(_graph); n != INVALID; ++n)
834
        (*_potential)[n] = bf.dist(n);
835

	
836
      // Handle non-zero lower bounds
837
      if (_lower) {
838
        for (ArcIt e(_graph); e != INVALID; ++e)
839
          (*_flow)[e] += (*_lower)[e];
840
      }
841
      return true;
842 1126
    }
843 1127

	
844 1128
  }; //class CostScaling
0 comments (0 inline)