gravatar
retvari@tmit.bme.hu
retvari@tmit.bme.hu
Fix LpBase::Constr two-side limit bug (#430)
0 2 0
default
2 files changed with 12 insertions and 4 deletions:
↑ Collapse diff ↑
Ignore white space 6 line context
... ...
@@ -1095,994 +1095,994 @@
1095 1095

	
1096 1096
    ///\param e is a dual linear expression (see \ref DualExpr)
1097 1097
    ///\param o is the corresponding component of the objective
1098 1098
    ///function. It is 0 by default.
1099 1099
    ///\return The created column.
1100 1100
    Col addCol(const DualExpr &e, Value o = 0) {
1101 1101
      Col c=addCol();
1102 1102
      col(c,e);
1103 1103
      objCoeff(c,o);
1104 1104
      return c;
1105 1105
    }
1106 1106

	
1107 1107
    ///Add a new empty row (i.e a new constraint) to the LP
1108 1108

	
1109 1109
    ///This function adds a new empty row (i.e a new constraint) to the LP.
1110 1110
    ///\return The created row
1111 1111
    Row addRow() { Row r; r._id = _addRowId(_addRow()); return r;}
1112 1112

	
1113 1113
    ///\brief Add several new rows (i.e constraints) at once
1114 1114
    ///
1115 1115
    ///This magic function takes a container as its argument and fills
1116 1116
    ///its elements with new row (i.e. variables)
1117 1117
    ///\param t can be
1118 1118
    ///- a standard STL compatible iterable container with
1119 1119
    ///\ref Row as its \c values_type like
1120 1120
    ///\code
1121 1121
    ///std::vector<LpBase::Row>
1122 1122
    ///std::list<LpBase::Row>
1123 1123
    ///\endcode
1124 1124
    ///- a standard STL compatible iterable container with
1125 1125
    ///\ref Row as its \c mapped_type like
1126 1126
    ///\code
1127 1127
    ///std::map<AnyType,LpBase::Row>
1128 1128
    ///\endcode
1129 1129
    ///- an iterable lemon \ref concepts::WriteMap "write map" like
1130 1130
    ///\code
1131 1131
    ///ListGraph::NodeMap<LpBase::Row>
1132 1132
    ///ListGraph::ArcMap<LpBase::Row>
1133 1133
    ///\endcode
1134 1134
    ///\return The number of rows created.
1135 1135
#ifdef DOXYGEN
1136 1136
    template<class T>
1137 1137
    int addRowSet(T &t) { return 0;}
1138 1138
#else
1139 1139
    template<class T>
1140 1140
    typename enable_if<typename T::value_type::LpRow,int>::type
1141 1141
    addRowSet(T &t, dummy<0> = 0) {
1142 1142
      int s=0;
1143 1143
      for(typename T::iterator i=t.begin();i!=t.end();++i) {*i=addRow();s++;}
1144 1144
      return s;
1145 1145
    }
1146 1146
    template<class T>
1147 1147
    typename enable_if<typename T::value_type::second_type::LpRow, int>::type
1148 1148
    addRowSet(T &t, dummy<1> = 1) {
1149 1149
      int s=0;
1150 1150
      for(typename T::iterator i=t.begin();i!=t.end();++i) {
1151 1151
        i->second=addRow();
1152 1152
        s++;
1153 1153
      }
1154 1154
      return s;
1155 1155
    }
1156 1156
    template<class T>
1157 1157
    typename enable_if<typename T::MapIt::Value::LpRow, int>::type
1158 1158
    addRowSet(T &t, dummy<2> = 2) {
1159 1159
      int s=0;
1160 1160
      for(typename T::MapIt i(t); i!=INVALID; ++i)
1161 1161
        {
1162 1162
          i.set(addRow());
1163 1163
          s++;
1164 1164
        }
1165 1165
      return s;
1166 1166
    }
1167 1167
#endif
1168 1168

	
1169 1169
    ///Set a row (i.e a constraint) of the LP
1170 1170

	
1171 1171
    ///\param r is the row to be modified
1172 1172
    ///\param l is lower bound (-\ref INF means no bound)
1173 1173
    ///\param e is a linear expression (see \ref Expr)
1174 1174
    ///\param u is the upper bound (\ref INF means no bound)
1175 1175
    void row(Row r, Value l, const Expr &e, Value u) {
1176 1176
      e.simplify();
1177 1177
      _setRowCoeffs(rows(id(r)), ExprIterator(e.comps.begin(), cols),
1178 1178
                    ExprIterator(e.comps.end(), cols));
1179 1179
      _setRowLowerBound(rows(id(r)),l - *e);
1180 1180
      _setRowUpperBound(rows(id(r)),u - *e);
1181 1181
    }
1182 1182

	
1183 1183
    ///Set a row (i.e a constraint) of the LP
1184 1184

	
1185 1185
    ///\param r is the row to be modified
1186 1186
    ///\param c is a linear expression (see \ref Constr)
1187 1187
    void row(Row r, const Constr &c) {
1188 1188
      row(r, c.lowerBounded()?c.lowerBound():-INF,
1189 1189
          c.expr(), c.upperBounded()?c.upperBound():INF);
1190 1190
    }
1191 1191

	
1192 1192

	
1193 1193
    ///Get a row (i.e a constraint) of the LP
1194 1194

	
1195 1195
    ///\param r is the row to get
1196 1196
    ///\return the expression associated to the row
1197 1197
    Expr row(Row r) const {
1198 1198
      Expr e;
1199 1199
      _getRowCoeffs(rows(id(r)), InsertIterator(e.comps, cols));
1200 1200
      return e;
1201 1201
    }
1202 1202

	
1203 1203
    ///Add a new row (i.e a new constraint) to the LP
1204 1204

	
1205 1205
    ///\param l is the lower bound (-\ref INF means no bound)
1206 1206
    ///\param e is a linear expression (see \ref Expr)
1207 1207
    ///\param u is the upper bound (\ref INF means no bound)
1208 1208
    ///\return The created row.
1209 1209
    Row addRow(Value l,const Expr &e, Value u) {
1210 1210
      Row r=addRow();
1211 1211
      row(r,l,e,u);
1212 1212
      return r;
1213 1213
    }
1214 1214

	
1215 1215
    ///Add a new row (i.e a new constraint) to the LP
1216 1216

	
1217 1217
    ///\param c is a linear expression (see \ref Constr)
1218 1218
    ///\return The created row.
1219 1219
    Row addRow(const Constr &c) {
1220 1220
      Row r=addRow();
1221 1221
      row(r,c);
1222 1222
      return r;
1223 1223
    }
1224 1224
    ///Erase a column (i.e a variable) from the LP
1225 1225

	
1226 1226
    ///\param c is the column to be deleted
1227 1227
    void erase(Col c) {
1228 1228
      _eraseCol(cols(id(c)));
1229 1229
      _eraseColId(cols(id(c)));
1230 1230
    }
1231 1231
    ///Erase a row (i.e a constraint) from the LP
1232 1232

	
1233 1233
    ///\param r is the row to be deleted
1234 1234
    void erase(Row r) {
1235 1235
      _eraseRow(rows(id(r)));
1236 1236
      _eraseRowId(rows(id(r)));
1237 1237
    }
1238 1238

	
1239 1239
    /// Get the name of a column
1240 1240

	
1241 1241
    ///\param c is the coresponding column
1242 1242
    ///\return The name of the colunm
1243 1243
    std::string colName(Col c) const {
1244 1244
      std::string name;
1245 1245
      _getColName(cols(id(c)), name);
1246 1246
      return name;
1247 1247
    }
1248 1248

	
1249 1249
    /// Set the name of a column
1250 1250

	
1251 1251
    ///\param c is the coresponding column
1252 1252
    ///\param name The name to be given
1253 1253
    void colName(Col c, const std::string& name) {
1254 1254
      _setColName(cols(id(c)), name);
1255 1255
    }
1256 1256

	
1257 1257
    /// Get the column by its name
1258 1258

	
1259 1259
    ///\param name The name of the column
1260 1260
    ///\return the proper column or \c INVALID
1261 1261
    Col colByName(const std::string& name) const {
1262 1262
      int k = _colByName(name);
1263 1263
      return k != -1 ? Col(cols[k]) : Col(INVALID);
1264 1264
    }
1265 1265

	
1266 1266
    /// Get the name of a row
1267 1267

	
1268 1268
    ///\param r is the coresponding row
1269 1269
    ///\return The name of the row
1270 1270
    std::string rowName(Row r) const {
1271 1271
      std::string name;
1272 1272
      _getRowName(rows(id(r)), name);
1273 1273
      return name;
1274 1274
    }
1275 1275

	
1276 1276
    /// Set the name of a row
1277 1277

	
1278 1278
    ///\param r is the coresponding row
1279 1279
    ///\param name The name to be given
1280 1280
    void rowName(Row r, const std::string& name) {
1281 1281
      _setRowName(rows(id(r)), name);
1282 1282
    }
1283 1283

	
1284 1284
    /// Get the row by its name
1285 1285

	
1286 1286
    ///\param name The name of the row
1287 1287
    ///\return the proper row or \c INVALID
1288 1288
    Row rowByName(const std::string& name) const {
1289 1289
      int k = _rowByName(name);
1290 1290
      return k != -1 ? Row(rows[k]) : Row(INVALID);
1291 1291
    }
1292 1292

	
1293 1293
    /// Set an element of the coefficient matrix of the LP
1294 1294

	
1295 1295
    ///\param r is the row of the element to be modified
1296 1296
    ///\param c is the column of the element to be modified
1297 1297
    ///\param val is the new value of the coefficient
1298 1298
    void coeff(Row r, Col c, Value val) {
1299 1299
      _setCoeff(rows(id(r)),cols(id(c)), val);
1300 1300
    }
1301 1301

	
1302 1302
    /// Get an element of the coefficient matrix of the LP
1303 1303

	
1304 1304
    ///\param r is the row of the element
1305 1305
    ///\param c is the column of the element
1306 1306
    ///\return the corresponding coefficient
1307 1307
    Value coeff(Row r, Col c) const {
1308 1308
      return _getCoeff(rows(id(r)),cols(id(c)));
1309 1309
    }
1310 1310

	
1311 1311
    /// Set the lower bound of a column (i.e a variable)
1312 1312

	
1313 1313
    /// The lower bound of a variable (column) has to be given by an
1314 1314
    /// extended number of type Value, i.e. a finite number of type
1315 1315
    /// Value or -\ref INF.
1316 1316
    void colLowerBound(Col c, Value value) {
1317 1317
      _setColLowerBound(cols(id(c)),value);
1318 1318
    }
1319 1319

	
1320 1320
    /// Get the lower bound of a column (i.e a variable)
1321 1321

	
1322 1322
    /// This function returns the lower bound for column (variable) \c c
1323 1323
    /// (this might be -\ref INF as well).
1324 1324
    ///\return The lower bound for column \c c
1325 1325
    Value colLowerBound(Col c) const {
1326 1326
      return _getColLowerBound(cols(id(c)));
1327 1327
    }
1328 1328

	
1329 1329
    ///\brief Set the lower bound of  several columns
1330 1330
    ///(i.e variables) at once
1331 1331
    ///
1332 1332
    ///This magic function takes a container as its argument
1333 1333
    ///and applies the function on all of its elements.
1334 1334
    ///The lower bound of a variable (column) has to be given by an
1335 1335
    ///extended number of type Value, i.e. a finite number of type
1336 1336
    ///Value or -\ref INF.
1337 1337
#ifdef DOXYGEN
1338 1338
    template<class T>
1339 1339
    void colLowerBound(T &t, Value value) { return 0;}
1340 1340
#else
1341 1341
    template<class T>
1342 1342
    typename enable_if<typename T::value_type::LpCol,void>::type
1343 1343
    colLowerBound(T &t, Value value,dummy<0> = 0) {
1344 1344
      for(typename T::iterator i=t.begin();i!=t.end();++i) {
1345 1345
        colLowerBound(*i, value);
1346 1346
      }
1347 1347
    }
1348 1348
    template<class T>
1349 1349
    typename enable_if<typename T::value_type::second_type::LpCol,
1350 1350
                       void>::type
1351 1351
    colLowerBound(T &t, Value value,dummy<1> = 1) {
1352 1352
      for(typename T::iterator i=t.begin();i!=t.end();++i) {
1353 1353
        colLowerBound(i->second, value);
1354 1354
      }
1355 1355
    }
1356 1356
    template<class T>
1357 1357
    typename enable_if<typename T::MapIt::Value::LpCol,
1358 1358
                       void>::type
1359 1359
    colLowerBound(T &t, Value value,dummy<2> = 2) {
1360 1360
      for(typename T::MapIt i(t); i!=INVALID; ++i){
1361 1361
        colLowerBound(*i, value);
1362 1362
      }
1363 1363
    }
1364 1364
#endif
1365 1365

	
1366 1366
    /// Set the upper bound of a column (i.e a variable)
1367 1367

	
1368 1368
    /// The upper bound of a variable (column) has to be given by an
1369 1369
    /// extended number of type Value, i.e. a finite number of type
1370 1370
    /// Value or \ref INF.
1371 1371
    void colUpperBound(Col c, Value value) {
1372 1372
      _setColUpperBound(cols(id(c)),value);
1373 1373
    };
1374 1374

	
1375 1375
    /// Get the upper bound of a column (i.e a variable)
1376 1376

	
1377 1377
    /// This function returns the upper bound for column (variable) \c c
1378 1378
    /// (this might be \ref INF as well).
1379 1379
    /// \return The upper bound for column \c c
1380 1380
    Value colUpperBound(Col c) const {
1381 1381
      return _getColUpperBound(cols(id(c)));
1382 1382
    }
1383 1383

	
1384 1384
    ///\brief Set the upper bound of  several columns
1385 1385
    ///(i.e variables) at once
1386 1386
    ///
1387 1387
    ///This magic function takes a container as its argument
1388 1388
    ///and applies the function on all of its elements.
1389 1389
    ///The upper bound of a variable (column) has to be given by an
1390 1390
    ///extended number of type Value, i.e. a finite number of type
1391 1391
    ///Value or \ref INF.
1392 1392
#ifdef DOXYGEN
1393 1393
    template<class T>
1394 1394
    void colUpperBound(T &t, Value value) { return 0;}
1395 1395
#else
1396 1396
    template<class T1>
1397 1397
    typename enable_if<typename T1::value_type::LpCol,void>::type
1398 1398
    colUpperBound(T1 &t, Value value,dummy<0> = 0) {
1399 1399
      for(typename T1::iterator i=t.begin();i!=t.end();++i) {
1400 1400
        colUpperBound(*i, value);
1401 1401
      }
1402 1402
    }
1403 1403
    template<class T1>
1404 1404
    typename enable_if<typename T1::value_type::second_type::LpCol,
1405 1405
                       void>::type
1406 1406
    colUpperBound(T1 &t, Value value,dummy<1> = 1) {
1407 1407
      for(typename T1::iterator i=t.begin();i!=t.end();++i) {
1408 1408
        colUpperBound(i->second, value);
1409 1409
      }
1410 1410
    }
1411 1411
    template<class T1>
1412 1412
    typename enable_if<typename T1::MapIt::Value::LpCol,
1413 1413
                       void>::type
1414 1414
    colUpperBound(T1 &t, Value value,dummy<2> = 2) {
1415 1415
      for(typename T1::MapIt i(t); i!=INVALID; ++i){
1416 1416
        colUpperBound(*i, value);
1417 1417
      }
1418 1418
    }
1419 1419
#endif
1420 1420

	
1421 1421
    /// Set the lower and the upper bounds of a column (i.e a variable)
1422 1422

	
1423 1423
    /// The lower and the upper bounds of
1424 1424
    /// a variable (column) have to be given by an
1425 1425
    /// extended number of type Value, i.e. a finite number of type
1426 1426
    /// Value, -\ref INF or \ref INF.
1427 1427
    void colBounds(Col c, Value lower, Value upper) {
1428 1428
      _setColLowerBound(cols(id(c)),lower);
1429 1429
      _setColUpperBound(cols(id(c)),upper);
1430 1430
    }
1431 1431

	
1432 1432
    ///\brief Set the lower and the upper bound of several columns
1433 1433
    ///(i.e variables) at once
1434 1434
    ///
1435 1435
    ///This magic function takes a container as its argument
1436 1436
    ///and applies the function on all of its elements.
1437 1437
    /// The lower and the upper bounds of
1438 1438
    /// a variable (column) have to be given by an
1439 1439
    /// extended number of type Value, i.e. a finite number of type
1440 1440
    /// Value, -\ref INF or \ref INF.
1441 1441
#ifdef DOXYGEN
1442 1442
    template<class T>
1443 1443
    void colBounds(T &t, Value lower, Value upper) { return 0;}
1444 1444
#else
1445 1445
    template<class T2>
1446 1446
    typename enable_if<typename T2::value_type::LpCol,void>::type
1447 1447
    colBounds(T2 &t, Value lower, Value upper,dummy<0> = 0) {
1448 1448
      for(typename T2::iterator i=t.begin();i!=t.end();++i) {
1449 1449
        colBounds(*i, lower, upper);
1450 1450
      }
1451 1451
    }
1452 1452
    template<class T2>
1453 1453
    typename enable_if<typename T2::value_type::second_type::LpCol, void>::type
1454 1454
    colBounds(T2 &t, Value lower, Value upper,dummy<1> = 1) {
1455 1455
      for(typename T2::iterator i=t.begin();i!=t.end();++i) {
1456 1456
        colBounds(i->second, lower, upper);
1457 1457
      }
1458 1458
    }
1459 1459
    template<class T2>
1460 1460
    typename enable_if<typename T2::MapIt::Value::LpCol, void>::type
1461 1461
    colBounds(T2 &t, Value lower, Value upper,dummy<2> = 2) {
1462 1462
      for(typename T2::MapIt i(t); i!=INVALID; ++i){
1463 1463
        colBounds(*i, lower, upper);
1464 1464
      }
1465 1465
    }
1466 1466
#endif
1467 1467

	
1468 1468
    /// Set the lower bound of a row (i.e a constraint)
1469 1469

	
1470 1470
    /// The lower bound of a constraint (row) has to be given by an
1471 1471
    /// extended number of type Value, i.e. a finite number of type
1472 1472
    /// Value or -\ref INF.
1473 1473
    void rowLowerBound(Row r, Value value) {
1474 1474
      _setRowLowerBound(rows(id(r)),value);
1475 1475
    }
1476 1476

	
1477 1477
    /// Get the lower bound of a row (i.e a constraint)
1478 1478

	
1479 1479
    /// This function returns the lower bound for row (constraint) \c c
1480 1480
    /// (this might be -\ref INF as well).
1481 1481
    ///\return The lower bound for row \c r
1482 1482
    Value rowLowerBound(Row r) const {
1483 1483
      return _getRowLowerBound(rows(id(r)));
1484 1484
    }
1485 1485

	
1486 1486
    /// Set the upper bound of a row (i.e a constraint)
1487 1487

	
1488 1488
    /// The upper bound of a constraint (row) has to be given by an
1489 1489
    /// extended number of type Value, i.e. a finite number of type
1490 1490
    /// Value or -\ref INF.
1491 1491
    void rowUpperBound(Row r, Value value) {
1492 1492
      _setRowUpperBound(rows(id(r)),value);
1493 1493
    }
1494 1494

	
1495 1495
    /// Get the upper bound of a row (i.e a constraint)
1496 1496

	
1497 1497
    /// This function returns the upper bound for row (constraint) \c c
1498 1498
    /// (this might be -\ref INF as well).
1499 1499
    ///\return The upper bound for row \c r
1500 1500
    Value rowUpperBound(Row r) const {
1501 1501
      return _getRowUpperBound(rows(id(r)));
1502 1502
    }
1503 1503

	
1504 1504
    ///Set an element of the objective function
1505 1505
    void objCoeff(Col c, Value v) {_setObjCoeff(cols(id(c)),v); };
1506 1506

	
1507 1507
    ///Get an element of the objective function
1508 1508
    Value objCoeff(Col c) const { return _getObjCoeff(cols(id(c))); };
1509 1509

	
1510 1510
    ///Set the objective function
1511 1511

	
1512 1512
    ///\param e is a linear expression of type \ref Expr.
1513 1513
    ///
1514 1514
    void obj(const Expr& e) {
1515 1515
      _setObjCoeffs(ExprIterator(e.comps.begin(), cols),
1516 1516
                    ExprIterator(e.comps.end(), cols));
1517 1517
      obj_const_comp = *e;
1518 1518
    }
1519 1519

	
1520 1520
    ///Get the objective function
1521 1521

	
1522 1522
    ///\return the objective function as a linear expression of type
1523 1523
    ///Expr.
1524 1524
    Expr obj() const {
1525 1525
      Expr e;
1526 1526
      _getObjCoeffs(InsertIterator(e.comps, cols));
1527 1527
      *e = obj_const_comp;
1528 1528
      return e;
1529 1529
    }
1530 1530

	
1531 1531

	
1532 1532
    ///Set the direction of optimization
1533 1533
    void sense(Sense sense) { _setSense(sense); }
1534 1534

	
1535 1535
    ///Query the direction of the optimization
1536 1536
    Sense sense() const {return _getSense(); }
1537 1537

	
1538 1538
    ///Set the sense to maximization
1539 1539
    void max() { _setSense(MAX); }
1540 1540

	
1541 1541
    ///Set the sense to maximization
1542 1542
    void min() { _setSense(MIN); }
1543 1543

	
1544 1544
    ///Clears the problem
1545 1545
    void clear() { _clear(); }
1546 1546

	
1547 1547
    /// Sets the message level of the solver
1548 1548
    void messageLevel(MessageLevel level) { _messageLevel(level); }
1549 1549

	
1550 1550
    ///@}
1551 1551

	
1552 1552
  };
1553 1553

	
1554 1554
  /// Addition
1555 1555

	
1556 1556
  ///\relates LpBase::Expr
1557 1557
  ///
1558 1558
  inline LpBase::Expr operator+(const LpBase::Expr &a, const LpBase::Expr &b) {
1559 1559
    LpBase::Expr tmp(a);
1560 1560
    tmp+=b;
1561 1561
    return tmp;
1562 1562
  }
1563 1563
  ///Substraction
1564 1564

	
1565 1565
  ///\relates LpBase::Expr
1566 1566
  ///
1567 1567
  inline LpBase::Expr operator-(const LpBase::Expr &a, const LpBase::Expr &b) {
1568 1568
    LpBase::Expr tmp(a);
1569 1569
    tmp-=b;
1570 1570
    return tmp;
1571 1571
  }
1572 1572
  ///Multiply with constant
1573 1573

	
1574 1574
  ///\relates LpBase::Expr
1575 1575
  ///
1576 1576
  inline LpBase::Expr operator*(const LpBase::Expr &a, const LpBase::Value &b) {
1577 1577
    LpBase::Expr tmp(a);
1578 1578
    tmp*=b;
1579 1579
    return tmp;
1580 1580
  }
1581 1581

	
1582 1582
  ///Multiply with constant
1583 1583

	
1584 1584
  ///\relates LpBase::Expr
1585 1585
  ///
1586 1586
  inline LpBase::Expr operator*(const LpBase::Value &a, const LpBase::Expr &b) {
1587 1587
    LpBase::Expr tmp(b);
1588 1588
    tmp*=a;
1589 1589
    return tmp;
1590 1590
  }
1591 1591
  ///Divide with constant
1592 1592

	
1593 1593
  ///\relates LpBase::Expr
1594 1594
  ///
1595 1595
  inline LpBase::Expr operator/(const LpBase::Expr &a, const LpBase::Value &b) {
1596 1596
    LpBase::Expr tmp(a);
1597 1597
    tmp/=b;
1598 1598
    return tmp;
1599 1599
  }
1600 1600

	
1601 1601
  ///Create constraint
1602 1602

	
1603 1603
  ///\relates LpBase::Constr
1604 1604
  ///
1605 1605
  inline LpBase::Constr operator<=(const LpBase::Expr &e,
1606 1606
                                   const LpBase::Expr &f) {
1607
    return LpBase::Constr(0, f - e, LpBase::INF);
1607
    return LpBase::Constr(0, f - e, LpBase::NaN);
1608 1608
  }
1609 1609

	
1610 1610
  ///Create constraint
1611 1611

	
1612 1612
  ///\relates LpBase::Constr
1613 1613
  ///
1614 1614
  inline LpBase::Constr operator<=(const LpBase::Value &e,
1615 1615
                                   const LpBase::Expr &f) {
1616 1616
    return LpBase::Constr(e, f, LpBase::NaN);
1617 1617
  }
1618 1618

	
1619 1619
  ///Create constraint
1620 1620

	
1621 1621
  ///\relates LpBase::Constr
1622 1622
  ///
1623 1623
  inline LpBase::Constr operator<=(const LpBase::Expr &e,
1624 1624
                                   const LpBase::Value &f) {
1625
    return LpBase::Constr(- LpBase::INF, e, f);
1625
    return LpBase::Constr(LpBase::NaN, e, f);
1626 1626
  }
1627 1627

	
1628 1628
  ///Create constraint
1629 1629

	
1630 1630
  ///\relates LpBase::Constr
1631 1631
  ///
1632 1632
  inline LpBase::Constr operator>=(const LpBase::Expr &e,
1633 1633
                                   const LpBase::Expr &f) {
1634
    return LpBase::Constr(0, e - f, LpBase::INF);
1634
    return LpBase::Constr(0, e - f, LpBase::NaN);
1635 1635
  }
1636 1636

	
1637 1637

	
1638 1638
  ///Create constraint
1639 1639

	
1640 1640
  ///\relates LpBase::Constr
1641 1641
  ///
1642 1642
  inline LpBase::Constr operator>=(const LpBase::Value &e,
1643 1643
                                   const LpBase::Expr &f) {
1644 1644
    return LpBase::Constr(LpBase::NaN, f, e);
1645 1645
  }
1646 1646

	
1647 1647

	
1648 1648
  ///Create constraint
1649 1649

	
1650 1650
  ///\relates LpBase::Constr
1651 1651
  ///
1652 1652
  inline LpBase::Constr operator>=(const LpBase::Expr &e,
1653 1653
                                   const LpBase::Value &f) {
1654
    return LpBase::Constr(f, e, LpBase::INF);
1654
    return LpBase::Constr(f, e, LpBase::NaN);
1655 1655
  }
1656 1656

	
1657 1657
  ///Create constraint
1658 1658

	
1659 1659
  ///\relates LpBase::Constr
1660 1660
  ///
1661 1661
  inline LpBase::Constr operator==(const LpBase::Expr &e,
1662 1662
                                   const LpBase::Value &f) {
1663 1663
    return LpBase::Constr(f, e, f);
1664 1664
  }
1665 1665

	
1666 1666
  ///Create constraint
1667 1667

	
1668 1668
  ///\relates LpBase::Constr
1669 1669
  ///
1670 1670
  inline LpBase::Constr operator==(const LpBase::Expr &e,
1671 1671
                                   const LpBase::Expr &f) {
1672 1672
    return LpBase::Constr(0, f - e, 0);
1673 1673
  }
1674 1674

	
1675 1675
  ///Create constraint
1676 1676

	
1677 1677
  ///\relates LpBase::Constr
1678 1678
  ///
1679 1679
  inline LpBase::Constr operator<=(const LpBase::Value &n,
1680 1680
                                   const LpBase::Constr &c) {
1681 1681
    LpBase::Constr tmp(c);
1682 1682
    LEMON_ASSERT(isNaN(tmp.lowerBound()), "Wrong LP constraint");
1683 1683
    tmp.lowerBound()=n;
1684 1684
    return tmp;
1685 1685
  }
1686 1686
  ///Create constraint
1687 1687

	
1688 1688
  ///\relates LpBase::Constr
1689 1689
  ///
1690 1690
  inline LpBase::Constr operator<=(const LpBase::Constr &c,
1691 1691
                                   const LpBase::Value &n)
1692 1692
  {
1693 1693
    LpBase::Constr tmp(c);
1694 1694
    LEMON_ASSERT(isNaN(tmp.upperBound()), "Wrong LP constraint");
1695 1695
    tmp.upperBound()=n;
1696 1696
    return tmp;
1697 1697
  }
1698 1698

	
1699 1699
  ///Create constraint
1700 1700

	
1701 1701
  ///\relates LpBase::Constr
1702 1702
  ///
1703 1703
  inline LpBase::Constr operator>=(const LpBase::Value &n,
1704 1704
                                   const LpBase::Constr &c) {
1705 1705
    LpBase::Constr tmp(c);
1706 1706
    LEMON_ASSERT(isNaN(tmp.upperBound()), "Wrong LP constraint");
1707 1707
    tmp.upperBound()=n;
1708 1708
    return tmp;
1709 1709
  }
1710 1710
  ///Create constraint
1711 1711

	
1712 1712
  ///\relates LpBase::Constr
1713 1713
  ///
1714 1714
  inline LpBase::Constr operator>=(const LpBase::Constr &c,
1715 1715
                                   const LpBase::Value &n)
1716 1716
  {
1717 1717
    LpBase::Constr tmp(c);
1718 1718
    LEMON_ASSERT(isNaN(tmp.lowerBound()), "Wrong LP constraint");
1719 1719
    tmp.lowerBound()=n;
1720 1720
    return tmp;
1721 1721
  }
1722 1722

	
1723 1723
  ///Addition
1724 1724

	
1725 1725
  ///\relates LpBase::DualExpr
1726 1726
  ///
1727 1727
  inline LpBase::DualExpr operator+(const LpBase::DualExpr &a,
1728 1728
                                    const LpBase::DualExpr &b) {
1729 1729
    LpBase::DualExpr tmp(a);
1730 1730
    tmp+=b;
1731 1731
    return tmp;
1732 1732
  }
1733 1733
  ///Substraction
1734 1734

	
1735 1735
  ///\relates LpBase::DualExpr
1736 1736
  ///
1737 1737
  inline LpBase::DualExpr operator-(const LpBase::DualExpr &a,
1738 1738
                                    const LpBase::DualExpr &b) {
1739 1739
    LpBase::DualExpr tmp(a);
1740 1740
    tmp-=b;
1741 1741
    return tmp;
1742 1742
  }
1743 1743
  ///Multiply with constant
1744 1744

	
1745 1745
  ///\relates LpBase::DualExpr
1746 1746
  ///
1747 1747
  inline LpBase::DualExpr operator*(const LpBase::DualExpr &a,
1748 1748
                                    const LpBase::Value &b) {
1749 1749
    LpBase::DualExpr tmp(a);
1750 1750
    tmp*=b;
1751 1751
    return tmp;
1752 1752
  }
1753 1753

	
1754 1754
  ///Multiply with constant
1755 1755

	
1756 1756
  ///\relates LpBase::DualExpr
1757 1757
  ///
1758 1758
  inline LpBase::DualExpr operator*(const LpBase::Value &a,
1759 1759
                                    const LpBase::DualExpr &b) {
1760 1760
    LpBase::DualExpr tmp(b);
1761 1761
    tmp*=a;
1762 1762
    return tmp;
1763 1763
  }
1764 1764
  ///Divide with constant
1765 1765

	
1766 1766
  ///\relates LpBase::DualExpr
1767 1767
  ///
1768 1768
  inline LpBase::DualExpr operator/(const LpBase::DualExpr &a,
1769 1769
                                    const LpBase::Value &b) {
1770 1770
    LpBase::DualExpr tmp(a);
1771 1771
    tmp/=b;
1772 1772
    return tmp;
1773 1773
  }
1774 1774

	
1775 1775
  /// \ingroup lp_group
1776 1776
  ///
1777 1777
  /// \brief Common base class for LP solvers
1778 1778
  ///
1779 1779
  /// This class is an abstract base class for LP solvers. This class
1780 1780
  /// provides a full interface for set and modify an LP problem,
1781 1781
  /// solve it and retrieve the solution. You can use one of the
1782 1782
  /// descendants as a concrete implementation, or the \c Lp
1783 1783
  /// default LP solver. However, if you would like to handle LP
1784 1784
  /// solvers as reference or pointer in a generic way, you can use
1785 1785
  /// this class directly.
1786 1786
  class LpSolver : virtual public LpBase {
1787 1787
  public:
1788 1788

	
1789 1789
    /// The problem types for primal and dual problems
1790 1790
    enum ProblemType {
1791 1791
      /// = 0. Feasible solution hasn't been found (but may exist).
1792 1792
      UNDEFINED = 0,
1793 1793
      /// = 1. The problem has no feasible solution.
1794 1794
      INFEASIBLE = 1,
1795 1795
      /// = 2. Feasible solution found.
1796 1796
      FEASIBLE = 2,
1797 1797
      /// = 3. Optimal solution exists and found.
1798 1798
      OPTIMAL = 3,
1799 1799
      /// = 4. The cost function is unbounded.
1800 1800
      UNBOUNDED = 4
1801 1801
    };
1802 1802

	
1803 1803
    ///The basis status of variables
1804 1804
    enum VarStatus {
1805 1805
      /// The variable is in the basis
1806 1806
      BASIC, 
1807 1807
      /// The variable is free, but not basic
1808 1808
      FREE,
1809 1809
      /// The variable has active lower bound 
1810 1810
      LOWER,
1811 1811
      /// The variable has active upper bound
1812 1812
      UPPER,
1813 1813
      /// The variable is non-basic and fixed
1814 1814
      FIXED
1815 1815
    };
1816 1816

	
1817 1817
  protected:
1818 1818

	
1819 1819
    virtual SolveExitStatus _solve() = 0;
1820 1820

	
1821 1821
    virtual Value _getPrimal(int i) const = 0;
1822 1822
    virtual Value _getDual(int i) const = 0;
1823 1823

	
1824 1824
    virtual Value _getPrimalRay(int i) const = 0;
1825 1825
    virtual Value _getDualRay(int i) const = 0;
1826 1826

	
1827 1827
    virtual Value _getPrimalValue() const = 0;
1828 1828

	
1829 1829
    virtual VarStatus _getColStatus(int i) const = 0;
1830 1830
    virtual VarStatus _getRowStatus(int i) const = 0;
1831 1831

	
1832 1832
    virtual ProblemType _getPrimalType() const = 0;
1833 1833
    virtual ProblemType _getDualType() const = 0;
1834 1834

	
1835 1835
  public:
1836 1836

	
1837 1837
    ///Allocate a new LP problem instance
1838 1838
    virtual LpSolver* newSolver() const = 0;
1839 1839
    ///Make a copy of the LP problem
1840 1840
    virtual LpSolver* cloneSolver() const = 0;
1841 1841

	
1842 1842
    ///\name Solve the LP
1843 1843

	
1844 1844
    ///@{
1845 1845

	
1846 1846
    ///\e Solve the LP problem at hand
1847 1847
    ///
1848 1848
    ///\return The result of the optimization procedure. Possible
1849 1849
    ///values and their meanings can be found in the documentation of
1850 1850
    ///\ref SolveExitStatus.
1851 1851
    SolveExitStatus solve() { return _solve(); }
1852 1852

	
1853 1853
    ///@}
1854 1854

	
1855 1855
    ///\name Obtain the Solution
1856 1856

	
1857 1857
    ///@{
1858 1858

	
1859 1859
    /// The type of the primal problem
1860 1860
    ProblemType primalType() const {
1861 1861
      return _getPrimalType();
1862 1862
    }
1863 1863

	
1864 1864
    /// The type of the dual problem
1865 1865
    ProblemType dualType() const {
1866 1866
      return _getDualType();
1867 1867
    }
1868 1868

	
1869 1869
    /// Return the primal value of the column
1870 1870

	
1871 1871
    /// Return the primal value of the column.
1872 1872
    /// \pre The problem is solved.
1873 1873
    Value primal(Col c) const { return _getPrimal(cols(id(c))); }
1874 1874

	
1875 1875
    /// Return the primal value of the expression
1876 1876

	
1877 1877
    /// Return the primal value of the expression, i.e. the dot
1878 1878
    /// product of the primal solution and the expression.
1879 1879
    /// \pre The problem is solved.
1880 1880
    Value primal(const Expr& e) const {
1881 1881
      double res = *e;
1882 1882
      for (Expr::ConstCoeffIt c(e); c != INVALID; ++c) {
1883 1883
        res += *c * primal(c);
1884 1884
      }
1885 1885
      return res;
1886 1886
    }
1887 1887
    /// Returns a component of the primal ray
1888 1888
    
1889 1889
    /// The primal ray is solution of the modified primal problem,
1890 1890
    /// where we change each finite bound to 0, and we looking for a
1891 1891
    /// negative objective value in case of minimization, and positive
1892 1892
    /// objective value for maximization. If there is such solution,
1893 1893
    /// that proofs the unsolvability of the dual problem, and if a
1894 1894
    /// feasible primal solution exists, then the unboundness of
1895 1895
    /// primal problem.
1896 1896
    ///
1897 1897
    /// \pre The problem is solved and the dual problem is infeasible.
1898 1898
    /// \note Some solvers does not provide primal ray calculation
1899 1899
    /// functions.
1900 1900
    Value primalRay(Col c) const { return _getPrimalRay(cols(id(c))); }
1901 1901

	
1902 1902
    /// Return the dual value of the row
1903 1903

	
1904 1904
    /// Return the dual value of the row.
1905 1905
    /// \pre The problem is solved.
1906 1906
    Value dual(Row r) const { return _getDual(rows(id(r))); }
1907 1907

	
1908 1908
    /// Return the dual value of the dual expression
1909 1909

	
1910 1910
    /// Return the dual value of the dual expression, i.e. the dot
1911 1911
    /// product of the dual solution and the dual expression.
1912 1912
    /// \pre The problem is solved.
1913 1913
    Value dual(const DualExpr& e) const {
1914 1914
      double res = 0.0;
1915 1915
      for (DualExpr::ConstCoeffIt r(e); r != INVALID; ++r) {
1916 1916
        res += *r * dual(r);
1917 1917
      }
1918 1918
      return res;
1919 1919
    }
1920 1920

	
1921 1921
    /// Returns a component of the dual ray
1922 1922
    
1923 1923
    /// The dual ray is solution of the modified primal problem, where
1924 1924
    /// we change each finite bound to 0 (i.e. the objective function
1925 1925
    /// coefficients in the primal problem), and we looking for a
1926 1926
    /// ositive objective value. If there is such solution, that
1927 1927
    /// proofs the unsolvability of the primal problem, and if a
1928 1928
    /// feasible dual solution exists, then the unboundness of
1929 1929
    /// dual problem.
1930 1930
    ///
1931 1931
    /// \pre The problem is solved and the primal problem is infeasible.
1932 1932
    /// \note Some solvers does not provide dual ray calculation
1933 1933
    /// functions.
1934 1934
    Value dualRay(Row r) const { return _getDualRay(rows(id(r))); }
1935 1935

	
1936 1936
    /// Return the basis status of the column
1937 1937

	
1938 1938
    /// \see VarStatus
1939 1939
    VarStatus colStatus(Col c) const { return _getColStatus(cols(id(c))); }
1940 1940

	
1941 1941
    /// Return the basis status of the row
1942 1942

	
1943 1943
    /// \see VarStatus
1944 1944
    VarStatus rowStatus(Row r) const { return _getRowStatus(rows(id(r))); }
1945 1945

	
1946 1946
    ///The value of the objective function
1947 1947

	
1948 1948
    ///\return
1949 1949
    ///- \ref INF or -\ref INF means either infeasibility or unboundedness
1950 1950
    /// of the primal problem, depending on whether we minimize or maximize.
1951 1951
    ///- \ref NaN if no primal solution is found.
1952 1952
    ///- The (finite) objective value if an optimal solution is found.
1953 1953
    Value primal() const { return _getPrimalValue()+obj_const_comp;}
1954 1954
    ///@}
1955 1955

	
1956 1956
  protected:
1957 1957

	
1958 1958
  };
1959 1959

	
1960 1960

	
1961 1961
  /// \ingroup lp_group
1962 1962
  ///
1963 1963
  /// \brief Common base class for MIP solvers
1964 1964
  ///
1965 1965
  /// This class is an abstract base class for MIP solvers. This class
1966 1966
  /// provides a full interface for set and modify an MIP problem,
1967 1967
  /// solve it and retrieve the solution. You can use one of the
1968 1968
  /// descendants as a concrete implementation, or the \c Lp
1969 1969
  /// default MIP solver. However, if you would like to handle MIP
1970 1970
  /// solvers as reference or pointer in a generic way, you can use
1971 1971
  /// this class directly.
1972 1972
  class MipSolver : virtual public LpBase {
1973 1973
  public:
1974 1974

	
1975 1975
    /// The problem types for MIP problems
1976 1976
    enum ProblemType {
1977 1977
      /// = 0. Feasible solution hasn't been found (but may exist).
1978 1978
      UNDEFINED = 0,
1979 1979
      /// = 1. The problem has no feasible solution.
1980 1980
      INFEASIBLE = 1,
1981 1981
      /// = 2. Feasible solution found.
1982 1982
      FEASIBLE = 2,
1983 1983
      /// = 3. Optimal solution exists and found.
1984 1984
      OPTIMAL = 3,
1985 1985
      /// = 4. The cost function is unbounded.
1986 1986
      ///The Mip or at least the relaxed problem is unbounded.
1987 1987
      UNBOUNDED = 4
1988 1988
    };
1989 1989

	
1990 1990
    ///Allocate a new MIP problem instance
1991 1991
    virtual MipSolver* newSolver() const = 0;
1992 1992
    ///Make a copy of the MIP problem
1993 1993
    virtual MipSolver* cloneSolver() const = 0;
1994 1994

	
1995 1995
    ///\name Solve the MIP
1996 1996

	
1997 1997
    ///@{
1998 1998

	
1999 1999
    /// Solve the MIP problem at hand
2000 2000
    ///
2001 2001
    ///\return The result of the optimization procedure. Possible
2002 2002
    ///values and their meanings can be found in the documentation of
2003 2003
    ///\ref SolveExitStatus.
2004 2004
    SolveExitStatus solve() { return _solve(); }
2005 2005

	
2006 2006
    ///@}
2007 2007

	
2008 2008
    ///\name Set Column Type
2009 2009
    ///@{
2010 2010

	
2011 2011
    ///Possible variable (column) types (e.g. real, integer, binary etc.)
2012 2012
    enum ColTypes {
2013 2013
      /// = 0. Continuous variable (default).
2014 2014
      REAL = 0,
2015 2015
      /// = 1. Integer variable.
2016 2016
      INTEGER = 1
2017 2017
    };
2018 2018

	
2019 2019
    ///Sets the type of the given column to the given type
2020 2020

	
2021 2021
    ///Sets the type of the given column to the given type.
2022 2022
    ///
2023 2023
    void colType(Col c, ColTypes col_type) {
2024 2024
      _setColType(cols(id(c)),col_type);
2025 2025
    }
2026 2026

	
2027 2027
    ///Gives back the type of the column.
2028 2028

	
2029 2029
    ///Gives back the type of the column.
2030 2030
    ///
2031 2031
    ColTypes colType(Col c) const {
2032 2032
      return _getColType(cols(id(c)));
2033 2033
    }
2034 2034
    ///@}
2035 2035

	
2036 2036
    ///\name Obtain the Solution
2037 2037

	
2038 2038
    ///@{
2039 2039

	
2040 2040
    /// The type of the MIP problem
2041 2041
    ProblemType type() const {
2042 2042
      return _getType();
2043 2043
    }
2044 2044

	
2045 2045
    /// Return the value of the row in the solution
2046 2046

	
2047 2047
    ///  Return the value of the row in the solution.
2048 2048
    /// \pre The problem is solved.
2049 2049
    Value sol(Col c) const { return _getSol(cols(id(c))); }
2050 2050

	
2051 2051
    /// Return the value of the expression in the solution
2052 2052

	
2053 2053
    /// Return the value of the expression in the solution, i.e. the
2054 2054
    /// dot product of the solution and the expression.
2055 2055
    /// \pre The problem is solved.
2056 2056
    Value sol(const Expr& e) const {
2057 2057
      double res = *e;
2058 2058
      for (Expr::ConstCoeffIt c(e); c != INVALID; ++c) {
2059 2059
        res += *c * sol(c);
2060 2060
      }
2061 2061
      return res;
2062 2062
    }
2063 2063
    ///The value of the objective function
2064 2064
    
2065 2065
    ///\return
2066 2066
    ///- \ref INF or -\ref INF means either infeasibility or unboundedness
2067 2067
    /// of the problem, depending on whether we minimize or maximize.
2068 2068
    ///- \ref NaN if no primal solution is found.
2069 2069
    ///- The (finite) objective value if an optimal solution is found.
2070 2070
    Value solValue() const { return _getSolValue()+obj_const_comp;}
2071 2071
    ///@}
2072 2072

	
2073 2073
  protected:
2074 2074

	
2075 2075
    virtual SolveExitStatus _solve() = 0;
2076 2076
    virtual ColTypes _getColType(int col) const = 0;
2077 2077
    virtual void _setColType(int col, ColTypes col_type) = 0;
2078 2078
    virtual ProblemType _getType() const = 0;
2079 2079
    virtual Value _getSol(int i) const = 0;
2080 2080
    virtual Value _getSolValue() const = 0;
2081 2081

	
2082 2082
  };
2083 2083

	
2084 2084

	
2085 2085

	
2086 2086
} //namespace lemon
2087 2087

	
2088 2088
#endif //LEMON_LP_BASE_H
Ignore white space 1024 line context
1 1
/* -*- mode: C++; indent-tabs-mode: nil; -*-
2 2
 *
3 3
 * This file is a part of LEMON, a generic C++ optimization library.
4 4
 *
5 5
 * Copyright (C) 2003-2009
6 6
 * Egervary Jeno Kombinatorikus Optimalizalasi Kutatocsoport
7 7
 * (Egervary Research Group on Combinatorial Optimization, EGRES).
8 8
 *
9 9
 * Permission to use, modify and distribute this software is granted
10 10
 * provided that this copyright notice appears in all copies. For
11 11
 * precise terms see the accompanying LICENSE file.
12 12
 *
13 13
 * This software is provided "AS IS" with no warranty of any kind,
14 14
 * express or implied, and with no claim as to its suitability for any
15 15
 * purpose.
16 16
 *
17 17
 */
18 18

	
19 19
#include <sstream>
20 20
#include <lemon/lp_skeleton.h>
21 21
#include "test_tools.h"
22 22
#include <lemon/tolerance.h>
23 23

	
24 24
#include <lemon/config.h>
25 25

	
26 26
#ifdef LEMON_HAVE_GLPK
27 27
#include <lemon/glpk.h>
28 28
#endif
29 29

	
30 30
#ifdef LEMON_HAVE_CPLEX
31 31
#include <lemon/cplex.h>
32 32
#endif
33 33

	
34 34
#ifdef LEMON_HAVE_SOPLEX
35 35
#include <lemon/soplex.h>
36 36
#endif
37 37

	
38 38
#ifdef LEMON_HAVE_CLP
39 39
#include <lemon/clp.h>
40 40
#endif
41 41

	
42 42
using namespace lemon;
43 43

	
44 44
void lpTest(LpSolver& lp)
45 45
{
46 46

	
47 47
  typedef LpSolver LP;
48 48

	
49 49
  std::vector<LP::Col> x(10);
50 50
  //  for(int i=0;i<10;i++) x.push_back(lp.addCol());
51 51
  lp.addColSet(x);
52 52
  lp.colLowerBound(x,1);
53 53
  lp.colUpperBound(x,1);
54 54
  lp.colBounds(x,1,2);
55 55

	
56 56
  std::vector<LP::Col> y(10);
57 57
  lp.addColSet(y);
58 58

	
59 59
  lp.colLowerBound(y,1);
60 60
  lp.colUpperBound(y,1);
61 61
  lp.colBounds(y,1,2);
62 62

	
63 63
  std::map<int,LP::Col> z;
64 64

	
65 65
  z.insert(std::make_pair(12,INVALID));
66 66
  z.insert(std::make_pair(2,INVALID));
67 67
  z.insert(std::make_pair(7,INVALID));
68 68
  z.insert(std::make_pair(5,INVALID));
69 69

	
70 70
  lp.addColSet(z);
71 71

	
72 72
  lp.colLowerBound(z,1);
73 73
  lp.colUpperBound(z,1);
74 74
  lp.colBounds(z,1,2);
75 75

	
76 76
  {
77 77
    LP::Expr e,f,g;
78 78
    LP::Col p1,p2,p3,p4,p5;
79 79
    LP::Constr c;
80 80

	
81 81
    p1=lp.addCol();
82 82
    p2=lp.addCol();
83 83
    p3=lp.addCol();
84 84
    p4=lp.addCol();
85 85
    p5=lp.addCol();
86 86

	
87 87
    e[p1]=2;
88 88
    *e=12;
89 89
    e[p1]+=2;
90 90
    *e+=12;
91 91
    e[p1]-=2;
92 92
    *e-=12;
93 93

	
94 94
    e=2;
95 95
    e=2.2;
96 96
    e=p1;
97 97
    e=f;
98 98

	
99 99
    e+=2;
100 100
    e+=2.2;
101 101
    e+=p1;
102 102
    e+=f;
103 103

	
104 104
    e-=2;
105 105
    e-=2.2;
106 106
    e-=p1;
107 107
    e-=f;
108 108

	
109 109
    e*=2;
110 110
    e*=2.2;
111 111
    e/=2;
112 112
    e/=2.2;
113 113

	
114 114
    e=((p1+p2)+(p1-p2)+(p1+12)+(12+p1)+(p1-12)+(12-p1)+
115 115
       (f+12)+(12+f)+(p1+f)+(f+p1)+(f+g)+
116 116
       (f-12)+(12-f)+(p1-f)+(f-p1)+(f-g)+
117 117
       2.2*f+f*2.2+f/2.2+
118 118
       2*f+f*2+f/2+
119 119
       2.2*p1+p1*2.2+p1/2.2+
120 120
       2*p1+p1*2+p1/2
121 121
       );
122 122

	
123 123

	
124 124
    c = (e  <= f  );
125 125
    c = (e  <= 2.2);
126 126
    c = (e  <= 2  );
127 127
    c = (e  <= p1 );
128 128
    c = (2.2<= f  );
129 129
    c = (2  <= f  );
130 130
    c = (p1 <= f  );
131 131
    c = (p1 <= p2 );
132 132
    c = (p1 <= 2.2);
133 133
    c = (p1 <= 2  );
134 134
    c = (2.2<= p2 );
135 135
    c = (2  <= p2 );
136 136

	
137 137
    c = (e  >= f  );
138 138
    c = (e  >= 2.2);
139 139
    c = (e  >= 2  );
140 140
    c = (e  >= p1 );
141 141
    c = (2.2>= f  );
142 142
    c = (2  >= f  );
143 143
    c = (p1 >= f  );
144 144
    c = (p1 >= p2 );
145 145
    c = (p1 >= 2.2);
146 146
    c = (p1 >= 2  );
147 147
    c = (2.2>= p2 );
148 148
    c = (2  >= p2 );
149 149

	
150 150
    c = (e  == f  );
151 151
    c = (e  == 2.2);
152 152
    c = (e  == 2  );
153 153
    c = (e  == p1 );
154 154
    c = (2.2== f  );
155 155
    c = (2  == f  );
156 156
    c = (p1 == f  );
157 157
    //c = (p1 == p2 );
158 158
    c = (p1 == 2.2);
159 159
    c = (p1 == 2  );
160 160
    c = (2.2== p2 );
161 161
    c = (2  == p2 );
162 162

	
163 163
    c = ((2 <= e) <= 3);
164 164
    c = ((2 <= p1) <= 3);
165 165

	
166 166
    c = ((2 >= e) >= 3);
167 167
    c = ((2 >= p1) >= 3);
168 168

	
169
    { //Tests for #430
170
      LP::Col v=lp.addCol();
171
      LP::Constr c = v >= -3;
172
      c = c <= 4;
173
      LP::Constr c2;
174
      c2 = -3 <= v <= 4;
175
    }
176

	
169 177
    e[x[3]]=2;
170 178
    e[x[3]]=4;
171 179
    e[x[3]]=1;
172 180
    *e=12;
173 181

	
174 182
    lp.addRow(-LP::INF,e,23);
175 183
    lp.addRow(-LP::INF,3.0*(x[1]+x[2]/2)-x[3],23);
176 184
    lp.addRow(-LP::INF,3.0*(x[1]+x[2]*2-5*x[3]+12-x[4]/3)+2*x[4]-4,23);
177 185

	
178 186
    lp.addRow(x[1]+x[3]<=x[5]-3);
179 187
    lp.addRow((-7<=x[1]+x[3]-12)<=3);
180 188
    lp.addRow(x[1]<=x[5]);
181 189

	
182 190
    std::ostringstream buf;
183 191

	
184 192

	
185 193
    e=((p1+p2)+(p1-0.99*p2));
186 194
    //e.prettyPrint(std::cout);
187 195
    //(e<=2).prettyPrint(std::cout);
188 196
    double tolerance=0.001;
189 197
    e.simplify(tolerance);
190 198
    buf << "Coeff. of p2 should be 0.01";
191 199
    check(e[p2]>0, buf.str());
192 200

	
193 201
    tolerance=0.02;
194 202
    e.simplify(tolerance);
195 203
    buf << "Coeff. of p2 should be 0";
196 204
    check(const_cast<const LpSolver::Expr&>(e)[p2]==0, buf.str());
197 205

	
198 206
    //Test for clone/new
199 207
    LP* lpnew = lp.newSolver();
200 208
    LP* lpclone = lp.cloneSolver();
201 209
    delete lpnew;
202 210
    delete lpclone;
203 211

	
204 212
  }
205 213

	
206 214
  {
207 215
    LP::DualExpr e,f,g;
208 216
    LP::Row p1 = INVALID, p2 = INVALID, p3 = INVALID,
209 217
      p4 = INVALID, p5 = INVALID;
210 218

	
211 219
    e[p1]=2;
212 220
    e[p1]+=2;
213 221
    e[p1]-=2;
214 222

	
215 223
    e=p1;
216 224
    e=f;
217 225

	
218 226
    e+=p1;
219 227
    e+=f;
220 228

	
221 229
    e-=p1;
222 230
    e-=f;
223 231

	
224 232
    e*=2;
225 233
    e*=2.2;
226 234
    e/=2;
227 235
    e/=2.2;
228 236

	
229 237
    e=((p1+p2)+(p1-p2)+
230 238
       (p1+f)+(f+p1)+(f+g)+
231 239
       (p1-f)+(f-p1)+(f-g)+
232 240
       2.2*f+f*2.2+f/2.2+
233 241
       2*f+f*2+f/2+
234 242
       2.2*p1+p1*2.2+p1/2.2+
235 243
       2*p1+p1*2+p1/2
236 244
       );
237 245
  }
238 246

	
239 247
}
240 248

	
241 249
void solveAndCheck(LpSolver& lp, LpSolver::ProblemType stat,
242 250
                   double exp_opt) {
243 251
  using std::string;
244 252
  lp.solve();
245 253

	
246 254
  std::ostringstream buf;
247 255
  buf << "PrimalType should be: " << int(stat) << int(lp.primalType());
248 256

	
249 257
  check(lp.primalType()==stat, buf.str());
250 258

	
251 259
  if (stat ==  LpSolver::OPTIMAL) {
252 260
    std::ostringstream sbuf;
253 261
    sbuf << "Wrong optimal value (" << lp.primal() <<") with "
254 262
         << lp.solverName() <<"\n     the right optimum is " << exp_opt;
255 263
    check(std::abs(lp.primal()-exp_opt) < 1e-3, sbuf.str());
256 264
  }
257 265
}
258 266

	
259 267
void aTest(LpSolver & lp)
260 268
{
261 269
  typedef LpSolver LP;
262 270

	
263 271
 //The following example is very simple
264 272

	
265 273
  typedef LpSolver::Row Row;
266 274
  typedef LpSolver::Col Col;
267 275

	
268 276

	
269 277
  Col x1 = lp.addCol();
270 278
  Col x2 = lp.addCol();
271 279

	
272 280

	
273 281
  //Constraints
274 282
  Row upright=lp.addRow(x1+2*x2 <=1);
275 283
  lp.addRow(x1+x2 >=-1);
276 284
  lp.addRow(x1-x2 <=1);
277 285
  lp.addRow(x1-x2 >=-1);
278 286
  //Nonnegativity of the variables
279 287
  lp.colLowerBound(x1, 0);
280 288
  lp.colLowerBound(x2, 0);
281 289
  //Objective function
282 290
  lp.obj(x1+x2);
283 291

	
284 292
  lp.sense(lp.MAX);
285 293

	
286 294
  //Testing the problem retrieving routines
287 295
  check(lp.objCoeff(x1)==1,"First term should be 1 in the obj function!");
288 296
  check(lp.sense() == lp.MAX,"This is a maximization!");
289 297
  check(lp.coeff(upright,x1)==1,"The coefficient in question is 1!");
290 298
  check(lp.colLowerBound(x1)==0,
291 299
        "The lower bound for variable x1 should be 0.");
292 300
  check(lp.colUpperBound(x1)==LpSolver::INF,
293 301
        "The upper bound for variable x1 should be infty.");
294 302
  check(lp.rowLowerBound(upright) == -LpSolver::INF,
295 303
        "The lower bound for the first row should be -infty.");
296 304
  check(lp.rowUpperBound(upright)==1,
297 305
        "The upper bound for the first row should be 1.");
298 306
  LpSolver::Expr e = lp.row(upright);
299 307
  check(e[x1] == 1, "The first coefficient should 1.");
300 308
  check(e[x2] == 2, "The second coefficient should 1.");
301 309

	
302 310
  lp.row(upright, x1+x2 <=1);
303 311
  e = lp.row(upright);
304 312
  check(e[x1] == 1, "The first coefficient should 1.");
305 313
  check(e[x2] == 1, "The second coefficient should 1.");
306 314

	
307 315
  LpSolver::DualExpr de = lp.col(x1);
308 316
  check(  de[upright] == 1, "The first coefficient should 1.");
309 317

	
310 318
  LpSolver* clp = lp.cloneSolver();
311 319

	
312 320
  //Testing the problem retrieving routines
313 321
  check(clp->objCoeff(x1)==1,"First term should be 1 in the obj function!");
314 322
  check(clp->sense() == clp->MAX,"This is a maximization!");
315 323
  check(clp->coeff(upright,x1)==1,"The coefficient in question is 1!");
316 324
  //  std::cout<<lp.colLowerBound(x1)<<std::endl;
317 325
  check(clp->colLowerBound(x1)==0,
318 326
        "The lower bound for variable x1 should be 0.");
319 327
  check(clp->colUpperBound(x1)==LpSolver::INF,
320 328
        "The upper bound for variable x1 should be infty.");
321 329

	
322 330
  check(lp.rowLowerBound(upright)==-LpSolver::INF,
323 331
        "The lower bound for the first row should be -infty.");
324 332
  check(lp.rowUpperBound(upright)==1,
325 333
        "The upper bound for the first row should be 1.");
326 334
  e = clp->row(upright);
327 335
  check(e[x1] == 1, "The first coefficient should 1.");
328 336
  check(e[x2] == 1, "The second coefficient should 1.");
329 337

	
330 338
  de = clp->col(x1);
331 339
  check(de[upright] == 1, "The first coefficient should 1.");
332 340

	
333 341
  delete clp;
334 342

	
335 343
  //Maximization of x1+x2
336 344
  //over the triangle with vertices (0,0) (0,1) (1,0)
337 345
  double expected_opt=1;
338 346
  solveAndCheck(lp, LpSolver::OPTIMAL, expected_opt);
339 347

	
340 348
  //Minimization
341 349
  lp.sense(lp.MIN);
342 350
  expected_opt=0;
343 351
  solveAndCheck(lp, LpSolver::OPTIMAL, expected_opt);
344 352

	
345 353
  //Vertex (-1,0) instead of (0,0)
346 354
  lp.colLowerBound(x1, -LpSolver::INF);
347 355
  expected_opt=-1;
348 356
  solveAndCheck(lp, LpSolver::OPTIMAL, expected_opt);
349 357

	
350 358
  //Erase one constraint and return to maximization
351 359
  lp.erase(upright);
352 360
  lp.sense(lp.MAX);
353 361
  expected_opt=LpSolver::INF;
354 362
  solveAndCheck(lp, LpSolver::UNBOUNDED, expected_opt);
355 363

	
356 364
  //Infeasibilty
357 365
  lp.addRow(x1+x2 <=-2);
358 366
  solveAndCheck(lp, LpSolver::INFEASIBLE, expected_opt);
359 367

	
360 368
}
361 369

	
362 370
template<class LP>
363 371
void cloneTest()
364 372
{
365 373
  //Test for clone/new
366 374

	
367 375
  LP* lp = new LP();
368 376
  LP* lpnew = lp->newSolver();
369 377
  LP* lpclone = lp->cloneSolver();
370 378
  delete lp;
371 379
  delete lpnew;
372 380
  delete lpclone;
373 381
}
374 382

	
375 383
int main()
376 384
{
377 385
  LpSkeleton lp_skel;
378 386
  lpTest(lp_skel);
379 387

	
380 388
#ifdef LEMON_HAVE_GLPK
381 389
  {
382 390
    GlpkLp lp_glpk1,lp_glpk2;
383 391
    lpTest(lp_glpk1);
384 392
    aTest(lp_glpk2);
385 393
    cloneTest<GlpkLp>();
386 394
  }
387 395
#endif
388 396

	
389 397
#ifdef LEMON_HAVE_CPLEX
390 398
  try {
391 399
    CplexLp lp_cplex1,lp_cplex2;
392 400
    lpTest(lp_cplex1);
393 401
    aTest(lp_cplex2);
394 402
    cloneTest<CplexLp>();
395 403
  } catch (CplexEnv::LicenseError& error) {
396 404
    check(false, error.what());
397 405
  }
398 406
#endif
399 407

	
400 408
#ifdef LEMON_HAVE_SOPLEX
401 409
  {
402 410
    SoplexLp lp_soplex1,lp_soplex2;
403 411
    lpTest(lp_soplex1);
404 412
    aTest(lp_soplex2);
405 413
    cloneTest<SoplexLp>();
406 414
  }
407 415
#endif
408 416

	
409 417
#ifdef LEMON_HAVE_CLP
410 418
  {
411 419
    ClpLp lp_clp1,lp_clp2;
412 420
    lpTest(lp_clp1);
413 421
    aTest(lp_clp2);
414 422
    cloneTest<ClpLp>();
415 423
  }
416 424
#endif
417 425

	
418 426
  return 0;
419 427
}
0 comments (0 inline)