Changeset e0fcac5f2c in tspsg for src/tspsolver.cpp
- Timestamp:
- Oct 24, 2009, 3:37:48 PM (15 years ago)
- Branches:
- 0.1.3.145-beta1-symbian, 0.1.4.170-beta2-bb10, appveyor, imgbot, master, readme
- Children:
- 8af59c4aa3
- Parents:
- 85ad815b0b
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
src/tspsolver.cpp
r85ad815b0b re0fcac5f2c 30 30 } 31 31 32 void CTSPSolver::cleanup() 33 { 34 route.clear(); 35 mayNotBeOptimal = false; 36 } 37 38 double CTSPSolver::findMinInRow(int nRow, tMatrix matrix, int exc) 39 { 40 double min = INFINITY; 41 for (int k = 0; k < nCities; k++) 42 if (((k != exc)) && (min > matrix.at(nRow).at(k))) 43 min = matrix.at(nRow).at(k); 44 return min == INFINITY ? 0 : min; 45 } 46 47 double CTSPSolver::findMinInCol(int nCol, tMatrix matrix, int exr) 48 { 49 double min = INFINITY; 50 for (int k = 0; k < nCities; k++) 51 if ((k != exr) && (min > matrix.at(k).at(nCol))) 52 min = matrix.at(k).at(nCol); 53 return min == INFINITY ? 0 : min; 54 } 55 56 void CTSPSolver::subRow(tMatrix &matrix, int nRow, double val) 57 { 58 for (int k = 0; k < nCities; k++) 59 if (k != nRow) 60 matrix[nRow][k] -= val; 61 } 62 63 void CTSPSolver::subCol(tMatrix &matrix, int nCol, double val) 64 { 65 for (int k = 0; k < nCities; k++) 66 if (k != nCol) 67 matrix[k][nCol] -= val; 68 } 32 /*! 33 * \brief Returns the sorted optimal path, starting from City 1. 34 * \return A string, containing sorted optimal path. 35 */ 36 QString CTSPSolver::getSortedPath() const 37 { 38 if (!root || route.isEmpty() || (route.size() != nCities)) 39 return QString(); 40 41 int i = 0; // We start from City 1 42 QString path = trUtf8("City %1").arg(1) + " -> "; 43 while ((i = route[i]) != 0) { 44 path += trUtf8("City %1").arg(i + 1) + " -> "; 45 } 46 // And finish in City 1, too 47 path += trUtf8("City %1").arg(1); 48 49 return path; 50 } 51 52 /*! 53 * \brief Returns CTSPSolver's version ID. 54 * \return A string: <b>\$Id$</b>. 55 */ 56 QString CTSPSolver::getVersionId() 57 { 58 return QString("$Id$"); 59 } 60 61 /*! 62 * \brief Returns whether or not the solution is definitely optimal. 63 * \return \c true if solution is definitely optimal, otherwise \c false. 64 * 65 * The solution may need some further interations to determine whether it is optimal. 66 * In such cases this function returns \c false. 67 */ 68 bool CTSPSolver::isOptimal() const 69 { 70 return !mayNotBeOptimal; 71 } 72 73 /*! 74 * \brief Solves the given task. 75 * \param numCities Number of cities in the task. 76 * \param task The matrix of city-to-city travel costs. 77 * \param parent The parent widget for displaying messages and dialogs. 78 * \return Pointer to the root of the solution tree. 79 * 80 * \todo TODO: Comment the algorithm. 81 */ 82 sStep *CTSPSolver::solve(int numCities, tMatrix task, QWidget *parent) 83 { 84 if (numCities <= 1) 85 return NULL; 86 cleanup(); 87 nCities = numCities; 88 QProgressDialog pd(parent); 89 QProgressBar *pb = new QProgressBar(&pd); 90 pb->setAlignment(Qt::AlignCenter); 91 pb->setFormat(trUtf8("%v of %m parts found")); 92 pd.setBar(pb); 93 pd.setMaximum(nCities); 94 pd.setMinimumDuration(1000); 95 pd.setLabelText(trUtf8("Calculating optimal route...")); 96 pd.setWindowTitle(trUtf8("Solution Progress")); 97 pd.setWindowModality(Qt::ApplicationModal); 98 pd.setValue(0); 99 100 sStep *step = new sStep(); 101 step->matrix = task; 102 step->price = align(step->matrix); 103 root = step; 104 105 sStep *left, *right; 106 int nRow, nCol; 107 bool firstStep = true; 108 double check; 109 while (this->route.size() < nCities) { 110 // forbidden.clear(); 111 step->alts = findCandidate(step->matrix,nRow,nCol); 112 while (hasSubCycles(nRow,nCol)) { 113 // forbidden[nRow] = nCol; 114 step->matrix[nRow][nCol] = INFINITY; 115 step->price += align(step->matrix); 116 step->alts = findCandidate(step->matrix,nRow,nCol); 117 } 118 if ((nRow == -1) || (nCol == -1) || pd.wasCanceled()) { 119 root = NULL; 120 break; 121 } 122 123 // Route with (nRow,nCol) path 124 right = new sStep(); 125 right->matrix = step->matrix; 126 for (int k = 0; k < nCities; k++) { 127 if (k != nCol) 128 right->matrix[nRow][k] = INFINITY; 129 if (k != nRow) 130 right->matrix[k][nCol] = INFINITY; 131 } 132 right->price = step->price + align(right->matrix); 133 // Forbid selected route to exclude its reuse in next steps. 134 right->matrix[nCol][nRow] = INFINITY; 135 right->matrix[nRow][nCol] = INFINITY; 136 137 // Route without (nRow,nCol) path 138 left = new sStep(); 139 left->matrix = step->matrix; 140 left->matrix[nRow][nCol] = INFINITY; 141 left->price = step->price + align(left->matrix); 142 143 step->candidate.nRow = nRow; 144 step->candidate.nCol = nCol; 145 step->plNode = left; 146 step->prNode = right; 147 148 if (right->price <= left->price) { 149 // Route with (nRow,nCol) path is cheaper 150 step = right; 151 this->route[nRow] = nCol; 152 pd.setValue(this->route.size()); 153 if (firstStep) { 154 check = left->price; 155 firstStep = false; 156 } 157 } else { 158 // Route without (nRow,nCol) path is cheaper 159 step = left; 160 qApp->processEvents(); 161 if (firstStep) { 162 check = right->price; 163 firstStep = false; 164 } 165 } 166 } 167 168 if (!root && !pd.wasCanceled()) { 169 pd.reset(); 170 QMessageBox(QMessageBox::Warning,trUtf8("Solution Result"),trUtf8("Unable to find solution.\nMaybe, this task has no solutions."),QMessageBox::Ok,parent).exec(); 171 } 172 173 qApp->processEvents(); 174 175 if (root) { 176 route = this->route; 177 mayNotBeOptimal = (check < step->price); 178 } 179 return root; 180 } 181 182 /* Privates **********************************************************/ 69 183 70 184 double CTSPSolver::align(tMatrix &matrix) … … 87 201 } 88 202 return r; 203 } 204 205 void CTSPSolver::cleanup() 206 { 207 route.clear(); 208 mayNotBeOptimal = false; 89 209 } 90 210 … … 112 232 } 113 233 234 double CTSPSolver::findMinInCol(int nCol, tMatrix matrix, int exr) 235 { 236 double min = INFINITY; 237 for (int k = 0; k < nCities; k++) 238 if ((k != exr) && (min > matrix.at(k).at(nCol))) 239 min = matrix.at(k).at(nCol); 240 return min == INFINITY ? 0 : min; 241 } 242 243 double CTSPSolver::findMinInRow(int nRow, tMatrix matrix, int exc) 244 { 245 double min = INFINITY; 246 for (int k = 0; k < nCities; k++) 247 if (((k != exc)) && (min > matrix.at(nRow).at(k))) 248 min = matrix.at(nRow).at(k); 249 return min == INFINITY ? 0 : min; 250 } 251 114 252 bool CTSPSolver::hasSubCycles(int nRow, int nCol) 115 253 { … … 126 264 } 127 265 128 /*! 129 * \brief Solves the given task. 130 * \param numCities Number of cities in the task. 131 * \param task The matrix of city-to-city travel costs. 132 * \param parent The parent widget for displaying messages and dialogs. 133 * \return Pointer to the root of the solution tree. 134 * 135 * \todo TODO: Comment the algorithm. 136 */ 137 sStep *CTSPSolver::solve(int numCities, tMatrix task, QWidget *parent) 138 { 139 if (numCities <= 1) 140 return NULL; 141 cleanup(); 142 nCities = numCities; 143 QProgressDialog pd(parent); 144 QProgressBar *pb = new QProgressBar(&pd); 145 pb->setAlignment(Qt::AlignCenter); 146 pb->setFormat(trUtf8("%v of %m parts found")); 147 pd.setBar(pb); 148 pd.setMaximum(nCities); 149 pd.setMinimumDuration(1000); 150 pd.setLabelText(trUtf8("Calculating optimal route...")); 151 pd.setWindowTitle(trUtf8("Solution Progress")); 152 pd.setWindowModality(Qt::ApplicationModal); 153 pd.setValue(0); 154 155 sStep *step = new sStep(); 156 step->matrix = task; 157 step->price = align(step->matrix); 158 root = step; 159 160 sStep *left, *right; 161 int nRow, nCol; 162 bool firstStep = true; 163 double check; 164 while (this->route.size() < nCities) { 165 // forbidden.clear(); 166 step->alts = findCandidate(step->matrix,nRow,nCol); 167 while (hasSubCycles(nRow,nCol)) { 168 // forbidden[nRow] = nCol; 169 step->matrix[nRow][nCol] = INFINITY; 170 step->price += align(step->matrix); 171 step->alts = findCandidate(step->matrix,nRow,nCol); 172 } 173 if ((nRow == -1) || (nCol == -1) || pd.wasCanceled()) { 174 root = NULL; 175 break; 176 } 177 178 // Route with (nRow,nCol) path 179 right = new sStep(); 180 right->matrix = step->matrix; 181 for (int k = 0; k < nCities; k++) { 182 if (k != nCol) 183 right->matrix[nRow][k] = INFINITY; 184 if (k != nRow) 185 right->matrix[k][nCol] = INFINITY; 186 } 187 right->price = step->price + align(right->matrix); 188 // Forbid selected route to exclude its reuse in next steps. 189 right->matrix[nCol][nRow] = INFINITY; 190 right->matrix[nRow][nCol] = INFINITY; 191 192 // Route without (nRow,nCol) path 193 left = new sStep(); 194 left->matrix = step->matrix; 195 left->matrix[nRow][nCol] = INFINITY; 196 left->price = step->price + align(left->matrix); 197 198 step->candidate.nRow = nRow; 199 step->candidate.nCol = nCol; 200 step->plNode = left; 201 step->prNode = right; 202 203 if (right->price <= left->price) { 204 // Route with (nRow,nCol) path is cheaper 205 step = right; 206 this->route[nRow] = nCol; 207 pd.setValue(this->route.size()); 208 if (firstStep) { 209 check = left->price; 210 firstStep = false; 211 } 212 } else { 213 // Route without (nRow,nCol) path is cheaper 214 step = left; 215 qApp->processEvents(); 216 if (firstStep) { 217 check = right->price; 218 firstStep = false; 219 } 220 } 221 } 222 223 if (!root && !pd.wasCanceled()) { 224 pd.reset(); 225 QMessageBox(QMessageBox::Warning,trUtf8("Solution Result"),trUtf8("Unable to find solution.\nMaybe, this task has no solutions."),QMessageBox::Ok,parent).exec(); 226 } 227 228 qApp->processEvents(); 229 230 if (root) { 231 route = this->route; 232 mayNotBeOptimal = (check < step->price); 233 } 234 return root; 235 } 236 237 /*! 238 * \brief Returns the sorted optimal path, starting from City 1. 239 * \return A string, containing sorted optimal path. 240 */ 241 QString CTSPSolver::getSortedPath() const 242 { 243 if (!root || route.isEmpty() || (route.size() != nCities)) 244 return QString(); 245 246 int i = 0; // We start from City 1 247 QString path = trUtf8("City %1").arg(1) + " -> "; 248 while ((i = route[i]) != 0) { 249 path += trUtf8("City %1").arg(i + 1) + " -> "; 250 } 251 // And finish in City 1, too 252 path += trUtf8("City %1").arg(1); 253 254 return path; 255 } 256 257 /*! 258 * \brief Returns CTSPSolver's version ID. 259 * \return A string: <b>\$Id$</b>. 260 */ 261 QString CTSPSolver::getVersionId() 262 { 263 return QString("$Id$"); 264 } 265 266 /*! 267 * \brief Returns whether or not the solution is definitely optimal. 268 * \return \c true if solution is definitely optimal, otherwise \c false. 269 * 270 * The solution may need some further interations to determine whether it is optimal. 271 * In such cases this function returns \c false. 272 */ 273 bool CTSPSolver::isOptimal() const 274 { 275 return !mayNotBeOptimal; 276 } 266 void CTSPSolver::subCol(tMatrix &matrix, int nCol, double val) 267 { 268 for (int k = 0; k < nCities; k++) 269 if (k != nCol) 270 matrix[k][nCol] -= val; 271 } 272 273 void CTSPSolver::subRow(tMatrix &matrix, int nRow, double val) 274 { 275 for (int k = 0; k < nCities; k++) 276 if (k != nRow) 277 matrix[nRow][k] -= val; 278 }
Note: See TracChangeset
for help on using the changeset viewer.