Пример #1
0
//-----------------------------------------------------------------------------
void CbcSolverLongThin::resolve()
{
  int * whichRow = NULL;
  int * whichColumn = NULL;
  // problem may be small enough to do nested search
  const double * colLower = modelPtr_->columnLower();
  const double * colUpper = modelPtr_->columnUpper();
  
  int numberIntegers = model_->numberIntegers();
  const int * integerVariable = model_->integerVariable();
  int numberRows=modelPtr_->numberRows();
  int numberColumns = modelPtr_->numberColumns();
  
  int i;
  int nFix=0;
  int nNewRow=0;
  int nNewCol=0;
  int sizeDynamic = COIN_INT_MAX;
  int smallOriginalNumberRows=0;
  if (algorithm_==0) {
    for (i=0;i<numberIntegers;i++) {
      int iColumn=integerVariable[i];
      if (colLower[iColumn]==colUpper[iColumn])
        nFix++;
    }
  } else {
    whichRow = new int[numberRows];
    whichColumn = new int [numberColumns];
    // more sophisticated
    OsiCuts cs;
    tryCut->generateCuts(*this,cs);
    int numberCuts = cs.sizeColCuts();
    if (numberCuts) {
      for ( i = 0 ; i < numberCuts ; i++) {
        const OsiColCut *thisCut = cs.colCutPtr(i) ;
	const CoinPackedVector & ubs = thisCut->ubs() ;
	int n = ubs.getNumElements() ;
	const int * which = ubs.getIndices() ;
	const double * values = ubs.getElements() ;
	for (int j = 0;j<n;j++) {
	  int iColumn = which[j] ;
	  this->setColUpper(iColumn,values[j]) ;
        }
      }
    }
#if 1
    const int * duplicate = tryCut->duplicate();
    sizeDynamic = tryCut->sizeDynamic();
    int nOrig = tryCut->numberOriginalRows();
    for (i=0;i<nOrig;i++) {
      if (duplicate[i]==-1)
        whichRow[nNewRow++]=i;
      else
        modelPtr_->setRowStatus(i,ClpSimplex::basic);
    }
    smallOriginalNumberRows=nNewRow;
    for (;i<numberRows;i++) {
      whichRow[nNewRow++]=i;
    }
#else
    for (i=0;i<numberRows;i++) 
      whichRow[i]=i;
    nNewRow=numberRows;
#endif
    for (i=0;i<numberIntegers;i++) {
      int iColumn=integerVariable[i];
      if (colLower[iColumn]==colUpper[iColumn])
        nFix++;
      bool choose;
      if (algorithm_==1)
        choose = true;
      else
        choose = (node_[i]>count_-memory_&&node_[i]>0);
      if ((choose&&colUpper[i])
	  ||(modelPtr_->getStatus(i)!=ClpSimplex::atLowerBound&&
             modelPtr_->getStatus(i)!=ClpSimplex::isFixed)
	  ||colLower[i]>0.0)
	whichColumn[nNewCol++]=i;
    }
  }
  if (nestedSearch_<1.0&&model_&&model_->phase()==2) {
    if (((double) sizeDynamic)*((double) nNewCol)<1000000000&&sizeDynamic<10000000) {
      // could do Dynamic Programming
      // back to original number of rows
      nNewRow = smallOriginalNumberRows;
      // and get rid of any basics
      int nNewCol=0;
      for (i=0;i<numberColumns;i++) {
        if (colUpper[i]||colLower[i]>0.0)
          whichColumn[nNewCol++]=i;
      }
      ClpSimplex temp(modelPtr_,nNewRow,whichRow,nNewCol,whichColumn);
      int returnCode;
      double * rowLower2 = temp.rowLower();
      double * rowUpper2 = temp.rowUpper();
      int numberColumns2 = temp.numberColumns();
      double * colLower2 = temp.columnLower();
      double * colUpper2 = temp.columnUpper();
      const CoinPackedMatrix * matrix = temp.matrix();
      const double * element = matrix->getElements();
      const int * row = matrix->getIndices();
      const CoinBigIndex * columnStart = matrix->getVectorStarts();
      const int * columnLength = matrix->getVectorLengths();
      double offset=0.0;
      const double * objective = temp.objective();
      bool feasible=true;
      for (i=0;i<numberColumns2;i++) {
        double value = colLower2[i];
        if (value) {
          offset += value*objective[i];
          colLower2[i]=0.0;
          colUpper2[i] -= value;
          for (int j=columnStart[i];
               j<columnStart[i]+columnLength[i];j++) {
            int iRow=row[j];
            rowLower2[iRow] -= value*element[j]; 
            rowUpper2[iRow] -= value*element[j]; 
            if (rowUpper2[iRow]<-1.0e-8) {
              feasible=false;
              printf("odd - problem is infeasible\n");
            }
          }
        }
      }
      temp.setObjectiveOffset(-offset);
      OsiClpSolverInterface temp2(&temp);
      double * solutionDP = NULL;
      if (feasible) {
        for (i=0;i<numberColumns2;i++) 
          temp2.setInteger(i);
        CbcModel modelSmall(temp2);
        modelSmall.messageHandler()->setLogLevel(0);
        CbcFathomDynamicProgramming fathom1(modelSmall);
        // Set maximum space allowed
        fathom1.setMaximumSize(100000000);
        temp2.writeMps("small");
        returnCode=fathom1.fathom(solutionDP);
        if (returnCode!=1) {
          printf("probably not enough memory\n");
          abort();
        }
      }
      if (solutionDP) {
        double objValue = 0.0;
        double * solution = modelPtr_->primalColumnSolution();
        const double * objective = modelPtr_->objective();
        for (i=0;i<numberColumns;i++) 
          solution[i]=colLower[i];
        for (i=0;i<nNewCol;i++) {
          int iColumn = whichColumn[i];
          solution[iColumn]+=solutionDP[i];
        }
        for (i=0;i<numberColumns;i++) 
          objValue += solution[i]*objective[i];
        if (objValue<model_->getCutoff()) {
          printf("good solution %g by dynamic programming\n",objValue);
          returnCode = 0;
          // paranoid check
          double * rowLower = modelPtr_->rowLower();
          double * rowUpper = modelPtr_->rowUpper();
          // Column copy
          const CoinPackedMatrix * matrix2 = modelPtr_->matrix();
          element = matrix2->getElements();
          row = matrix2->getIndices();
          columnStart = matrix2->getVectorStarts();
          columnLength = matrix2->getVectorLengths();
          double * rowActivity = new double [numberRows];
          memset(rowActivity,0,numberRows*sizeof(double));
          for (i=0;i<numberColumns;i++) {
            int j;
            double value = solution[i];
            assert (value>=colLower[i]&&value<=colUpper[i]);
            if (value) {
              printf("%d has value %g\n",i,value);
              for (j=columnStart[i];
                   j<columnStart[i]+columnLength[i];j++) {
                int iRow=row[j];
                rowActivity[iRow] += value*element[j];
              }
            }
          }
          // check was feasible
          bool feasible=true;
          for (i=0;i<numberRows;i++) {
            if(rowActivity[i]<rowLower[i]) {
              if (rowActivity[i]<rowLower[i]-1.0e-8)
                feasible = false;
            } else if(rowActivity[i]>rowUpper[i]) {
              if (rowActivity[i]>rowUpper[i]+1.0e-8)
                feasible = false;
            }
          }
          if (!feasible) {
            printf("** Bad solution by dynamic programming\n");
            abort();
          }
          delete [] rowActivity;
          model_->setBestSolution(CBC_TREE_SOL,objValue,solution);
        } else {
          returnCode=2;
        }
      } else {
        returnCode=2;
      }
      temp2.releaseClp();
      modelPtr_->setProblemStatus(1);
      delete [] whichRow;
      delete [] whichColumn;
      return;
    }
    if (nFix>nestedSearch_*numberIntegers) {
      // Do nested search
      // back to original number of rows
      nNewRow = smallOriginalNumberRows;
      // and get rid of any basics
      int nNewCol=0;
      for (i=0;i<numberColumns;i++) {
        if (colUpper[i]||colLower[i]>0.0)
          whichColumn[nNewCol++]=i;
      }
#if 0
      // We clone from continuous solver so set some stuff
      OsiSolverInterface * solver = model_->continuousSolver();
      CbcSolverLongThin * osiclp = dynamic_cast< CbcSolverLongThin*> (solver);
      assert (osiclp);
      // up special options
      if (osiclp->specialOptions()==3)
	osiclp->setSpecialOptions(7);
      double saveNested = osiclp->getNested();
      int saveAlgorithm = osiclp->getAlgorithm();
      osiclp->setNested(1.0);
      osiclp->setAlgorithm(0);
      int numberObjects = model_->numberObjects();
      if (numberObjects>model_->numberIntegers()) {
	// for now only integers
	//assert (numberObjects == model_->numberIntegers()+1);
	model_->setNumberObjects(model_->numberIntegers());
        // try follow on
	//model_->setNumberObjects(model_->numberIntegers()+1);
      }
      double saveMaxTime = model_->getDblParam(CbcModel::CbcMaximumSeconds);
      model_->setDblParam(CbcModel::CbcMaximumSeconds,1.0e5);
      // up special options
#if 1
      int returnCode= model_->subBranchAndBound(colLower,colUpper,2000);
#else
      CbcModel * model3 = model_->cleanModel(colLower,colUpper);
      // integer presolve
      int returnCode=0;
      CbcModel * model2 = model3->integerPresolve(false);
      if (!model2||!model2->getNumRows()) {
        delete model2;
        delete model3;
        returnCode= 2;
      } else {
        if (handler_->logLevel()>1)
          printf("Reduced model has %d rows and %d columns\n",
                 model2->getNumRows(),model2->getNumCols());
        if (true) {
          OsiSolverInterface * solver = model2->solver();
          OsiSolverInterface * osiclp = dynamic_cast< OsiSolverInterface*> (solver);
          assert (osiclp);
          int * priority = new int [numberColumns+1];
          int n=0;
          int iColumn;
          for ( iColumn=0;iColumn<numberColumns;iColumn++) {
            if (solver->isInteger(iColumn)) {
              priority[n++]=10000;
            }
          }
          priority[n]=1;
          CbcObject * newObject =new CbcFollowOn2(model2);
          model2->addObjects(1,&newObject);
          delete newObject;
          model2->passInPriorities(priority,false);
          delete [] priority;
        }
        returnCode= model_->subBranchAndBound(model3,model2,4000);
      }
#endif
      model_->setDblParam(CbcModel::CbcMaximumSeconds,saveMaxTime);
      model_->setNumberObjects(numberObjects);
      osiclp->setNested(saveNested);
      osiclp->setAlgorithm(saveAlgorithm);
#else
      // start again very simply
      ClpSimplex temp(modelPtr_,nNewRow,whichRow,nNewCol,whichColumn);
      int returnCode;
      OsiClpSolverInterface temp2(&temp);
      temp2.setupForRepeatedUse(2);
      int numberColumns2 = temp.numberColumns();
      const double * colUpper2 = temp2.getColUpper();
      const double * colLower2 = temp2.getColLower();
      const double * solution2 = temp.getColSolution();
      double * cleanSolution2 = new double [numberColumns2];
      for (i=0;i<numberColumns2;i++) {
        temp2.setInteger(i);
        double value = solution2[i];
        value = CoinMin(CoinMax(value,colLower2[i]),colUpper2[i]);
        cleanSolution2[i] = value;
      }
      temp2.setColSolution(cleanSolution2);
      delete [] cleanSolution2;
      CbcModel modelSmall(temp2);
      modelSmall.setNumberStrong(0);
      CglProbing generator1;
      generator1.setUsingObjective(true);
      generator1.setMaxPass(3);
      generator1.setMaxProbe(100);
      generator1.setMaxLook(50);
      generator1.setRowCuts(3);
      
      CglGomory generator2;
      // try larger limit
      generator2.setLimit(300);
      
      CglKnapsackCover generator3;
      
      CglOddHole generator4;
      generator4.setMinimumViolation(0.005);
      generator4.setMinimumViolationPer(0.00002);
      // try larger limit
      generator4.setMaximumEntries(200);
      
      CglClique generator5;
      generator5.setStarCliqueReport(false);
      generator5.setRowCliqueReport(false);
      
      CglMixedIntegerRounding mixedGen;
      CglFlowCover flowGen;
      
      // Add in generators
      modelSmall.addCutGenerator(&generator1,-1,"Probing",true,false,false,-1);
      modelSmall.addCutGenerator(&generator2,-99,"Gomory",true,false,false,-99);
      modelSmall.addCutGenerator(&generator3,-99,"Knapsack",true,false,false,-99);
      modelSmall.addCutGenerator(&generator4,-99,"OddHole",true,false,false,-99);
      modelSmall.addCutGenerator(&generator5,-99,"Clique",true,false,false,-99);
      modelSmall.addCutGenerator(&flowGen,-99,"FlowCover",true,false,false,-99);
      modelSmall.addCutGenerator(&mixedGen,-99,"MixedIntegerRounding",true,false,false,-100);
#if 1
      const CoinPackedMatrix * matrix = temp2.getMatrixByCol();
      const int * columnLength = matrix->getVectorLengths();
      int * priority = new int [numberColumns2+1];
      // do pseudo costs and priorities - take a reasonable guess
      CbcObject ** objects = new CbcObject * [numberColumns2+1];
      int n=0;
      const double * objective = modelSmall.getObjCoefficients();
      for (i=0;i<numberColumns2;i++) {
        CbcSimpleIntegerPseudoCost * newObject =
          new CbcSimpleIntegerPseudoCost(&modelSmall,n,i,objective[i],0.5*objective[i]);
        newObject->setMethod(3);
        objects[n]= newObject;
        priority[n++]=10000-columnLength[i];
      }
      priority[n]=1;
      objects[n++]=new CbcFollowOn2(&modelSmall);
      modelSmall.addObjects(n,objects);
      for (i=0;i<n;i++)
        delete objects[i];
      delete [] objects;
      modelSmall.passInPriorities(priority,false);
      delete [] priority;
#endif
      modelSmall.setCutoff(model_->getCutoff());
      //if (!onPathX&&modelSmall.getCutoff()>480.5)
      //modelSmall.setCutoff(480.5);
      //printf("cutoff %g\n",model_->getCutoff());
      modelSmall.messageHandler()->setLogLevel(1);
      modelSmall.solver()->messageHandler()->setLogLevel(0);
      modelSmall.messagesPointer()->setDetailMessage(3,9);
      modelSmall.messagesPointer()->setDetailMessage(3,6);
      modelSmall.messagesPointer()->setDetailMessage(3,4);
      modelSmall.messagesPointer()->setDetailMessage(3,13);
      modelSmall.messagesPointer()->setDetailMessage(3,14);
      modelSmall.messagesPointer()->setDetailMessage(3,1);
      modelSmall.messagesPointer()->setDetailMessage(3,3007);
      modelSmall.branchAndBound();
      temp2.releaseClp();
      if (modelSmall.bestSolution()) {
        double objValue = 0.0;
        const double * solution2 = modelSmall.bestSolution();
        double * solution = modelPtr_->primalColumnSolution();
        const double * objective = modelPtr_->objective();
        for (i=0;i<numberColumns;i++) 
          solution[i]=colLower[i];
        for (i=0;i<nNewCol;i++) {
          int iColumn = whichColumn[i];
          solution[iColumn]=solution2[i];
        }
        for (i=0;i<numberColumns;i++) 
          objValue += solution[i]*objective[i];
        assert (objValue<model_->getCutoff());
        if (objValue<model_->getCutoff()) {
          //printf("good solution \n");
          model_->setBestSolution(CBC_TREE_SOL,objValue,solution);
          returnCode = 0;
        } else {
          returnCode=2;
        }
      } else {
        returnCode=2;
      }
#endif
      if (returnCode!=0&&returnCode!=2) {
	printf("pretending entire search done\n");
	returnCode=0;
      }
      if (returnCode==0||returnCode==2) {
	modelPtr_->setProblemStatus(1);
        delete [] whichRow;
        delete [] whichColumn;
	return;
     }
    }
  }
  if ((count_<100&&algorithm_==2)||!algorithm_) {
    delete [] whichRow;
    delete [] whichColumn;
    assert(!modelPtr_->specialOptions());
    int saveOptions = modelPtr_->specialOptions();
    int startFinishOptions;
    bool takeHint;
    OsiHintStrength strength;
    bool gotHint = (getHintParam(OsiDoInBranchAndCut,takeHint,strength));
    assert (gotHint);
    if (strength!=OsiHintIgnore&&takeHint) {
      // could do something - think about it
      //printf("thin hint %d %c\n",strength,takeHint ? 'T' :'F');
    }
    if((specialOptions_&1)==0) {
      startFinishOptions=0;
      modelPtr_->setSpecialOptions(saveOptions|(64|1024));
    } else {
      startFinishOptions=1+2+4;
      if((specialOptions_&4)==0) 
	modelPtr_->setSpecialOptions(saveOptions|(64|128|512|1024|4096));
      else
	modelPtr_->setSpecialOptions(saveOptions|(64|128|512|1024|2048|4096));
    }
    //printf("thin options %d size %d\n",modelPtr_->specialOptions(),modelPtr_->numberColumns());
    setBasis(basis_,modelPtr_);
    //modelPtr_->setLogLevel(1);
    modelPtr_->dual(0,0);
    basis_ = getBasis(modelPtr_);
    modelPtr_->setSpecialOptions(saveOptions);
    if (modelPtr_->status()==0) {
      count_++;
      double * solution = modelPtr_->primalColumnSolution();
      int i;
      for (i=0;i<numberColumns;i++) {
	if (solution[i]>1.0e-6||modelPtr_->getStatus(i)==ClpSimplex::basic) {
	  node_[i]=CoinMax(count_,node_[i]);
	  howMany_[i]++;
	}
      }
    } else {
      if (!algorithm_==2)
	printf("infeasible early on\n");
    }
  } else {
    // use counts
    int i;
    const double * lower = modelPtr_->columnLower();
    const double * upper = modelPtr_->columnUpper();
    setBasis(basis_,modelPtr_);
    ClpSimplex * temp = new ClpSimplex(modelPtr_,nNewRow,whichRow,nNewCol,whichColumn);
    //temp->setLogLevel(2);
    //printf("small has %d rows and %d columns\n",nNewRow,nNewCol);
    temp->setSpecialOptions(128+512);
    temp->setDualObjectiveLimit(1.0e50);
    temp->dual();
    if (temp->status()) {
      // In some cases we know that it must be infeasible
      if (believeInfeasible_||algorithm_==1) {
	modelPtr_->setProblemStatus(1);
	printf("assuming infeasible!\n");
	//modelPtr_->writeMps("infeas.mps");
	//temp->writeMps("infeas2.mps");
	//abort();
	delete temp;
        delete [] whichRow;
	delete [] whichColumn;
	return;
      }
    }
    double * solution = modelPtr_->primalColumnSolution();
    if (!temp->status()) {
      const double * solution2 = temp->primalColumnSolution();
      memset(solution,0,numberColumns*sizeof(double));
      for (i=0;i<nNewCol;i++) {
	int iColumn = whichColumn[i];
	solution[iColumn]=solution2[i];
	modelPtr_->setStatus(iColumn,temp->getStatus(i));
      }
      double * rowSolution = modelPtr_->primalRowSolution();
      const double * rowSolution2 = temp->primalRowSolution();
      double * dual = modelPtr_->dualRowSolution();
      const double * dual2 = temp->dualRowSolution();
      memset(dual,0,numberRows*sizeof(double));
      for (i=0;i<nNewRow;i++) {
        int iRow=whichRow[i];
        modelPtr_->setRowStatus(iRow,temp->getRowStatus(i));
        rowSolution[iRow]=rowSolution2[i];
        dual[iRow]=dual2[i];
      }
      // See if optimal
      double * dj = modelPtr_->dualColumnSolution();
      // get reduced cost for large problem
      // this assumes minimization
      memcpy(dj,modelPtr_->objective(),numberColumns*sizeof(double));
      modelPtr_->transposeTimes(-1.0,dual,dj);
      modelPtr_->setObjectiveValue(temp->objectiveValue());
      modelPtr_->setProblemStatus(0);
      int nBad=0;
      
      for (i=0;i<numberColumns;i++) {
	if (modelPtr_->getStatus(i)==ClpSimplex::atLowerBound
	    &&upper[i]>lower[i]&&dj[i]<-1.0e-5)
	  nBad++;
      }
      //modelPtr_->writeMps("bada.mps");
      //temp->writeMps("badb.mps");
      if (nBad) {
        assert (algorithm_==2);
	//printf("%d bad\n",nBad);
	timesBad_++;
	modelPtr_->primal();
      }
    } else {
      // infeasible - do all
      modelPtr_->setSpecialOptions(64+128+512);
      setBasis(basis_,modelPtr_);
      //modelPtr_->setLogLevel(1);
      modelPtr_->dual(0,0);
      basis_ = getBasis(modelPtr_);
      modelPtr_->setSpecialOptions(0);
      if (modelPtr_->status()) {
	printf("really infeasible!\n");
	delete temp;
        delete [] whichRow;
	delete [] whichColumn;
	return;
      } else {
	printf("initially infeasible\n");
      }
    }
    delete temp;
    delete [] whichRow;
    delete [] whichColumn;
    basis_ = getBasis(modelPtr_);
    modelPtr_->setSpecialOptions(0);
    count_++;
    if ((count_%100)==0&&algorithm_==2)
      printf("count %d, bad %d\n",count_,timesBad_);
    for (i=0;i<numberColumns;i++) {
      if (solution[i]>1.0e-6||modelPtr_->getStatus(i)==ClpSimplex::basic) {
	node_[i]=CoinMax(count_,node_[i]);
	howMany_[i]++;
      }
    }
    if (modelPtr_->objectiveValue()>=modelPtr_->dualObjectiveLimit())
      modelPtr_->setProblemStatus(1);
  }
}
Пример #2
0
int main (int argc, const char *argv[])
{

  CbcSolver3 solver1;

  // Read in model using argv[1]
  // and assert that it is a clean model
  std::string mpsFileName;
  if (argc>=2) mpsFileName = argv[1];
  int numMpsReadErrors = solver1.readMps(mpsFileName.c_str(),"");
  assert(numMpsReadErrors==0);
  double time1 = CoinCpuTime();

  /* Options are:
     preprocess to do preprocessing
     time in minutes
     if 2 parameters and numeric taken as time
  */
  bool preProcess=false;
  double minutes=-1.0;
  int nGoodParam=0;
  for (int iParam=2; iParam<argc;iParam++) {
    if (!strcmp(argv[iParam],"preprocess")) {
      preProcess=true;
      nGoodParam++;
    } else if (!strcmp(argv[iParam],"time")) {
      if (iParam+1<argc&&isdigit(argv[iParam+1][0])) {
        minutes=atof(argv[iParam+1]);
        if (minutes>=0.0) {
          nGoodParam+=2;
          iParam++; // skip time
        }
      }
    }
  }
  if (nGoodParam==0&&argc==3&&isdigit(argv[2][0])) {
    // If time is given then stop after that number of minutes
    minutes = atof(argv[2]);
    if (minutes>=0.0) 
      nGoodParam=1;
  }
  if (nGoodParam!=argc-2) {
    printf("Usage <file> [preprocess] [time <minutes>] or <file> <minutes>\n");
    exit(1);
  }
  solver1.initialSolve();
  // Reduce printout
  solver1.setHintParam(OsiDoReducePrint,true,OsiHintTry);
  // Say we want scaling
  //solver1.setHintParam(OsiDoScale,true,OsiHintTry);
  //solver1.setCleanupScaling(1);
  // See if we want preprocessing
  OsiSolverInterface * solver2=&solver1;
  CglPreProcess process;
  if (preProcess) {
    /* Do not try and produce equality cliques and
       do up to 5 passes */
    solver2 = process.preProcess(solver1,false,5);
    if (!solver2) {
      printf("Pre-processing says infeasible\n");
      exit(2);
    }
    solver2->resolve();
  }
  CbcModel model(*solver2);
  // Point to solver
  OsiSolverInterface * solver3 = model.solver();
  CbcSolver3 * osiclp = dynamic_cast< CbcSolver3*> (solver3);
  assert (osiclp);
  const double fractionFix=0.985;
  osiclp->initialize(&model,NULL);
  osiclp->setAlgorithm(2);
  osiclp->setMemory(1000);
  osiclp->setNested(fractionFix);
  //osiclp->setNested(1.0); //off
  // Set up some cut generators and defaults
  // Probing first as gets tight bounds on continuous

  CglProbing generator1;
  generator1.setUsingObjective(true);
  generator1.setMaxPass(3);
  // Number of unsatisfied variables to look at
  generator1.setMaxProbe(10);
  // How far to follow the consequences
  generator1.setMaxLook(50);
  // Only look at rows with fewer than this number of elements
  generator1.setMaxElements(200);
  generator1.setRowCuts(3);

  CglGomory generator2;
  // try larger limit
  generator2.setLimit(300);

  CglKnapsackCover generator3;

  CglOddHole generator4;
  generator4.setMinimumViolation(0.005);
  generator4.setMinimumViolationPer(0.00002);
  // try larger limit
  generator4.setMaximumEntries(200);

  CglClique generator5;
  generator5.setStarCliqueReport(false);
  generator5.setRowCliqueReport(false);

  CglMixedIntegerRounding mixedGen;
  /* This is same as default constructor - 
     (1,true,1)
      I presume if maxAggregate larger then slower but maybe better
      criterion can be  1 through 3
      Reference: 
      Hugues Marchand and Laurence A. Wolsey
      Aggregation and Mixed Integer Rounding to Solve MIPs
      Operations Research, 49(3), May-June 2001.
   */
  int maxAggregate=1;
  bool multiply=true;
  int criterion=1;
  CglMixedIntegerRounding2 mixedGen2(maxAggregate,multiply,criterion);
  CglFlowCover flowGen;
  
  // Add in generators
  // Experiment with -1 and -99 etc
  model.addCutGenerator(&generator1,-99,"Probing");
  //model.addCutGenerator(&generator2,-1,"Gomory");
  //model.addCutGenerator(&generator3,-1,"Knapsack");
  //model.addCutGenerator(&generator4,-1,"OddHole");
  //model.addCutGenerator(&generator5,-1,"Clique");
  //model.addCutGenerator(&flowGen,-1,"FlowCover");
  //model.addCutGenerator(&mixedGen,-1,"MixedIntegerRounding");
  //model.addCutGenerator(&mixedGen2,-1,"MixedIntegerRounding2");
  // Say we want timings
  int numberGenerators = model.numberCutGenerators();
  int iGenerator;
  for (iGenerator=0;iGenerator<numberGenerators;iGenerator++) {
    CbcCutGenerator * generator = model.cutGenerator(iGenerator);
    generator->setTiming(true);
  }

  // Allow rounding heuristic

  CbcRounding heuristic1(model);
  model.addHeuristic(&heuristic1);

  // And Greedy heuristic

  CbcHeuristicGreedyCover heuristic2(model);
  // Use original upper and perturb more
  heuristic2.setAlgorithm(11);
  model.addHeuristic(&heuristic2);

  // Redundant definition of default branching (as Default == User)
  CbcBranchUserDecision branch;
  model.setBranchingMethod(&branch);

  // Definition of node choice
  CbcCompareUser compare;
  model.setNodeComparison(compare);

  int iColumn;
  int numberColumns = solver3->getNumCols();
  // do pseudo costs
  CbcObject ** objects = new CbcObject * [numberColumns+1];
  const CoinPackedMatrix * matrix = solver3->getMatrixByCol();
  // Column copy
  const int * columnLength = matrix->getVectorLengths();
  const double * objective = model.getObjCoefficients();
  int n=0;
  for (iColumn=0;iColumn<numberColumns;iColumn++) {
    if (solver3->isInteger(iColumn)) {
      double costPer = objective[iColumn]/ ((double) columnLength[iColumn]);
      CbcSimpleIntegerPseudoCost * newObject =
        new CbcSimpleIntegerPseudoCost(&model,n,iColumn,
                                       costPer,costPer);
      newObject->setMethod(3);
      objects[n++]= newObject;
    }
  }
  // and special fix lots branch
  objects[n++]=new CbcBranchToFixLots(&model,-1.0e-6,fractionFix+0.01,1,0,NULL);
  model.addObjects(n,objects);
  for (iColumn=0;iColumn<n;iColumn++)
    delete objects[iColumn];
  delete [] objects;
  // High priority for odd object
  int followPriority=1;
  model.passInPriorities(&followPriority,true);

  // Do initial solve to continuous
  model.initialSolve();

  // Could tune more
  model.setMinimumDrop(CoinMin(1.0,
			     fabs(model.getMinimizationObjValue())*1.0e-3+1.0e-4));

  if (model.getNumCols()<500)
    model.setMaximumCutPassesAtRoot(-100); // always do 100 if possible
  else if (model.getNumCols()<5000)
    model.setMaximumCutPassesAtRoot(100); // use minimum drop
  else
    model.setMaximumCutPassesAtRoot(20);
  //model.setMaximumCutPasses(1);

  // Do more strong branching if small
  //if (model.getNumCols()<5000)
  //model.setNumberStrong(10);
  // Switch off strong branching if wanted
  model.setNumberStrong(0);

  model.solver()->setIntParam(OsiMaxNumIterationHotStart,100);

  // If time is given then stop after that number of minutes
  if (minutes>=0.0) {
    std::cout<<"Stopping after "<<minutes<<" minutes"<<std::endl;
    model.setDblParam(CbcModel::CbcMaximumSeconds,60.0*minutes);
  }
  // Switch off most output
  if (model.getNumCols()<300000) {
    model.messageHandler()->setLogLevel(1);
    //model.solver()->messageHandler()->setLogLevel(0);
  } else {
    model.messageHandler()->setLogLevel(2);
    model.solver()->messageHandler()->setLogLevel(1);
  }
  //model.messageHandler()->setLogLevel(2);
  //model.solver()->messageHandler()->setLogLevel(2);
  //model.setPrintFrequency(50);
#define DEBUG_CUTS
#ifdef DEBUG_CUTS
  // Set up debugger by name (only if no preprocesing)
  if (!preProcess) {
    std::string problemName ;
    //model.solver()->getStrParam(OsiProbName,problemName) ;
    //model.solver()->activateRowCutDebugger(problemName.c_str()) ;
    model.solver()->activateRowCutDebugger("cap6000a") ;
  }
#endif

  // Do complete search
  try {
    model.branchAndBound();
  }
  catch (CoinError e) {
    e.print();
    if (e.lineNumber()>=0)
      std::cout<<"This was from a CoinAssert"<<std::endl;
    exit(0);
  }
  //void printHowMany();
  //printHowMany();
  std::cout<<mpsFileName<<" took "<<CoinCpuTime()-time1<<" seconds, "
	   <<model.getNodeCount()<<" nodes with objective "
	   <<model.getObjValue()
	   <<(!model.status() ? " Finished" : " Not finished")
	   <<std::endl;

  // Print more statistics
  std::cout<<"Cuts at root node changed objective from "<<model.getContinuousObjective()
	   <<" to "<<model.rootObjectiveAfterCuts()<<std::endl;

  for (iGenerator=0;iGenerator<numberGenerators;iGenerator++) {
    CbcCutGenerator * generator = model.cutGenerator(iGenerator);
    std::cout<<generator->cutGeneratorName()<<" was tried "
	     <<generator->numberTimesEntered()<<" times and created "
	     <<generator->numberCutsInTotal()<<" cuts of which "
	     <<generator->numberCutsActive()<<" were active after adding rounds of cuts";
    if (generator->timing())
      std::cout<<" ( "<<generator->timeInCutGenerator()<<" seconds)"<<std::endl;
    else
      std::cout<<std::endl;
  }
  // Print solution if finished - we can't get names from Osi!

  if (model.getMinimizationObjValue()<1.0e50) {
    // post process
    if (preProcess)
      process.postProcess(*model.solver());
    int numberColumns = model.solver()->getNumCols();
    
    const double * solution = model.solver()->getColSolution();
    
    int iColumn;
    std::cout<<std::setiosflags(std::ios::fixed|std::ios::showpoint)<<std::setw(14);
    
    std::cout<<"--------------------------------------"<<std::endl;
    for (iColumn=0;iColumn<numberColumns;iColumn++) {
      double value=solution[iColumn];
      if (fabs(value)>1.0e-7&&model.solver()->isInteger(iColumn)) 
	std::cout<<std::setw(6)<<iColumn<<" "<<value<<std::endl;
    }
    std::cout<<"--------------------------------------"<<std::endl;
  
    std::cout<<std::resetiosflags(std::ios::fixed|std::ios::showpoint|std::ios::scientific);
  }
  return 0;
}