// Restore result
void OsiSolverResult::restoreResult(OsiSolverInterface &solver) const
{
//solver.setObjValue(objectiveValue_)*solver.getObjSense();
solver.setWarmStart(&basis_);
solver.setColSolution(primalSolution_);
solver.setRowPrice(dualSolution_);
fixed_.applyBounds(solver, -1);
}
/** Create a set of candidate branching objects. */
int
BlisBranchStrategyPseudo::createCandBranchObjects(int numPassesLeft,
double ub)
{
int bStatus = 0;
int i, pass, colInd;
int preferDir, saveLimit;
int numFirsts = 0;
int numInfs = 0;
int minCount = 0;
int numLowerTightens = 0;
int numUpperTightens = 0;
double lpX, score, infeasibility, downDeg, upDeg, sumDeg = 0.0;
bool roundAgain, downKeep, downGood, upKeep, upGood;
int *lbInd = NULL;
int *ubInd = NULL;
double *newLB = NULL;
double *newUB = NULL;
double *saveUpper = NULL;
double *saveLower = NULL;
double *saveSolution = NULL;
BlisModel *model = dynamic_cast<BlisModel *>(model_);
OsiSolverInterface *solver = model->solver();
int numCols = model->getNumCols();
int numObjects = model->numObjects();
int aveIterations = model->getAveIterations();
//std::cout << "aveIterations = " << aveIterations << std::endl;
//------------------------------------------------------
// Check if max time is reached or no pass is left.
//------------------------------------------------------
double timeLimit = model->AlpsPar()->entry(AlpsParams::timeLimit);
AlpsKnowledgeBroker *broker = model->getKnowledgeBroker();
bool maxTimeReached = (broker->timer().getTime() > timeLimit);
bool selectNow = false;
if (maxTimeReached || !numPassesLeft) {
selectNow = true;
#ifdef BLIS_DEBUG
printf("PSEUDO: CREATE: maxTimeReached %d, numPassesLeft %d\n",
maxTimeReached, numPassesLeft);
#endif
}
// Store first time objects.
std::vector<BlisObjectInt *> firstObjects;
// Store infeasible objects.
std::vector<BlisObjectInt *> infObjects;
// TODO: check if sorting is expensive.
std::multimap<double, BcpsBranchObject*, BlisPseuoGreater> candObjects;
double objValue = solver->getObjSense() * solver->getObjValue();
const double * lower = solver->getColLower();
const double * upper = solver->getColUpper();
saveSolution = new double[numCols];
memcpy(saveSolution, solver->getColSolution(), numCols*sizeof(double));
//--------------------------------------------------
// Find the infeasible objects.
// NOTE: we might go round this loop twice if we are feed in
// a "feasible" solution.
//--------------------------------------------------
for (pass = 0; pass < 2; ++pass) {
numInfs = 0;
BcpsObject * object = NULL;
BlisObjectInt * intObject = NULL;
infObjects.clear();
firstObjects.clear();
for (i = 0; i < numObjects; ++i) {
object = model->objects(i);
infeasibility = object->infeasibility(model, preferDir);
if (infeasibility) {
++numInfs;
intObject = dynamic_cast<BlisObjectInt *>(object);
if (intObject) {
infObjects.push_back(intObject);
if (!selectNow) {
minCount =
ALPS_MIN(intObject->pseudocost().getDownCount(),
intObject->pseudocost().getUpCount());
if (minCount < 1) {
firstObjects.push_back(intObject);
}
}
#ifdef BLIS_DEBUG
if (intObject->columnIndex() == 40) {
std::cout << "x[40] = " << saveSolution[40]
<< std::endl;
}
#endif
intObject = NULL;
}
else {
// TODO: currently all are integer objects.
#ifdef BLIS_DEBU
assert(0);
#endif
}
}
}
if (numInfs) {
#if 0
std::cout << "PSEUDO: numInfs = " << numInfs
<< std::endl;
#endif
break;
}
else if (pass == 0) {
// The first pass and is IP feasible.
#if 1
std::cout << "ERROR: PSEUDO: given a integer feasible sol, no fraction variable" << std::endl;
assert(0);
#endif
roundAgain = false;
CoinWarmStartBasis * ws =
dynamic_cast<CoinWarmStartBasis*>(solver->getWarmStart());
if (!ws) break;
// Force solution values within bounds
for (i = 0; i < numCols; ++i) {
lpX = saveSolution[i];
if (lpX < lower[i]) {
saveSolution[i] = lower[i];
roundAgain = true;
ws->setStructStatus(i, CoinWarmStartBasis::atLowerBound);
}
else if (lpX > upper[i]) {
saveSolution[i] = upper[i];
roundAgain = true;
ws->setStructStatus(i, CoinWarmStartBasis::atUpperBound);
}
}
if (roundAgain) {
// Need resolve and do the second round selection.
solver->setWarmStart(ws);
delete ws;
// Resolve.
solver->resolve();
if (!solver->isProvenOptimal()) {
// Become infeasible, can do nothing.
bStatus = -2;
goto TERM_CREATE;
}
else {
// Save new lp solution.
memcpy(saveSolution, solver->getColSolution(),
numCols * sizeof(double));
objValue = solver->getObjSense() * solver->getObjValue();
}
}
else {
delete ws;
break;
}
}
} // EOF 2 pass
//--------------------------------------------------
// If we have a set of first time object,
// branch up and down to initialize pseudo-cost.
//--------------------------------------------------
numFirsts = static_cast<int> (firstObjects.size());
//std::cout << "PSEUDO: numFirsts = " << numFirsts << std::endl;
if (numFirsts > 0) {
//std::cout << "PSEUDO: numFirsts = " << numFirsts << std::endl;
//--------------------------------------------------
// Backup solver status and mark hot start.
//--------------------------------------------------
saveLower = new double[numCols];
saveUpper = new double[numCols];
memcpy(saveLower, lower, numCols * sizeof(double));
memcpy(saveUpper, upper, numCols * sizeof(double));
CoinWarmStart * ws = solver->getWarmStart();
solver->getIntParam(OsiMaxNumIterationHotStart, saveLimit);
aveIterations = ALPS_MIN(50, aveIterations);
solver->setIntParam(OsiMaxNumIterationHotStart, aveIterations);
solver->markHotStart();
lbInd = new int [numFirsts];
ubInd = new int [numFirsts];
newLB = new double [numFirsts];
newUB = new double [numFirsts];
for (i = 0; i < numFirsts && bStatus != -2; ++i) {
colInd = firstObjects[i]->columnIndex();
lpX = saveSolution[colInd];
BlisStrongBranch(model, objValue, colInd, lpX,
saveLower, saveUpper,
downKeep, downGood, downDeg,
upKeep, upGood, upDeg);
if(!downKeep && !upKeep) {
// Both branch can be fathomed
bStatus = -2;
}
else if (!downKeep) {
// Down branch can be fathomed.
lbInd[numLowerTightens] = colInd;
newLB[numLowerTightens++] = ceil(lpX);
}
else if (!upKeep) {
// Up branch can be fathomed.
ubInd[numUpperTightens] = colInd;
newUB[numUpperTightens++] = floor(lpX);
}
}
//--------------------------------------------------
// Set new bounds in lp solver for resolving
//--------------------------------------------------
if (bStatus != -2) {
if (numUpperTightens > 0) {
bStatus = -1;
for (i = 0; i < numUpperTightens; ++i) {
solver->setColUpper(ubInd[i], newUB[i]);
}
}
if (numLowerTightens > 0) {
bStatus = -1;
for (i = 0; i < numLowerTightens; ++i) {
solver->setColLower(lbInd[i], newLB[i]);
}
}
}
//--------------------------------------------------
// Unmark hotstart and recover LP solver.
//--------------------------------------------------
solver->unmarkHotStart();
solver->setColSolution(saveSolution);
solver->setIntParam(OsiMaxNumIterationHotStart, saveLimit);
solver->setWarmStart(ws);
delete ws;
}
if (bStatus < 0) {
goto TERM_CREATE;
}
else {
// Create a set of candidate branching objects.
numBranchObjects_ = numInfs;
branchObjects_ = new BcpsBranchObject* [numInfs];
// NOTE: it set model->savedLpSolution.
sumDeg = 0.0;
for (i = 0; i < numInfs; ++i) {
if (infObjects[i]->pseudocost().getUpCost() <
infObjects[i]->pseudocost().getDownCost()) {
preferDir = 1;
}
else {
preferDir = -1;
}
branchObjects_[i] = infObjects[i]->createBranchObject(model,
preferDir);
score = infObjects[i]->pseudocost().getScore();
branchObjects_[i]->setUpScore(score);
sumDeg += score;
#ifdef BLIS_DEBUG_MORE
std::cout << "col[" << infObjects[i]->columnIndex() << "]: score="
<< score << ", dir=" << branchObjects_[i]->getDirection()
<< ", up=" << infObjects[i]->pseudocost().getUpCost()
<< ", down=" << infObjects[i]->pseudocost().getDownCost()
<< std::endl;
#endif
}
model->setSolEstimate(objValue + sumDeg);
}
TERM_CREATE:
//------------------------------------------------------
// Cleanup.
//------------------------------------------------------
delete [] lbInd;
delete [] ubInd;
delete [] newLB;
delete [] newUB;
delete [] saveSolution;
delete [] saveLower;
delete [] saveUpper;
return bStatus;
}
//--------------------------------------------------------------------------
void
CglKnapsackCoverUnitTest(
const OsiSolverInterface * baseSiP,
const std::string mpsDir )
{
int i;
CoinRelFltEq eq(0.000001);
// Test default constructor
{
CglKnapsackCover kccGenerator;
}
// Test copy & assignment
{
CglKnapsackCover rhs;
{
CglKnapsackCover kccGenerator;
CglKnapsackCover cgC(kccGenerator);
rhs=kccGenerator;
}
}
// test exactSolveKnapsack
{
CglKnapsackCover kccg;
const int n=7;
double c=50;
double p[n] = {70,20,39,37,7,5,10};
double w[n] = {31, 10, 20, 19, 4, 3, 6};
double z;
int x[n];
int exactsol = kccg.exactSolveKnapsack(n, c, p, w, z, x);
assert(exactsol==1);
assert (z == 107);
assert (x[0]==1);
assert (x[1]==0);
assert (x[2]==0);
assert (x[3]==1);
assert (x[4]==0);
assert (x[5]==0);
assert (x[6]==0);
}
/*
// Testcase /u/rlh/osl2/mps/scOneInt.mps
// Model has 3 continous, 2 binary, and 1 general
// integer variable.
{
OsiSolverInterface * siP = baseSiP->clone();
int * complement=NULL;
double * xstar=NULL;
siP->readMps("../Mps/scOneInt","mps");
CglKnapsackCover kccg;
int nCols=siP->getNumCols();
// Test the siP methods for detecting
// variable type
int numCont=0, numBinary=0, numIntNonBinary=0, numInt=0;
for (int thisCol=0; thisCol<nCols; thisCol++) {
if ( siP->isContinuous(thisCol) ) numCont++;
if ( siP->isBinary(thisCol) ) numBinary++;
if ( siP->isIntegerNonBinary(thisCol) ) numIntNonBinary++;
if ( siP->isInteger(thisCol) ) numInt++;
}
assert(numCont==3);
assert(numBinary==2);
assert(numIntNonBinary==1);
assert(numInt==3);
// Test initializeCutGenerator
siP->initialSolve();
assert(xstar !=NULL);
for (i=0; i<nCols; i++){
assert(complement[i]==0);
}
int nRows=siP->getNumRows();
for (i=0; i<nRows; i++){
int vectorsize = siP->getMatrixByRow()->vectorSize(i);
assert(vectorsize==2);
}
kccg.cleanUpCutGenerator(complement,xstar);
delete siP;
}
*/
// Testcase /u/rlh/osl2/mps/tp3.mps
// Models has 3 cols, 3 rows
// Row 0 yields a knapsack, others do not.
{
// setup
OsiSolverInterface * siP = baseSiP->clone();
std::string fn(mpsDir+"tp3");
siP->readMps(fn.c_str(),"mps");
// All integer variables should be binary.
// Assert that this is true.
for ( i = 0; i < siP->getNumCols(); i++ )
if ( siP->isInteger(i) )
assert(siP->getColUpper()[i]==1.0 && siP->isBinary(i));
OsiCuts cs;
CoinPackedVector krow;
double b=0;
int nCols=siP->getNumCols();
int * complement=new int [nCols];
double * xstar=new double [nCols];
CglKnapsackCover kccg;
// solve LP relaxation
// a "must" before calling initialization
siP->initialSolve();
double lpRelaxBefore=siP->getObjValue();
std::cout<<"Initial LP value: "<<lpRelaxBefore<<std::endl;
assert( eq(siP->getObjValue(), 97.185) );
double mycs[] = {.627, .667558333333, .038};
siP->setColSolution(mycs);
const double *colsol = siP->getColSolution();
int k;
for (k=0; k<nCols; k++){
xstar[k]=colsol[k];
complement[k]=0;
}
// test deriveAKnapsack
int rind = ( siP->getRowSense()[0] == 'N' ) ? 1 : 0;
const CoinShallowPackedVector reqdBySunCC = siP->getMatrixByRow()->getVector(rind) ;
int deriveaknap = kccg.deriveAKnapsack(*siP, cs, krow,b,complement,xstar,rind,reqdBySunCC);
assert(deriveaknap ==1);
assert(complement[0]==0);
assert(complement[1]==1);
assert(complement[2]==1);
int inx[3] = {0,1,2};
double el[3] = {161, 120, 68};
CoinPackedVector r;
r.setVector(3,inx,el);
assert (krow == r);
//assert (b == 183.0); ????? but x1 and x2 at 1 is valid
// test findGreedyCover
CoinPackedVector cover,remainder;
#if 0
int findgreedy = kccg.findGreedyCover( 0, krow, b, xstar, cover, remainder );
assert( findgreedy == 1 );
int coveri = cover.getNumElements();
assert( cover.getNumElements() == 2);
coveri = cover.getIndices()[0];
assert( cover.getIndices()[0] == 0);
assert( cover.getIndices()[1] == 1);
assert( cover.getElements()[0] == 161.0);
assert( cover.getElements()[1] == 120.0);
assert( remainder.getNumElements() == 1);
assert( remainder.getIndices()[0] == 2);
assert( remainder.getElements()[0] == 68.0);
// test liftCoverCut
CoinPackedVector cut;
double * rowupper = ekk_rowupper(model);
double cutRhs = cover.getNumElements() - 1.0;
kccg.liftCoverCut(b, krow.getNumElements(),
cover, remainder,
cut);
assert ( cut.getNumElements() == 3 );
assert ( cut.getIndices()[0] == 0 );
assert ( cut.getIndices()[1] == 1 );
assert ( cut.getIndices()[2] == 2 );
assert( cut.getElements()[0] == 1 );
assert( cut.getElements()[1] == 1 );
assert( eq(cut.getElements()[2], 0.087719) );
// test liftAndUncomplementAndAdd
OsiCuts cuts;
kccg.liftAndUncomplementAndAdd(*siP.getRowUpper()[0],krow,b,complement,0,
cover,remainder,cuts);
int sizerowcuts = cuts.sizeRowCuts();
assert ( sizerowcuts== 1 );
OsiRowCut testRowCut = cuts.rowCut(0);
CoinPackedVector testRowPV = testRowCut.row();
OsiRowCut sampleRowCut;
const int sampleSize = 3;
int sampleCols[sampleSize]={0,1,2};
double sampleElems[sampleSize]={1.0,-1.0,-0.087719};
sampleRowCut.setRow(sampleSize,sampleCols,sampleElems);
sampleRowCut.setLb(-DBL_MAX);
sampleRowCut.setUb(-0.087719);
bool equiv = testRowPV.equivalent(sampleRowCut.row(),CoinRelFltEq(1.0e-05) );
assert ( equiv );
#endif
// test find PseudoJohnAndEllisCover
cover.setVector(0,NULL, NULL);
remainder.setVector(0,NULL,NULL);
rind = ( siP->getRowSense()[0] == 'N' ) ? 1 : 0;
int findPJE = kccg.findPseudoJohnAndEllisCover( rind, krow,
b, xstar, cover, remainder );
assert( findPJE == 1 );
assert ( cover.getIndices()[0] == 0 );
assert ( cover.getIndices()[1] == 2 );
assert ( cover.getElements()[0] == 161 );
assert ( cover.getElements()[1] == 68 );
assert ( remainder.getIndices()[0] == 1 );
assert ( remainder.getElements()[0] == 120 );
OsiCuts cuts;
kccg.liftAndUncomplementAndAdd((*siP).getRowUpper()[rind],krow,b, complement, rind,
cover,remainder,cuts);
assert (cuts.sizeRowCuts() == 1 );
OsiRowCut testRowCut = cuts.rowCut(0);
CoinPackedVector testRowPV = testRowCut.row();
const int sampleSize = 3;
int sampleCols[sampleSize]={0,1,2};
double sampleElems[sampleSize]={1.0, -1.0, -1.0};
OsiRowCut sampleRowCut;
sampleRowCut.setRow(sampleSize,sampleCols,sampleElems);
sampleRowCut.setLb(-COIN_DBL_MAX);
sampleRowCut.setUb(-1.0);
// test for 'close enough'
assert( testRowPV.isEquivalent(sampleRowCut.row(),CoinRelFltEq(1.0e-05) ) );
// Reset complement & test next row
for (i=0; i<nCols; i++){
complement[i]=0;
}
rind++;
const CoinShallowPackedVector reqdBySunCC2 = siP->getMatrixByRow()->getVector(rind) ;
deriveaknap = kccg.deriveAKnapsack(*siP,cuts,krow,b,complement,xstar,rind,reqdBySunCC2);
assert(deriveaknap==0);
// Reset complement & test next row
for (i=0; i<nCols; i++){
complement[i]=0;
}
const CoinShallowPackedVector reqdBySunCC3 = siP->getMatrixByRow()->getVector(2) ;
deriveaknap = kccg.deriveAKnapsack(*siP,cuts,krow,b,complement,xstar,2,
reqdBySunCC3);
assert(deriveaknap == 0);
// Clean up
delete [] complement;
delete [] xstar;
delete siP;
}
#if 0
// Testcase /u/rlh/osl2/mps/tp4.mps
// Models has 6 cols, 1 knapsack row and
// 3 rows explicily bounding variables
// Row 0 yields a knapsack cover cut
// using findGreedyCover which moves the
// LP objective function value.
{
// Setup
EKKContext * env=ekk_initializeContext();
EKKModel * model = ekk_newModel(env,"");
OsiSolverInterface si(model);
ekk_importModel(model, "tp4.mps");
CglKnapsackCover kccg;
kccg.ekk_validateIntType(si);
// Solve the LP relaxation of the model and
// print out ofv for sake of comparison
ekk_allSlackBasis(model);
ekk_crash(model,1);
ekk_primalSimplex(model,1);
double lpRelaxBefore=ekk_getRobjvalue(model);
#ifdef CGL_DEBUG
printf("\n\nOrig LP min=%f\n",lpRelaxBefore);
#endif
// Determine if lp sol is ip optimal
// Note: no ekk_function to do this
int nCols=ekk_getInumcols(model);
double * optLpSol = ekk_colsol(model);
int ipOpt = 1;
i=0;
while (i++<nCols && ipOpt){
if(optLpSol[i] < 1.0-1.0e-08 && optLpSol[i]> 1.0e-08) ipOpt = 0;
}
if (ipOpt){
#ifdef CGL_DEBUG
printf("Lp solution is within ip optimality tolerance\n");
#endif
}
else {
OsiSolverInterface iModel(model);
OsiCuts cuts;
// Test generateCuts method
kccg.generateCuts(iModel,cuts);
OsiSolverInterface::ApplyCutsReturnCode rc = iModel.applyCuts(cuts);
ekk_mergeBlocks(model,1);
ekk_dualSimplex(model);
double lpRelaxAfter=ekk_getRobjvalue(model);
#ifdef CGL_DEBUG
printf("\n\nFinal LP min=%f\n",lpRelaxAfter);
#endif
assert( lpRelaxBefore < lpRelaxAfter );
// This may need to be updated as other
// minimal cover finders are added
assert( cuts.sizeRowCuts() == 1 );
OsiRowCut testRowCut = cuts.rowCut(0);
CoinPackedVector testRowPV = testRowCut.row();
OsiRowCut sampleRowCut;
const int sampleSize = 6;
int sampleCols[sampleSize]={0,1,2,3,4,5};
double sampleElems[sampleSize]={1.0,1.0,1.0,1.0,0.5, 2.0};
sampleRowCut.setRow(sampleSize,sampleCols,sampleElems);
sampleRowCut.setLb(-DBL_MAX);
sampleRowCut.setUb(3.0);
bool equiv = testRowPV.equivalent(sampleRowCut.row(),CoinRelFltEq(1.0e-05) );
assert( testRowPV.equivalent(sampleRowCut.row(),CoinRelFltEq(1.0e-05) ) );
}
// Exit out of OSL
ekk_deleteModel(model);
ekk_endContext(env);
}
#endif
// Testcase /u/rlh/osl2/mps/tp5.mps
// Models has 6 cols, 1 knapsack row and
// 3 rows explicily bounding variables
// Row 0 yields a knapsack cover cut
// using findGreedyCover which moves the
// LP objective function value.
{
// Setup
OsiSolverInterface * siP = baseSiP->clone();
std::string fn(mpsDir+"tp5");
siP->readMps(fn.c_str(),"mps");
// All integer variables should be binary.
// Assert that this is true.
for ( i = 0; i < siP->getNumCols(); i++ )
if ( siP->isInteger(i) )
assert(siP->getColUpper()[i]==1.0 && siP->isBinary(i));
CglKnapsackCover kccg;
// Solve the LP relaxation of the model and
// print out ofv for sake of comparison
siP->initialSolve();
double lpRelaxBefore=siP->getObjValue();
assert( eq(lpRelaxBefore, -51.66666666667) );
double mycs[] = {.8999999999, .899999999999, .89999999999, 1.110223e-16, .5166666666667, 0};
siP->setColSolution(mycs);
#ifdef CGL_DEBUG
printf("\n\nOrig LP min=%f\n",lpRelaxBefore);
#endif
// Determine if lp sol is 0/1 optimal
int nCols=siP->getNumCols();
const double * optLpSol = siP->getColSolution();
bool ipOpt = true;
i=0;
while (i++<nCols && ipOpt){
if(optLpSol[i] > kccg.epsilon_ && optLpSol[i] < kccg.onetol_) ipOpt = false;
}
if (ipOpt){
#ifdef CGL_DEBUG
printf("Lp solution is within ip optimality tolerance\n");
#endif
}
else {
// set up
OsiCuts cuts;
CoinPackedVector krow;
double b=0.0;
int * complement=new int[nCols];
double * xstar=new double[nCols];
// initialize cut generator
const double *colsol = siP->getColSolution();
for (i=0; i<nCols; i++){
xstar[i]=colsol[i];
complement[i]=0;
}
int row = ( siP->getRowSense()[0] == 'N' ) ? 1 : 0;
// transform row into canonical knapsack form
const CoinShallowPackedVector reqdBySunCC = siP->getMatrixByRow()->getVector(row) ;
if (kccg.deriveAKnapsack(*siP, cuts, krow, b, complement, xstar, row,reqdBySunCC)){
CoinPackedVector cover, remainder;
// apply greedy logic to detect violated minimal cover inequalities
if (kccg.findGreedyCover(row, krow, b, xstar, cover, remainder) == 1){
// lift, uncomplements, and add cut to cut set
kccg.liftAndUncomplementAndAdd((*siP).getRowUpper()[row],krow, b, complement, row, cover, remainder, cuts);
}
// reset optimal column solution (xstar) information in OSL
const double * rowupper = siP->getRowUpper();
int k;
if (fabs(b-rowupper[row]) > 1.0e-05) {
for(k=0; k<krow.getNumElements(); k++) {
if (complement[krow.getIndices()[k]]){
xstar[krow.getIndices()[k]]= 1.0-xstar[krow.getIndices()[k]];
complement[krow.getIndices()[k]]=0;
}
}
}
// clean up
delete [] complement;
delete [] xstar;
}
// apply the cuts
OsiSolverInterface::ApplyCutsReturnCode rc = siP->applyCuts(cuts);
siP->resolve();
double lpRelaxAfter=siP->getObjValue();
assert( eq(lpRelaxAfter, -30.0) );
#ifdef CGL_DEBUG
printf("\n\nFinal LP min=%f\n",lpRelaxAfter);
#endif
// test that expected cut was detected
assert( lpRelaxBefore < lpRelaxAfter );
assert( cuts.sizeRowCuts() == 1 );
OsiRowCut testRowCut = cuts.rowCut(0);
CoinPackedVector testRowPV = testRowCut.row();
OsiRowCut sampleRowCut;
const int sampleSize = 6;
int sampleCols[sampleSize]={0,1,2,3,4,5};
double sampleElems[sampleSize]={1.0,1.0,1.0,0.25,1.0,2.0};
sampleRowCut.setRow(sampleSize,sampleCols,sampleElems);
sampleRowCut.setLb(-COIN_DBL_MAX);
sampleRowCut.setUb(3.0);
assert(testRowPV.isEquivalent(sampleRowCut.row(),CoinRelFltEq(1.0e-05)));
}
delete siP;
}
// Testcase /u/rlh/osl2/mps/p0033
// Miplib3 problem p0033
// Test that no cuts chop off the optimal solution
{
// Setup
OsiSolverInterface * siP = baseSiP->clone();
std::string fn(mpsDir+"p0033");
siP->readMps(fn.c_str(),"mps");
// All integer variables should be binary.
// Assert that this is true.
for ( i = 0; i < siP->getNumCols(); i++ )
if ( siP->isInteger(i) )
assert(siP->getColUpper()[i]==1.0 && siP->isBinary(i));
int nCols=siP->getNumCols();
CglKnapsackCover kccg;
// Solve the LP relaxation of the model and
// print out ofv for sake of comparison
siP->initialSolve();
double lpRelaxBefore=siP->getObjValue();
assert( eq(lpRelaxBefore, 2520.5717391304347) );
double mycs[] = {0, 1, 0, 0, -2.0837010502455788e-19, 1, 0, 0, 1,
0.021739130434782594, 0.35652173913043478,
-6.7220534694101275e-18, 5.3125906451789717e-18,
1, 0, 1.9298798670241979e-17, 0, 0, 0,
7.8875708048320448e-18, 0.5, 0,
0.85999999999999999, 1, 1, 0.57999999999999996,
1, 0, 1, 0, 0.25, 0, 0.67500000000000004};
siP->setColSolution(mycs);
#ifdef CGL_DEBUG
printf("\n\nOrig LP min=%f\n",lpRelaxBefore);
#endif
OsiCuts cuts;
// Test generateCuts method
kccg.generateCuts(*siP,cuts);
OsiSolverInterface::ApplyCutsReturnCode rc = siP->applyCuts(cuts);
siP->resolve();
double lpRelaxAfter=siP->getObjValue();
assert( eq(lpRelaxAfter, 2829.0597826086955) );
#ifdef CGL_DEBUG
printf("\n\nOrig LP min=%f\n",lpRelaxBefore);
printf("\n\nFinal LP min=%f\n",lpRelaxAfter);
#endif
assert( lpRelaxBefore < lpRelaxAfter );
// the CoinPackedVector p0033 is the optimal
// IP solution to the miplib problem p0033
int objIndices[14] = {
0, 6, 7, 9, 13, 17, 18,
22, 24, 25, 26, 27, 28, 29 };
CoinPackedVector p0033(14,objIndices,1.0);
// Sanity check
const double * objective=siP->getObjCoefficients();
double ofv =0 ;
int r;
for (r=0; r<nCols; r++){
ofv=ofv + p0033[r]*objective[r];
}
CoinRelFltEq eq;
assert( eq(ofv,3089.0) );
int nRowCuts = cuts.sizeRowCuts();
OsiRowCut rcut;
CoinPackedVector rpv;
for (i=0; i<nRowCuts; i++){
rcut = cuts.rowCut(i);
rpv = rcut.row();
double p0033Sum = (rpv*p0033).sum();
assert (p0033Sum <= rcut.ub() );
}
delete siP;
}
// if a debug file is there then look at it
{
FILE * fp = fopen("knapsack.debug","r");
if (fp) {
int ncol,nel;
double up;
int x = fscanf(fp,"%d %d %lg",&ncol,&nel,&up);
if (x<=0)
throw("bad fscanf");
printf("%d columns, %d elements, upper %g\n",ncol,nel,up);
double * sol1 = new double[nel];
double * el1 = new double[nel];
int * col1 = new int[nel];
CoinBigIndex * start = new CoinBigIndex [ncol+1];
memset(start,0,ncol*sizeof(CoinBigIndex ));
int * row = new int[nel];
int i;
for (i=0;i<nel;i++) {
x=fscanf(fp,"%d %lg %lg",col1+i,el1+i,sol1+i);
if (x<=0)
throw("bad fscanf");
printf("[%d, e=%g, v=%g] ",col1[i],el1[i],sol1[i]);
start[col1[i]]=1;
row[i]=0;
}
printf("\n");
// Setup
OsiSolverInterface * siP = baseSiP->clone();
double lo=-1.0e30;
double * upper = new double[ncol];
start[ncol]=nel;
int last=0;
for (i=0;i<ncol;i++) {
upper[i]=1.0;
int marked=start[i];
start[i]=last;
if (marked)
last++;
}
siP->loadProblem(ncol,1,start,row,el1,NULL,upper,NULL,&lo,&up);
// use upper for solution
memset(upper,0,ncol*sizeof(double));
for (i=0;i<nel;i++) {
int icol=col1[i];
upper[icol]=sol1[i];
siP->setInteger(icol);
}
siP->setColSolution(upper);
delete [] sol1;
delete [] el1;
delete [] col1;
delete [] start;
delete [] row;
delete [] upper;
CglKnapsackCover kccg;
OsiCuts cuts;
// Test generateCuts method
kccg.generateCuts(*siP,cuts);
// print out and compare to known cuts
int numberCuts = cuts.sizeRowCuts();
if (numberCuts) {
for (i=0;i<numberCuts;i++) {
OsiRowCut * thisCut = cuts.rowCutPtr(i);
int n=thisCut->row().getNumElements();
printf("Cut %d has %d entries, rhs %g %g =>",i,n,thisCut->lb(),
thisCut->ub());
int j;
const int * index = thisCut->row().getIndices();
const double * element = thisCut->row().getElements();
for (j=0;j<n;j++) {
printf(" (%d,%g)",index[j],element[j]);
}
printf("\n");
}
}
fclose(fp);
}
}
// Testcase /u/rlh/osl2/mps/p0201
// Miplib3 problem p0282
// Test that no cuts chop off the optimal ip solution
{
// Setup
OsiSolverInterface * siP = baseSiP->clone();
std::string fn(mpsDir+"p0201");
siP->readMps(fn.c_str(),"mps");
// All integer variables should be binary.
// Assert that this is true.
for ( i = 0; i < siP->getNumCols(); i++ )
if ( siP->isInteger(i) )
assert(siP->getColUpper()[i]==1.0 && siP->isBinary(i));
const int nCols=siP->getNumCols();
CglKnapsackCover kccg;
// Solve the LP relaxation of the model and
// print out ofv for sake of comparisn
siP->initialSolve();
double lpRelaxBefore=siP->getObjValue();
assert( eq(lpRelaxBefore, 6875.) );
double mycs[] =
{0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0.5, 0, 0, 0, 0.5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0.5,
0, 0, 0, 0.5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0.5, 0, 0,
0, 0.5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0.5, 0, 0, 0, 0.5,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0.5, 0, 0, 0, 0.5, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0.5, 0, 0, 0, 0.5, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 1, 0.5, 0, 0, 0, 0.5, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 1, 0.5, 0, 0, 0, 0.5, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 1, 0.5, 0, 0, 0, 0.5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 1, 0.5, 0, 0, 0, 0.5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
1};
siP->setColSolution(mycs);
#ifdef CGL_DEBUG
printf("\n\nOrig LP min=%f\n",lpRelaxBefore);
#endif
OsiCuts cuts;
// Test generateCuts method
kccg.generateCuts(*siP,cuts);
OsiSolverInterface::ApplyCutsReturnCode rc = siP->applyCuts(cuts);
siP->resolve();
double lpRelaxAfter=siP->getObjValue();
assert( eq(lpRelaxAfter, 7125) );
#ifdef CGL_DEBUG
printf("\n\nOrig LP min=%f\n",lpRelaxBefore);
printf("\n\nFinal LP min=%f\n",lpRelaxAfter);
#endif
assert( lpRelaxBefore < lpRelaxAfter );
// Optimal IP solution to p0201
int objIndices[22] = { 8, 10, 21, 38, 39, 56,
60, 74, 79, 92, 94, 110, 111, 128, 132, 146,
151,164, 166, 182,183, 200 };
CoinPackedVector p0201(22,objIndices,1.0);
// Sanity check
const double * objective=siP->getObjCoefficients();
double ofv =0 ;
int r;
for (r=0; r<nCols; r++){
ofv=ofv + p0201[r]*objective[r];
}
CoinRelFltEq eq;
assert( eq(ofv,7615.0) );
//printf("p0201 optimal ofv = %g\n",ofv);
int nRowCuts = cuts.sizeRowCuts();
OsiRowCut rcut;
CoinPackedVector rpv;
for (i=0; i<nRowCuts; i++){
rcut = cuts.rowCut(i);
rpv = rcut.row();
double p0201Sum = (rpv*p0201).sum();
assert (p0201Sum <= rcut.ub() );
}
delete siP;
}
// see if I get the same covers that N&W get
{
OsiSolverInterface * siP=baseSiP->clone();
std::string fn(mpsDir+"nw460");
siP->readMps(fn.c_str(),"mps");
// All integer variables should be binary.
// Assert that this is true.
for ( i = 0; i < siP->getNumCols(); i++ )
if ( siP->isInteger(i) )
assert(siP->getColUpper()[i]==1.0 && siP->isBinary(i));
CglKnapsackCover kccg;
// Solve the LP relaxation of the model and
// print out ofv for sake of comparison
siP->initialSolve();
double lpRelaxBefore=siP->getObjValue();
assert( eq(lpRelaxBefore, -225.68951787852194) );
double mycs[] = {0.7099213482046447, 0, 0.34185802225477174, 1, 1, 0, 1, 1, 0};
siP->setColSolution(mycs);
OsiCuts cuts;
// Test generateCuts method
kccg.generateCuts(*siP,cuts);
OsiSolverInterface::ApplyCutsReturnCode rc = siP->applyCuts(cuts);
siP->resolve();
double lpRelaxAfter=siP->getObjValue();
assert( eq(lpRelaxAfter, -176) );
#ifdef CGL_DEBUG
printf("\n\nOrig LP min=%f\n",lpRelaxBefore);
printf("\n\nFinal LP min=%f\n",lpRelaxAfter);
#endif
#ifdef MJS
assert( lpRelaxBefore < lpRelaxAfter );
#endif
int nRowCuts = cuts.sizeRowCuts();
OsiRowCut rcut;
CoinPackedVector rpv;
for (i=0; i<nRowCuts; i++){
rcut = cuts.rowCut(i);
rpv = rcut.row();
int j;
printf("Row cut number %i has rhs = %g\n",i,rcut.ub());
for (j=0; j<rpv.getNumElements(); j++){
printf("index %i, element %g\n", rpv.getIndices()[j], rpv.getElements()[j]);
}
printf("\n");
}
delete siP;
}
// Debugging: try "exmip1.mps"
{
// Setup
OsiSolverInterface * siP = baseSiP->clone();
std::string fn(mpsDir+"exmip1");
siP->readMps(fn.c_str(),"mps");
// All integer variables should be binary.
// Assert that this is true.
for ( i = 0; i < siP->getNumCols(); i++ )
if ( siP->isInteger(i) )
assert(siP->getColUpper()[i]==1.0 && siP->isBinary(i));
CglKnapsackCover kccg;
// Solve the LP relaxation of the model and
// print out ofv for sake of comparison
siP->initialSolve();
double lpRelaxBefore=siP->getObjValue();
assert( eq(lpRelaxBefore, 3.2368421052631575) );
double mycs[] = {2.5, 0, 0, 0.6428571428571429, 0.5, 4, 0, 0.26315789473684253};
siP->setColSolution(mycs);
// Test generateCuts method
OsiCuts cuts;
kccg.generateCuts(*siP,cuts);
OsiSolverInterface::ApplyCutsReturnCode rc = siP->applyCuts(cuts);
siP->resolve();
double lpRelaxAfter=siP->getObjValue();
assert( eq(lpRelaxAfter, 3.2368421052631575) );
#ifdef CGL_DEBUG
printf("\n\nOrig LP min=%f\n",lpRelaxBefore);
printf("\n\nFinal LP min=%f\n",lpRelaxAfter);
#endif
assert( lpRelaxBefore <= lpRelaxAfter );
delete siP;
}
#ifdef CGL_DEBUG
// See what findLPMostViolatedMinCover for knapsack with 2 elements does
{
int nCols = 2;
int row = 1;
CoinPackedVector krow;
double e[2] = {5,10};
int ii[2] = {0,1};
krow.setVector(nCols,ii,e);
double b=11;
double xstar[2] = {.2,.9};
CoinPackedVector cover;
CoinPackedVector remainder;
CglKnapsackCover kccg;
kccg.findLPMostViolatedMinCover(nCols, row, krow, b, xstar, cover, remainder);
printf("num in cover = %i\n",cover.getNumElements());
int j;
for (j=0; j<cover.getNumElements(); j++){
printf(" index %i element % g\n", cover.getIndices()[j], cover.getElements()[j]);
}
}
#endif
#ifdef CGL_DEBUG
// see what findLPMostViolatedMinCover does
{
int nCols = 5;
int row = 1;
CoinPackedVector krow;
double e[5] = {1,1,1,1,10};
int ii[5] = {0,1,2,3,4};
krow.setVector(nCols,ii,e);
double b=11;
double xstar[5] = {.9,.9,1,1,.1};
CoinPackedVector cover;
CoinPackedVector remainder;
CglKnapsackCover kccg;
kccg.findLPMostViolatedMinCover(nCols, row, krow, b, xstar, cover, remainder);
printf("num in cover = %i\n",cover.getNumElements());
int j;
for (j=0; j<cover.getNumElements(); j++){
printf(" index %i element % g\n", cover.getIndices()[j], cover.getElements()[j]);
}
}
#endif
}
/** Create a set of candidate branching objects. */
int
BlisBranchStrategyRel::createCandBranchObjects(int numPassesLeft)
{
int bStatus = 0;
int i, pass, colInd;
int preferDir, saveLimit;
int numFirsts = 0;
int numInfs = 0;
int minCount = 0;
int numLowerTightens = 0;
int numUpperTightens = 0;
double lpX, score, infeasibility, downDeg, upDeg, sumDeg = 0.0;
bool roundAgain, downKeep, downGood, upKeep, upGood;
int *lbInd = NULL;
int *ubInd = NULL;
double *newLB = NULL;
double *newUB = NULL;
double * saveUpper = NULL;
double * saveLower = NULL;
double * saveSolution = NULL;
BlisModel *model = dynamic_cast<BlisModel *>(model_);
OsiSolverInterface * solver = model->solver();
int numCols = model->getNumCols();
int numObjects = model->numObjects();
//int lookAhead = dynamic_cast<BlisParams*>
// (model->blisPar())->entry(BlisParams::lookAhead);
//------------------------------------------------------
// Check if max time is reached or no pass is left.
//------------------------------------------------------
double timeLimit = model->AlpsPar()->entry(AlpsParams::timeLimit);
bool maxTimeReached = (CoinCpuTime() - model->startTime_ > timeLimit);
bool selectNow = false;
if (maxTimeReached || !numPassesLeft) {
selectNow = true;
#ifdef BLIS_DEBUG
printf("REL: CREATE: maxTimeReached %d, numPassesLeft %d\n",
maxTimeReached, numPassesLeft);
#endif
}
// Store first time objects.
std::vector<BlisObjectInt *> firstObjects;
// Store infeasible objects.
std::vector<BlisObjectInt *> infObjects;
// TODO: check if sorting is expensive.
std::multimap<double, BlisObjectInt*, BlisPseuoGreater> sortedObjects;
double objValue = solver->getObjSense() * solver->getObjValue();
const double * lower = solver->getColLower();
const double * upper = solver->getColUpper();
int lookAhead = dynamic_cast<BlisParams*>
(model->BlisPar())->entry(BlisParams::lookAhead);
BlisObjectInt * intObject = NULL;
//------------------------------------------------------
// Backup solver status and mark hot start.
//-----------------------------------------------------
saveSolution = new double[numCols];
memcpy(saveSolution, solver->getColSolution(), numCols*sizeof(double));
saveLower = new double[numCols];
saveUpper = new double[numCols];
memcpy(saveLower, lower, numCols * sizeof(double));
memcpy(saveUpper, upper, numCols * sizeof(double));
//------------------------------------------------------
// Find the infeasible objects.
// NOTE: we might go round this loop twice if we are feed in
// a "feasible" solution.
//------------------------------------------------------
for (pass = 0; pass < 2; ++pass) {
numInfs = 0;
BcpsObject * object = NULL;
infObjects.clear();
firstObjects.clear();
for (i = 0; i < numObjects; ++i) {
object = model->objects(i);
infeasibility = object->infeasibility(model, preferDir);
if (infeasibility) {
++numInfs;
intObject = dynamic_cast<BlisObjectInt *>(object);
if (intObject) {
//score = object->pseudocost().getScore();
//tempBO = object->createBranchObject(model, preferDir);
//candObjects.insert(std::make_pair(score, tempBO));
//tempBO = NULL;
infObjects.push_back(intObject);
if (!selectNow) {
minCount =
ALPS_MIN(intObject->pseudocost().getDownCount(),
intObject->pseudocost().getUpCount());
if (minCount < 1) {
firstObjects.push_back(intObject);
}
}
#ifdef BLIS_DEBUG_MORE
if (intObject->columnIndex() == 15) {
std::cout << "x[15] = " << saveSolution[15]
<< std::endl;
}
#endif
intObject = NULL;
}
else {
// TODO: currently all are integer objects.
#ifdef BLIS_DEBU
assert(0);
#endif
}
}
}
if (numInfs) {
#ifdef BLIS_DEBUG_MORE
std::cout << "REL: numInfs = " << numInfs
<< std::endl;
#endif
break;
}
else if (pass == 0) {
// The first pass and is IP feasible.
#ifdef BLIS_DEBUG
std::cout << "REL: given a feasible sol" << std::endl;
#endif
roundAgain = false;
CoinWarmStartBasis * ws =
dynamic_cast<CoinWarmStartBasis*>(solver->getWarmStart());
if (!ws) break;
// Force solution values within bounds
for (i = 0; i < numCols; ++i) {
lpX = saveSolution[i];
if (lpX < lower[i]) {
saveSolution[i] = lower[i];
roundAgain = true;
ws->setStructStatus(i, CoinWarmStartBasis::atLowerBound);
}
else if (lpX > upper[i]) {
saveSolution[i] = upper[i];
roundAgain = true;
ws->setStructStatus(i, CoinWarmStartBasis::atUpperBound);
}
}
if (roundAgain) {
// Need resolve and do the second round selection.
solver->setWarmStart(ws);
delete ws;
// Resolve.
solver->resolve();
if (!solver->isProvenOptimal()) {
// Become infeasible, can do nothing.
bStatus = -2;
goto TERM_CREATE;
}
else {
// Save new lp solution.
memcpy(saveSolution, solver->getColSolution(),
numCols * sizeof(double));
objValue = solver->getObjSense() * solver->getObjValue();
}
}
else {
delete ws;
break;
}
}
} // EOF 2 pass
//--------------------------------------------------
// If we have a set of first time object,
// branch up and down to initialize pseudo-cost.
//--------------------------------------------------
numFirsts = static_cast<int> (firstObjects.size());
if (numFirsts > 0) {
CoinWarmStart * ws = solver->getWarmStart();
solver->getIntParam(OsiMaxNumIterationHotStart, saveLimit);
int maxIter = ALPS_MAX(model->getAveIterations(), 50);
solver->setIntParam(OsiMaxNumIterationHotStart, maxIter);
solver->markHotStart();
lbInd = new int [numFirsts];
ubInd = new int [numFirsts];
newLB = new double [numFirsts];
newUB = new double [numFirsts];
for (i = 0; i < numFirsts && bStatus != -2; ++i) {
colInd = firstObjects[i]->columnIndex();
lpX = saveSolution[colInd];
BlisStrongBranch(model, objValue, colInd, lpX,
saveLower, saveUpper,
downKeep, downGood, downDeg,
upKeep, upGood, upDeg);
if(!downKeep && !upKeep) {
// Both branch can be fathomed
bStatus = -2;
}
else if (!downKeep) {
// Down branch can be fathomed.
lbInd[numLowerTightens] = colInd;
newLB[numLowerTightens++] = ceil(lpX);
//break;
}
else if (!upKeep) {
// Up branch can be fathomed.
ubInd[numUpperTightens] = colInd;
newUB[numUpperTightens++] = floor(lpX);
// break;
}
// Update pseudocost.
if(downGood) {
firstObjects[i]->pseudocost().update(-1, downDeg, lpX);
}
if(downGood) {
firstObjects[i]->pseudocost().update(1, upDeg, lpX);
}
}
//--------------------------------------------------
// Set new bounds in lp solver for resolving
//--------------------------------------------------
if (bStatus != -2) {
if (numUpperTightens > 0) {
bStatus = -1;
for (i = 0; i < numUpperTightens; ++i) {
solver->setColUpper(ubInd[i], newUB[i]);
}
}
if (numLowerTightens > 0) {
bStatus = -1;
for (i = 0; i < numLowerTightens; ++i) {
solver->setColLower(lbInd[i], newLB[i]);
}
}
}
//--------------------------------------------------
// Unmark hotstart and recover LP solver.
//--------------------------------------------------
solver->unmarkHotStart();
solver->setColSolution(saveSolution);
solver->setIntParam(OsiMaxNumIterationHotStart, saveLimit);
solver->setWarmStart(ws);
delete ws;
}
//std::cout << "REL: bStatus = " << bStatus << std::endl;
if (bStatus < 0) {
// Infeasible or monotone.
goto TERM_CREATE;
}
else {
// All object's pseudocost have been initialized.
// Sort them, and do strong branch for the unreliable one
// NOTE: it set model->savedLpSolution.
// model->feasibleSolution(numIntegerInfs, numObjectInfs);
sumDeg = 0.0;
for (i = 0; i < numInfs; ++i) {
score = infObjects[i]->pseudocost().getScore();
sumDeg += score;
std::pair<const double, BlisObjectInt*> sa(score, infObjects[i]);
sortedObjects.insert(sa);
#ifdef BLIS_DEBUG_MORE
std::cout << "col[" << infObjects[i]->columnIndex() << "]="
<< score << ", "<< std::endl;
#endif
}
int numNotChange = 0;
std::multimap< double, BlisObjectInt*, BlisPseuoGreater >::iterator pos;
CoinWarmStart * ws = solver->getWarmStart();
solver->getIntParam(OsiMaxNumIterationHotStart, saveLimit);
int maxIter = ALPS_MAX(model->getAveIterations(), 50);
solver->setIntParam(OsiMaxNumIterationHotStart, maxIter);
solver->markHotStart();
BlisObjectInt *bestObject = NULL;
double bestScore = -10.0;
for (pos = sortedObjects.begin(); pos != sortedObjects.end(); ++pos) {
intObject = pos->second;
colInd = intObject->columnIndex();
#ifdef BLIS_DEBUG_MORE
std::cout << "col[" << colInd << "]: "
<< "score=" << pos->first
<< ", upCount=" << intObject->pseudocost().getUpCount()
<<", downCount="<< intObject->pseudocost().getDownCount()
<< std::endl;
#endif
// Check if reliable.
int objRelibility=ALPS_MIN(intObject->pseudocost().getUpCount(),
intObject->pseudocost().getDownCount());
if (objRelibility < relibility_) {
// Unrelible object. Do strong branching.
lpX = saveSolution[colInd];
BlisStrongBranch(model, objValue, colInd, lpX,
saveLower, saveUpper,
downKeep, downGood, downDeg,
upKeep, upGood, upDeg);
// Update pseudocost.
if(downGood) {
intObject->pseudocost().update(-1, downDeg, lpX);
}
if(downGood) {
intObject->pseudocost().update(1, upDeg, lpX);
}
}
// Compare with the best.
if (intObject->pseudocost().getScore() > bestScore) {
bestScore = intObject->pseudocost().getScore();
bestObject = intObject;
// Reset
numNotChange = 0;
}
else {
// If best doesn't change for "lookAhead" comparisons, then
// the best is reliable.
if (++numNotChange > lookAhead) {
if (bestObject->pseudocost().getUpCost() >
bestObject->pseudocost().getDownCost()) {
preferDir = 1;
}
else {
preferDir = -1;
}
break;
}
}
}
solver->unmarkHotStart();
solver->setColSolution(saveSolution);
solver->setIntParam(OsiMaxNumIterationHotStart, saveLimit);
solver->setWarmStart(ws);
delete ws;
model->setSolEstimate(objValue + sumDeg);
assert(bestObject != NULL);
bestBranchObject_ = bestObject->createBranchObject(model, preferDir);
}
TERM_CREATE:
//------------------------------------------------------
// Cleanup.
//------------------------------------------------------
delete [] lbInd;
delete [] ubInd;
delete [] newLB;
delete [] newUB;
delete [] saveSolution;
delete [] saveLower;
delete [] saveUpper;
return bStatus;
}
int
CbcHeuristicNaive::solution(double & solutionValue,
double * betterSolution)
{
numCouldRun_++;
// See if to do
bool atRoot = model_->getNodeCount() == 0;
int passNumber = model_->getCurrentPassNumber();
if (!when() || (when() == 1 && model_->phase() != 1) || !atRoot || passNumber != 1)
return 0; // switched off
// Don't do if it was this heuristic which found solution!
if (this == model_->lastHeuristic())
return 0;
numRuns_++;
double cutoff;
model_->solver()->getDblParam(OsiDualObjectiveLimit, cutoff);
double direction = model_->solver()->getObjSense();
cutoff *= direction;
cutoff = CoinMin(cutoff, solutionValue);
OsiSolverInterface * solver = model_->continuousSolver();
if (!solver)
solver = model_->solver();
const double * colLower = solver->getColLower();
const double * colUpper = solver->getColUpper();
const double * objective = solver->getObjCoefficients();
int numberColumns = model_->getNumCols();
int numberIntegers = model_->numberIntegers();
const int * integerVariable = model_->integerVariable();
int i;
bool solutionFound = false;
CoinWarmStartBasis saveBasis;
CoinWarmStartBasis * basis =
dynamic_cast<CoinWarmStartBasis *>(solver->getWarmStart()) ;
if (basis) {
saveBasis = * basis;
delete basis;
}
// First just fix all integers as close to zero as possible
OsiSolverInterface * newSolver = cloneBut(7); // wassolver->clone();
for (i = 0; i < numberIntegers; i++) {
int iColumn = integerVariable[i];
double lower = colLower[iColumn];
double upper = colUpper[iColumn];
double value;
if (lower > 0.0)
value = lower;
else if (upper < 0.0)
value = upper;
else
value = 0.0;
newSolver->setColLower(iColumn, value);
newSolver->setColUpper(iColumn, value);
}
newSolver->initialSolve();
if (newSolver->isProvenOptimal()) {
double solValue = newSolver->getObjValue() * direction ;
if (solValue < cutoff) {
// we have a solution
solutionFound = true;
solutionValue = solValue;
memcpy(betterSolution, newSolver->getColSolution(),
numberColumns*sizeof(double));
COIN_DETAIL_PRINT(printf("Naive fixing close to zero gave solution of %g\n", solutionValue));
cutoff = solValue - model_->getCutoffIncrement();
}
}
// Now fix all integers as close to zero if zero or large cost
int nFix = 0;
for (i = 0; i < numberIntegers; i++) {
int iColumn = integerVariable[i];
double lower = colLower[iColumn];
double upper = colUpper[iColumn];
double value;
if (fabs(objective[i]) > 0.0 && fabs(objective[i]) < large_) {
nFix++;
if (lower > 0.0)
value = lower;
else if (upper < 0.0)
value = upper;
else
value = 0.0;
newSolver->setColLower(iColumn, value);
newSolver->setColUpper(iColumn, value);
} else {
// set back to original
newSolver->setColLower(iColumn, lower);
newSolver->setColUpper(iColumn, upper);
}
}
const double * solution = solver->getColSolution();
if (nFix) {
newSolver->setWarmStart(&saveBasis);
newSolver->setColSolution(solution);
newSolver->initialSolve();
if (newSolver->isProvenOptimal()) {
double solValue = newSolver->getObjValue() * direction ;
if (solValue < cutoff) {
// try branch and bound
double * newSolution = new double [numberColumns];
COIN_DETAIL_PRINT(printf("%d fixed after fixing costs\n", nFix));
int returnCode = smallBranchAndBound(newSolver,
numberNodes_, newSolution,
solutionValue,
solutionValue, "CbcHeuristicNaive1");
if (returnCode < 0)
returnCode = 0; // returned on size
if ((returnCode&2) != 0) {
// could add cut
returnCode &= ~2;
}
if (returnCode == 1) {
// solution
solutionFound = true;
memcpy(betterSolution, newSolution,
numberColumns*sizeof(double));
COIN_DETAIL_PRINT(printf("Naive fixing zeros gave solution of %g\n", solutionValue));
cutoff = solutionValue - model_->getCutoffIncrement();
}
delete [] newSolution;
}
}
}
#if 1
newSolver->setObjSense(-direction); // maximize
newSolver->setWarmStart(&saveBasis);
newSolver->setColSolution(solution);
for (int iColumn = 0; iColumn < numberColumns; iColumn++) {
double value = solution[iColumn];
double lower = colLower[iColumn];
double upper = colUpper[iColumn];
double newLower;
double newUpper;
if (newSolver->isInteger(iColumn)) {
newLower = CoinMax(lower, floor(value) - 2.0);
newUpper = CoinMin(upper, ceil(value) + 2.0);
} else {
newLower = CoinMax(lower, value - 1.0e5);
newUpper = CoinMin(upper, value + 1.0e-5);
}
newSolver->setColLower(iColumn, newLower);
newSolver->setColUpper(iColumn, newUpper);
}
newSolver->initialSolve();
if (newSolver->isProvenOptimal()) {
double solValue = newSolver->getObjValue() * direction ;
if (solValue < cutoff) {
nFix = 0;
newSolver->setObjSense(direction); // correct direction
//const double * thisSolution = newSolver->getColSolution();
for (int iColumn = 0; iColumn < numberColumns; iColumn++) {
double value = solution[iColumn];
double lower = colLower[iColumn];
double upper = colUpper[iColumn];
double newLower = lower;
double newUpper = upper;
if (newSolver->isInteger(iColumn)) {
if (value < lower + 1.0e-6) {
nFix++;
newUpper = lower;
} else if (value > upper - 1.0e-6) {
nFix++;
newLower = upper;
} else {
newLower = CoinMax(lower, floor(value) - 2.0);
newUpper = CoinMin(upper, ceil(value) + 2.0);
}
}
newSolver->setColLower(iColumn, newLower);
newSolver->setColUpper(iColumn, newUpper);
}
// try branch and bound
double * newSolution = new double [numberColumns];
COIN_DETAIL_PRINT(printf("%d fixed after maximizing\n", nFix));
int returnCode = smallBranchAndBound(newSolver,
numberNodes_, newSolution,
solutionValue,
solutionValue, "CbcHeuristicNaive1");
if (returnCode < 0)
returnCode = 0; // returned on size
if ((returnCode&2) != 0) {
// could add cut
returnCode &= ~2;
}
if (returnCode == 1) {
// solution
solutionFound = true;
memcpy(betterSolution, newSolution,
numberColumns*sizeof(double));
COIN_DETAIL_PRINT(printf("Naive maximizing gave solution of %g\n", solutionValue));
cutoff = solutionValue - model_->getCutoffIncrement();
}
delete [] newSolution;
}
}
#endif
delete newSolver;
return solutionFound ? 1 : 0;
}
