void profit_create_vars(graph g, IloModel model, IloNumVarArray x, IloNumVarArray p, IloNumVarArray z, int **columns) {
IloEnv env = model.getEnv();
string s;
assignment_vars(g, model, x, columns);
for(int j = 0; j < g->items; j++) {
s = "p_" + itos(j);
p.add(IloNumVar(env, 0.0, boundp(j), ILOFLOAT, s.c_str()));
// p.add(IloNumVar(env, lowerp(j), boundp(j), ILOFLOAT, s.c_str()));
}
for(int i = 0; i < g->bidders; i++) {
s = "z_" + itos(i);
z.add(IloNumVar(env, 0.0, boundu(i), ILOFLOAT, s.c_str()));
}
}
void addInitColumn(IloNumVarArray lambda, IloObjective rmpObj,
IloRangeArray maintConEng, IloRangeArray removeMod, IloRangeArray convex,
IloNumArray2 addXCol, IloNumArray addZCol,
const IloNumArray compCosts, const IloNumArray convexityCoef) {
// loop counter
IloInt t;
// counter for objective function coefficient for lambda
// variable to be added.
IloNum lambdaObjCoef = 0;
// function assumes addXCol and addZCol contains proper values.
// calculate objective function coefficient lambdaObjCoef
for (t = 0; t < TIME_SPAN; t++) {
// for each fixed t: scalar product of x[m]* vector and
// component costs vector compCosts[m]:
lambdaObjCoef += IloScalProd(addXCol[t],compCosts);
// also clear the addXCol subarrays as soon as they
// have been used
addXCol[t].clear();
}
// now add this column and it's associated lambda variable to the RMP.
lambda.add(IloNumVar(rmpObj(lambdaObjCoef) + maintConEng(addZCol) +
removeMod(addZCol) + convex(convexityCoef), 0.0, 1.0));
// clear addZCol num array.
addZCol.clear();
} // END of addColumn
static void
populatebycolumn (IloModel model, IloNumVarArray x, IloRangeArray c)
{
IloEnv env = model.getEnv();
IloObjective obj = IloMaximize(env);
c.add(IloRange(env, -IloInfinity, 20.0));
c.add(IloRange(env, -IloInfinity, 30.0));
x.add(IloNumVar(obj(1.0) + c[0](-1.0) + c[1]( 1.0), 0.0, 40.0));
x.add(obj(2.0) + c[0]( 1.0) + c[1](-3.0));
x.add(obj(3.0) + c[0]( 1.0) + c[1]( 1.0));
model.add(obj);
model.add(c);
} // END populatebycolumn
void addColumn(IloCplex subSolver, IloNumVarArray2 x, IloNumVarArray z, IloNumVarArray lambda,
IloObjective rmpObj, IloRangeArray maintConEng, IloRangeArray removeMod,
IloRangeArray convex, IloNumArray2 addXCol, IloNumArray addZCol,
const IloNumArray compCosts, const IloNumArray convexityCoef) {
// loop counter
IloInt t;
// counter for objective function coefficient for lambda
// variable to be added.
IloNum lambdaObjCoef = 0;
// extract subproblem-optimal solution values
// (into IloNumArrays addXCol (2d) and addZCol (1d)).
// z values:
subSolver.getValues(addZCol,z);
//cout << endl << endl << "z = " << endl << addZCol << endl;
//cin.get();
// !!! OBS !!!
// here we might want to save these z values some column pool's custom-nitted class
// array. Or to be specific, we want to add the indexes for NON-ZERO-ENTRIES in addZCol
// to our class that keep place of columns.
// E.g., given variable lambda(m)_(q_m), we want to know in our own class object,
// given (m)(q_m), the indexes of non-zeros in that Z column.
// and for each t...
for (t = 0; t < TIME_SPAN; t++) {
// x values:
subSolver.getValues(addXCol[t],x[t]);
//cout << endl << endl << "x[t=" << t << "] =" << endl << addXCol[t] << endl;
}
//cin.get();
// calculate objective function coefficient lambdaObjCoef
for (t = 0; t < TIME_SPAN; t++) {
// for each fixed t: scalar product of x[m]* vector and
// component costs vector compCosts[m]:
lambdaObjCoef += IloScalProd(addXCol[t],compCosts);
// also clear the addXCol subarrays as soon as they
// have been used
addXCol[t].clear();
}
// now add this column and it's associated lambda variable to the RMP.
lambda.add(IloNumVar(rmpObj(lambdaObjCoef) + maintConEng(addZCol) +
removeMod(addZCol) + convex(convexityCoef), 0.0, 1.0));
// clear addZCol num array.
addZCol.clear();
} // END of addColumn
static void
populatebyrow (IloModel model, IloNumVarArray x, IloRangeArray c)
{
IloEnv env = model.getEnv();
x.add(IloNumVar(env, 0.0, 40.0));
x.add(IloNumVar(env));
x.add(IloNumVar(env));
x.add(IloNumVar(env, 2.0, 3.0, ILOINT));
model.add(IloMaximize(env, x[0] + 2 * x[1] + 3 * x[2] + x[3]));
c.add( - x[0] + x[1] + x[2] + 10 * x[3] <= 20);
c.add( x[0] - 3 * x[1] + x[2] <= 30);
c.add( x[1] - 3.5* x[3] == 0);
model.add(c);
} // END populatebyrow
void PopulateFromGraph(IloModel model, IloNumVarArray s, IloRangeArray c){
IloEnv env = model.getEnv();
// Used n+1 for accomodating an extra variable
// For being able to write the objective function
for(int i = 0; i <= n; i++)
s.add(IloNumVar(env));
}
static void
populatebyrow (IloModel model, IloNumVarArray x, IloRangeArray c)
{
IloEnv env = model.getEnv();
x.add(IloNumVar(env, 0.0, 40.0));
x.add(IloNumVar(env));
x.add(IloNumVar(env));
model.add(IloMaximize(env, x[0] + 2 * x[1] + 3 * x[2]
- 0.5 * (33*x[0]*x[0] + 22*x[1]*x[1] +
11*x[2]*x[2] - 12*x[0]*x[1] -
23*x[1]*x[2] ) ));
c.add( - x[0] + x[1] + x[2] <= 20);
c.add( x[0] - 3 * x[1] + x[2] <= 30);
model.add(c);
} // END populatebyrow
IloExpr minBeamletStr(IloEnv & env, IloNumVarArray & w, Sparse_Matrix<IloInt> & vox)
{
IloExpr obj(env);
for(int i=0;i<w.getSize();++i)
{
obj += w[i]/IloNum(100);
}
return obj;
}
void generateProblem(const ILPModel& m, IloModel& model, IloNumVarArray& x, IloRangeArray& con) {
IloEnv env = model.getEnv();
IloObjective obj = (m.obj == MINIMIZE ? IloMinimize(env) : IloMaximize(env));
for (unsigned long v = 0; v < m.numberOfVariables(); ++v) {
switch (m.x[v].type) {
case FLT:
x.add(IloNumVar(env, m.x[v].lowerBound, m.x[v].upperBound, IloNumVar::Float));
break;
case BIN:
x.add(IloNumVar(env, m.x[v].lowerBound, m.x[v].upperBound, IloNumVar::Bool));
break;
default:
x.add(IloNumVar(env, m.x[v].lowerBound, m.x[v].upperBound, IloNumVar::Int));
}
obj.setLinearCoef(x[v], m.c[v]);
x[v].setName(m.varDesc[v].c_str());
}
for (unsigned long c = 0; c < m.numberOfConstraints(); ++c) {
switch (m.ops[c]) {
case LESS_EQUAL:
con.add(IloRange(env, -IloInfinity, m.b[c]));
break;
case EQUAL:
con.add(IloRange(env, m.b[c], m.b[c]));
break;
case GREATER_EQUAL:
con.add(IloRange(env, m.b[c], IloInfinity));
}
for (const pair<uint32_t, double>& p : m.A[c])
con[c].setLinearCoef(x[p.first], p.second);
con[c].setName(m.conDesc[c].c_str());
}
model.add(obj);
model.add(con);
}
static void
populatebyrow (IloModel model, IloNumVarArray x, IloRangeArray c)
{
IloEnv env = model.getEnv();
x.add(IloNumVar(env, -1.0, 1.0));
x.add(IloNumVar(env, 0.0, 1.0));
model.add(IloMinimize(env, 0.5 * (-3*x[0]*x[0] - 3*x[1]*x[1] +
- 1*x[0]*x[1] ) ));
c.add( - x[0] + x[1] >= 0);
c.add( x[0] + x[1] >= 0);
model.add(c);
} // END populatebyrow
/** Destructor. */
~Worker() {
if ( handle ) {
handle.kill();
handle.join();
}
if ( cplex.getImpl() ) {
cplex.userfunction(USERACTION_REMOVECALLBACK, 0, NULL, 0, 0, NULL);
cplex.end();
}
rng.end();
x.end();
obj.end();
model.end();
}
static void
populatebyrow (IloModel model, IloNumVarArray x, IloRangeArray c)
{
IloEnv env = model.getEnv();
x.add(IloNumVar(env, 0.0, 40.0));
x.add(IloNumVar(env, 0.0, IloInfinity, ILOINT));
x.add(IloNumVar(env, 0.0, IloInfinity, ILOINT));
x.add(IloNumVar(env, 2.0, 3.0, ILOINT));
model.add(IloMaximize(env, x[0] + 2 * x[1] + 3 * x[2] + x[3]));
c.add( - x[0] + x[1] + x[2] + 10 * x[3] <= 20);
c.add( x[0] - 3 * x[1] + x[2] <= 30);
c.add( x[1] - 3.5* x[3] == 0);
model.add(c);
IloNumVarArray sosvar(env, 2);
IloNumArray sosval(env, 2);
sosvar[0] = x[2]; sosvar[1] = x[3];
sosval[0] = 25.0; sosval[1] = 18.0;
model.add(IloSOS1(model.getEnv(), sosvar, sosval));
} // END populatebyrow
/** Extract sub block number <code>n</code> from <code>problem</code>.
* The constructor creates a representation of block number <code>n</code>
* as described in <code>problem</code>.
* The constructor will also connect the newly created block to a remote
* object solver instance.
* @param problem The problem from which the block is to be extracted.
* @param n Index of the block to be extracted.
* @param argc Argument for IloCplex constructor.
* @param argv Argument for IloCplex constructor.
* @param machines List of machines to which to connect. If the code is
* compiled for the TCP/IP transport then the block will
* be connected to <code>machines[n]</code>.
*/
BendersOpt::Block::Block(Problem const *problem, IloInt n, int argc, char const *const *argv,
std::vector<char const *> const &machines)
: env(), number(n), vars(0), rows(0), cplex(0), cb(0)
{
IloNumVarArray problemVars = problem->getVariables();
IloRangeArray problemRanges = problem->getRows();
// Create a map that maps variables in the original model to their
// respective index in problemVars.
std::map<IloNumVar,IloInt,ExtractableLess<IloNumVar> > origIdxMap;
for (IloInt j = 0; j < problemVars.getSize(); ++j)
origIdxMap.insert(std::map<IloNumVar,IloInt,ExtractableLess<IloNumVar> >::value_type(problemVars[j], j));
// Copy non-fixed variables from original problem into primal problem.
IloExpr primalObj(env);
IloNumVarArray primalVars(env);
IloRangeArray primalRows(env);
IdxMap idxMap; // Index of original variable in block's primal model
RowSet rowSet;
for (IloInt j = 0; j < problemVars.getSize(); ++j) {
IloNumVar x = problemVars[j];
if ( problem->getBlock(x) == number ) {
// Create column in block LP with exactly the same data.
if ( x.getType() != IloNumVar::Float ) {
std::stringstream s;
s << "Cannot create non-continuous block variable " << x;
std::cerr << s.str() << std::endl;
throw s.str();
}
IloNumVar v(env, x.getLB(), x.getUB(), x.getType(), x.getName());
// Normalize objective function to 'minimize'
double coef = problem->getObjCoef(x);
if ( problem->getObjSense() != IloObjective::Minimize )
coef *= -1.0;
primalObj += coef * v;
// Record the index that the copied variable has in the
// block model.
idxMap.insert(IdxMap::value_type(x, primalVars.getSize()));
primalVars.add(v);
// Mark the rows that are intersected by this column
// so that we can collect them later.
RowSet const &intersected = problem->getIntersectedRows(x);
for (RowSet::const_iterator it = intersected.begin();
it != intersected.end(); ++it)
rowSet.insert(*it);
}
else
idxMap.insert(IdxMap::value_type(x, -1));
}
// Now copy all rows that intersect block variables.
for (IloInt i = 0; i < problemRanges.getSize(); ++i) {
IloRange r = problemRanges[i];
if ( rowSet.find(r) == rowSet.end() )
continue;
// Create a copy of the row, normalizing it to '<='
double factor = 1.0;
if ( r.getLB() > -IloInfinity )
factor = -1.0;
IloRange primalR(env,
factor < 0 ? -r.getUB() : r.getLB(),
factor < 0 ? -r.getLB() : r.getUB(), r.getName());
IloExpr lhs(env);
for (IloExpr::LinearIterator it = r.getLinearIterator(); it.ok(); ++it)
{
IloNumVar v = it.getVar();
double const val = factor * it.getCoef();
if ( problem->getBlock(v) != number ) {
// This column is not explicitly in this block. This means
// that it is a column that will be fixed by the master.
// We collect all such columns so that we can adjust the
// dual objective function according to concrete fixings.
// Store information about variables in this block that
// will be fixed by master solves.
fixed.push_back(FixData(primalRows.getSize(), origIdxMap[v], -val));
}
else {
// The column is an ordinary in this block. Just copy it.
lhs += primalVars[idxMap[v]] * val;
}
}
primalR.setExpr(lhs);
primalRows.add(primalR);
lhs.end();
}
// Create the dual of the primal model we just created.
// Note that makeDual _always_ returns a 'maximize' objective.
IloObjective objective(env, primalObj, IloObjective::Minimize);
makeDual(objective, primalVars, primalRows,
&obj, &vars, &rows);
objective.end();
primalRows.endElements();
primalRows.end();
primalVars.endElements();
primalVars.end();
primalObj.end();
// Create a model.
IloModel model(env);
model.add(obj);
model.add(vars);
model.add(rows);
for (IloExpr::LinearIterator it = obj.getLinearIterator(); it.ok(); ++it)
objMap.insert(ObjMap::value_type(it.getVar(), it.getCoef()));
// Finally create the IloCplex instance that will solve
// the problems associated with this block.
char const **transargv = new char const *[argc + 3];
for (int i = 0; i < argc; ++i)
transargv[i] = argv[i];
#if defined(USE_MPI)
char extra[128];
sprintf (extra, "-remoterank=%d", static_cast<int>(number + 1));
transargv[argc++] = extra;
(void)machines;
#elif defined(USE_PROCESS)
char extra[128];
sprintf (extra, "-logfile=block%04d.log", static_cast<int>(number));
transargv[argc++] = extra;
(void)machines;
#elif defined(USE_TCPIP)
transargv[argc++] = machines[number];
#endif
cplex = IloCplex(model, TRANSPORT, argc, transargv);
delete[] transargv;
// Suppress output from this block's solver.
cplex.setOut(env.getNullStream());
cplex.setWarning(env.getNullStream());
}
// This routine set up the IloCplex algorithm to solve the worker LP, and
// creates the worker LP (i.e., the dual of flow constraints and
// capacity constraints of the flow MILP)
//
// Modeling variables:
// forall k in V0, i in V:
// u(k,i) = dual variable associated with flow constraint (k,i)
//
// forall k in V0, forall (i,j) in A:
// v(k,i,j) = dual variable associated with capacity constraint (k,i,j)
//
// Objective:
// minimize sum(k in V0) sum((i,j) in A) x(i,j) * v(k,i,j)
// - sum(k in V0) u(k,0) + sum(k in V0) u(k,k)
//
// Constraints:
// forall k in V0, forall (i,j) in A: u(k,i) - u(k,j) <= v(k,i,j)
//
// Nonnegativity on variables v(k,i,j)
// forall k in V0, forall (i,j) in A: v(k,i,j) >= 0
//
void
createWorkerLP(IloCplex cplex, IloNumVarArray v, IloNumVarArray u,
IloObjective obj, IloInt numNodes)
{
IloInt i, j, k;
IloEnv env = cplex.getEnv();
IloModel mod(env, "atsp_worker");
// Set up IloCplex algorithm to solve the worker LP
cplex.extract(mod);
cplex.setOut(env.getNullStream());
// Turn off the presolve reductions and set the CPLEX optimizer
// to solve the worker LP with primal simplex method.
cplex.setParam(IloCplex::Reduce, 0);
cplex.setParam(IloCplex::RootAlg, IloCplex::Primal);
// Create variables v(k,i,j) forall k in V0, (i,j) in A
// For simplicity, also dummy variables v(k,i,i) are created.
// Those variables are fixed to 0 and do not partecipate to
// the constraints.
IloInt numArcs = numNodes * numNodes;
IloInt vNumVars = (numNodes-1) * numArcs;
IloNumVarArray vTemp(env, vNumVars, 0, IloInfinity);
for (k = 1; k < numNodes; ++k) {
for (i = 0; i < numNodes; ++i) {
vTemp[(k-1)*numArcs + i *numNodes + i].setBounds(0, 0);
}
}
v.clear();
v.add(vTemp);
vTemp.end();
mod.add(v);
// Set names for variables v(k,i,j)
for (k = 1; k < numNodes; ++k) {
for(i = 0; i < numNodes; ++i) {
for(j = 0; j < numNodes; ++j) {
char varName[100];
sprintf(varName, "v.%d.%d.%d", (int) k, (int) i, (int) j);
v[(k-1)*numArcs + i*numNodes + j].setName(varName);
}
}
}
// Associate indices to variables v(k,i,j)
IloIntArray vIndex(env, vNumVars);
for (j = 0; j < vNumVars; ++j)
{
vIndex[j] = j;
v[j].setObject(&vIndex[j]);
}
// Create variables u(k,i) forall k in V0, i in V
IloInt uNumVars = (numNodes-1) * numNodes;
IloNumVarArray uTemp(env, uNumVars, -IloInfinity, IloInfinity);
u.clear();
u.add(uTemp);
uTemp.end();
mod.add(u);
// Set names for variables u(k,i)
for (k = 1; k < numNodes; ++k) {
for(i = 0; i < numNodes; ++i) {
char varName[100];
sprintf(varName, "u.%d.%d", (int) k, (int) i);
u[(k-1)*numNodes + i].setName(varName);
}
}
// Associate indices to variables u(k,i)
IloIntArray uIndex(env, uNumVars);
for (j = 0; j < uNumVars; ++j)
{
uIndex[j] = vNumVars + j;
u[j].setObject(&uIndex[j]);
}
// Initial objective function is empty
obj.setSense(IloObjective::Minimize);
mod.add(obj);
// Add constraints:
// forall k in V0, forall (i,j) in A: u(k,i) - u(k,j) <= v(k,i,j)
for (k = 1; k < numNodes; ++k) {
for(i = 0; i < numNodes; ++i) {
for(j = 0; j < numNodes; ++j) {
if ( i != j ) {
IloExpr expr(env);
expr -= v[(k-1)*numArcs + i*(numNodes) + j];
expr += u[(k-1)*numNodes + i];
expr -= u[(k-1)*numNodes + j];
mod.add(expr <= 0);
expr.end();
}
}
}
}
}// END createWorkerLP
Example(IloEnv env)
: nblocks(0), model(env), vars(env), ranges(env)
{
// Model data.
// fixed[] is the fixed cost for opening a facility,
// cost[i,j] is the cost for serving customer i from facility j.
static double const fixed[] = { 2.0, 3.0, 3.0 };
static double const cost[] = { 2.0, 3.0, 4.0, 5.0, 7.0,
4.0, 3.0, 1.0, 2.0, 6.0,
5.0, 4.0, 2.0, 1.0, 3.0 };
#define NFACTORY ((CPXDIM)(sizeof(fixed) / sizeof(fixed[0])))
#define NCUSTOMER ((CPXDIM)((sizeof(cost) / sizeof(cost[0])) / NFACTORY))
nblocks = NCUSTOMER;
IloExpr obj(env);
// Create integer y variables.
IloNumVarArray y(env);
for (IloInt f = 0; f < NFACTORY; ++f) {
std::stringstream s;
s << "y" << f;
IloIntVar v(env, 0, 1, s.str().c_str());
obj += fixed[f] * v;
objMap[v] = fixed[f];
y.add(v);
blockMap.insert(BlockMap::value_type(v, -1));
intersectMap.insert(IntersectMap::value_type(v, RowSet()));
}
// Create continuous x variables.
IloNumVarArray x(env);
for (IloInt f = 0; f < NFACTORY; ++f) {
for (IloInt c = 0; c < NCUSTOMER; ++c) {
std::stringstream s;
s << "x" << f << "#" << c;
IloNumVar v(env, 0.0, IloInfinity, s.str().c_str());
obj += v * cost[f * NCUSTOMER + c];
objMap[v] = cost[f * NCUSTOMER + c];
x.add(v);
blockMap.insert(BlockMap::value_type(v, c));
intersectMap.insert(IntersectMap::value_type(v, RowSet()));
}
}
vars.add(y);
vars.add(x);
model.add(vars);
// Add objective function.
model.add(IloMinimize(env, obj, "obj"));
objSense = IloObjective::Minimize;
obj.end();
// Satisfy each customer's demand.
for (IloInt c = 0; c < NCUSTOMER; ++c) {
std::stringstream s;
s << "c1_" << c;
IloRange r(env, 1.0, IloInfinity, s.str().c_str());
IloExpr lhs(env);
for (IloInt f = 0; f < NFACTORY; ++f) {
lhs += x[f * NCUSTOMER + c];
intersectMap[x[f * NCUSTOMER + c]].insert(r);
}
r.setExpr(lhs);
ranges.add(r);
lhs.end();
}
// A factory must be open if we service from it.
for (IloInt c = 0; c < NCUSTOMER; ++c) {
for (IloInt f = 0; f < NFACTORY; ++f) {
std::stringstream s;
s << "c2_" << c << "#" << f;
IloRange r(env, 0.0, IloInfinity, s.str().c_str());
intersectMap[x[f * NCUSTOMER + c]].insert(r);
intersectMap[y[f]].insert(r);
r.setExpr(-x[f * NCUSTOMER + c] + y[f]);
ranges.add(r);
}
}
// Capacity constraint.
IloRange r(env, -IloInfinity, NFACTORY - 1, "c3");
IloExpr lhs(env);
for (IloInt f = 0; f < NFACTORY; ++f) {
lhs += y[f];
intersectMap[y[f]].insert(r);
}
r.setExpr(lhs);
ranges.add(r);
lhs.end();
model.add(ranges);
#undef NFACTORY
#undef NCUSTOMER
}
void
makeCuts(const IloNumVarArray vars, IloExprArray lhs, IloNumArray rhs) {
IloNumVar x11, x12, x13, x14, x15;
IloNumVar x21, x22, x23, x24, x25;
IloNumVar x31, x32, x33, x34, x35;
IloNumVar x41, x42, x43, x44, x45;
IloNumVar x51, x52, x53, x54, x55;
IloNumVar w11, w12, w13, w14, w15;
IloNumVar w21, w22, w23, w24, w25;
IloNumVar w31, w32, w33, w34, w35;
IloNumVar w41, w42, w43, w44, w45;
IloNumVar w51, w52, w53, w54, w55;
IloInt num = vars.getSize();
for (IloInt i = 0; i < num; i++) {
if ( strcmp(vars[i].getName(), "X11") == 0 ) x11 = vars[i];
else if ( strcmp(vars[i].getName(), "X12") == 0 ) x12 = vars[i];
else if ( strcmp(vars[i].getName(), "X13") == 0 ) x13 = vars[i];
else if ( strcmp(vars[i].getName(), "X14") == 0 ) x14 = vars[i];
else if ( strcmp(vars[i].getName(), "X15") == 0 ) x15 = vars[i];
else if ( strcmp(vars[i].getName(), "X21") == 0 ) x21 = vars[i];
else if ( strcmp(vars[i].getName(), "X22") == 0 ) x22 = vars[i];
else if ( strcmp(vars[i].getName(), "X23") == 0 ) x23 = vars[i];
else if ( strcmp(vars[i].getName(), "X24") == 0 ) x24 = vars[i];
else if ( strcmp(vars[i].getName(), "X25") == 0 ) x25 = vars[i];
else if ( strcmp(vars[i].getName(), "X31") == 0 ) x31 = vars[i];
else if ( strcmp(vars[i].getName(), "X32") == 0 ) x32 = vars[i];
else if ( strcmp(vars[i].getName(), "X33") == 0 ) x33 = vars[i];
else if ( strcmp(vars[i].getName(), "X34") == 0 ) x34 = vars[i];
else if ( strcmp(vars[i].getName(), "X35") == 0 ) x35 = vars[i];
else if ( strcmp(vars[i].getName(), "X41") == 0 ) x41 = vars[i];
else if ( strcmp(vars[i].getName(), "X42") == 0 ) x42 = vars[i];
else if ( strcmp(vars[i].getName(), "X43") == 0 ) x43 = vars[i];
else if ( strcmp(vars[i].getName(), "X44") == 0 ) x44 = vars[i];
else if ( strcmp(vars[i].getName(), "X45") == 0 ) x45 = vars[i];
else if ( strcmp(vars[i].getName(), "X51") == 0 ) x51 = vars[i];
else if ( strcmp(vars[i].getName(), "X52") == 0 ) x52 = vars[i];
else if ( strcmp(vars[i].getName(), "X53") == 0 ) x53 = vars[i];
else if ( strcmp(vars[i].getName(), "X54") == 0 ) x54 = vars[i];
else if ( strcmp(vars[i].getName(), "X55") == 0 ) x55 = vars[i];
else if ( strcmp(vars[i].getName(), "W11") == 0 ) w11 = vars[i];
else if ( strcmp(vars[i].getName(), "W12") == 0 ) w12 = vars[i];
else if ( strcmp(vars[i].getName(), "W13") == 0 ) w13 = vars[i];
else if ( strcmp(vars[i].getName(), "W14") == 0 ) w14 = vars[i];
else if ( strcmp(vars[i].getName(), "W15") == 0 ) w15 = vars[i];
else if ( strcmp(vars[i].getName(), "W21") == 0 ) w21 = vars[i];
else if ( strcmp(vars[i].getName(), "W22") == 0 ) w22 = vars[i];
else if ( strcmp(vars[i].getName(), "W23") == 0 ) w23 = vars[i];
else if ( strcmp(vars[i].getName(), "W24") == 0 ) w24 = vars[i];
else if ( strcmp(vars[i].getName(), "W25") == 0 ) w25 = vars[i];
else if ( strcmp(vars[i].getName(), "W31") == 0 ) w31 = vars[i];
else if ( strcmp(vars[i].getName(), "W32") == 0 ) w32 = vars[i];
else if ( strcmp(vars[i].getName(), "W33") == 0 ) w33 = vars[i];
else if ( strcmp(vars[i].getName(), "W34") == 0 ) w34 = vars[i];
else if ( strcmp(vars[i].getName(), "W35") == 0 ) w35 = vars[i];
else if ( strcmp(vars[i].getName(), "W41") == 0 ) w41 = vars[i];
else if ( strcmp(vars[i].getName(), "W42") == 0 ) w42 = vars[i];
else if ( strcmp(vars[i].getName(), "W43") == 0 ) w43 = vars[i];
else if ( strcmp(vars[i].getName(), "W44") == 0 ) w44 = vars[i];
else if ( strcmp(vars[i].getName(), "W45") == 0 ) w45 = vars[i];
else if ( strcmp(vars[i].getName(), "W51") == 0 ) w51 = vars[i];
else if ( strcmp(vars[i].getName(), "W52") == 0 ) w52 = vars[i];
else if ( strcmp(vars[i].getName(), "W53") == 0 ) w53 = vars[i];
else if ( strcmp(vars[i].getName(), "W54") == 0 ) w54 = vars[i];
else if ( strcmp(vars[i].getName(), "W55") == 0 ) w55 = vars[i];
}
lhs.add(x21 - x22);
rhs.add(0.0);
lhs.add(x22 - x23);
rhs.add(0.0);
lhs.add(x23 - x24);
rhs.add(0.0);
lhs.add(2.08*x11 + 2.98*x21 + 3.47*x31 + 2.24*x41 + 2.08*x51 +
0.25*w11 + 0.25*w21 + 0.25*w31 + 0.25*w41 + 0.25*w51);
rhs.add(20.25);
lhs.add(2.08*x12 + 2.98*x22 + 3.47*x32 + 2.24*x42 + 2.08*x52 +
0.25*w12 + 0.25*w22 + 0.25*w32 + 0.25*w42 + 0.25*w52);
rhs.add(20.25);
lhs.add(2.08*x13 + 2.98*x23 + 3.47*x33 + 2.24*x43 + 2.08*x53 +
0.25*w13 + 0.25*w23 + 0.25*w33 + 0.25*w43 + 0.25*w53);
rhs.add(20.25);
lhs.add(2.08*x14 + 2.98*x24 + 3.47*x34 + 2.24*x44 + 2.08*x54 +
0.25*w14 + 0.25*w24 + 0.25*w34 + 0.25*w44 + 0.25*w54);
rhs.add(20.25);
lhs.add(2.08*x15 + 2.98*x25 + 3.47*x35 + 2.24*x45 + 2.08*x55 +
0.25*w15 + 0.25*w25 + 0.25*w35 + 0.25*w45 + 0.25*w55);
rhs.add(16.25);
}
// This function creates the following model:
// Minimize
// obj: x1 + x2 + x3 + x4 + x5 + x6
// Subject To
// c1: x1 + x2 + x5 = 8
// c2: x3 + x5 + x6 = 10
// q1: [ -x1^2 + x2^2 + x3^2 ] <= 0
// q2: [ -x4^2 + x5^2 ] <= 0
// Bounds
// x2 Free
// x3 Free
// x5 Free
// End
// which is a second order cone program in standard form.
// The function returns objective, variables and constraints in the
// values obj, vars and rngs.
// The function also sets up cone so that for a column j we have
// cone[j] >= 0 Column j is in a cone constraint and is the
// cone's head variable.
// cone[j] == NOT_CONE_HEAD Column j is in a cone constraint but is
// not the cone's head variable..
// cone[j] == NOT_IN_CONE Column j is not contained in any cone constraint.
static void
createmodel (IloModel& model, IloObjective &obj, IloNumVarArray &vars,
IloRangeArray &rngs, IloIntArray& cone)
{
// The indices we assign as user objects to the modeling objects.
// We define them as static data so that we don't have to worry about
// dynamic memory allocation/leakage.
static int indices[] = { 0, 1, 2, 3, 4, 5, 6 };
IloEnv env = model.getEnv();
// Create variables.
IloNumVar x1(env, 0, IloInfinity, "x1");
IloNumVar x2(env, -IloInfinity, IloInfinity, "x2");
IloNumVar x3(env, -IloInfinity, IloInfinity, "x3");
IloNumVar x4(env, 0, IloInfinity, "x4");
IloNumVar x5(env, -IloInfinity, IloInfinity, "x5");
IloNumVar x6(env, 0, IloInfinity, "x6");
// Create objective function and immediately store it in return value.
obj = IloMinimize(env, x1 + x2 + x3 + x4 + x5 + x6);
// Create constraints.
IloRange c1(env, 8, x1 + x2 + x5, 8, "c1");
IloRange c2(env, 10, x3 + x5 + x6, 10, "c2");
IloRange q1(env, -IloInfinity, -x1*x1 + x2*x2 + x3*x3, 0, "q1");
cone.add(2); // x1, cone head of constraint at index 2
cone.add(NOT_CONE_HEAD); // x2
cone.add(NOT_CONE_HEAD); // x3
IloRange q2(env, -IloInfinity, -x4*x4 + x5*x5, 0, "q2");
cone.add(3); // x4, cone head of constraint at index 3
cone.add(NOT_CONE_HEAD); // x5
cone.add(NOT_IN_CONE); // x6
// Setup model.
model.add(obj);
model.add(obj);
model.add(c1);
model.add(c2);
model.add(q1);
model.add(q2);
// Setup return values.
vars.add(x1);
vars.add(x2);
vars.add(x3);
vars.add(x4);
vars.add(x5);
vars.add(x6);
rngs.add(c1);
rngs.add(c2);
rngs.add(q1);
rngs.add(q2);
// We set the user object for each modeling object to its index in the
// respective array. This makes the code in checkkkt a little simpler.
for (IloInt i = 0; i < vars.getSize(); ++i)
vars[i].setObject(&indices[i]);
for (IloInt i = 0; i < rngs.getSize(); ++i)
rngs[i].setObject(&indices[i]);
}
void LpSolver::populatebyrow (CplexConverter& cplexConverter,
IloModel model, IloNumVarArray x, IloRangeArray c)
{
IloEnv env = model.getEnv();
// CAPITAL LETTERS MEAN I NEED YOUR HELP, here is help
// IloExpr cost(env);
// Create Variables
// cout << "size of var: " << cplexConverter.variables.size() << endl;
for (int i = 0; i < cplexConverter.variables.size(); ++i){
IloNumVar iloVar(env, 0.0, cplexConverter.capacities[i], IloNumVar::Int);
// cout << iloVar << endl;
x.add(iloVar);
}
//Capacity Constraints
for (auto &it : cplexConverter.atomicIdToVarIdDict){
IloExpr t(env);
// cout << "adding constraint ";
for (int j = 0; j < it.second.size(); j++){
// cout << "x[" << it.second[j] << "] + ";
t += x[it.second[j]];
}
// cout << endl;
c.add(t <= cplexConverter.graph->atomicEdges[it.first]->capacity);
// cout << c << endl;
t.end();
}
// other constraints
for (auto nodePair : cplexConverter.graph->nodes){
// For all nodes
Node* n = nodePair.second;
if(n == cplexConverter.src){
// source constraints
// IloExpr inFlow(env);
IloExpr outFlow(env);
for(auto &atoIn : n->atomicEdge_in){
int aeId = atoIn.second->atomicEdgeId;
for (int j = 0; j < cplexConverter.atomicIdToVarIdDict[aeId].size(); j++){
// var Id
int vId = cplexConverter.atomicIdToVarIdDict[aeId][j];
outFlow += x[vId];
// cost += cplexConverter.graph->atomicEdges[cplexConverter.variables[vId].atomicEdgeId]->interest_rate * x[vId];
}
}
for (auto &atoOut : n->atomicEdge_out){
int aeId = atoOut.second->atomicEdgeId;
for (int j = 0; j < cplexConverter.atomicIdToVarIdDict[aeId].size(); j++){
// var Id
int vId = cplexConverter.atomicIdToVarIdDict[aeId][j];
// inFlow += x[vId];
c.add(x[vId] == 0);
// cost -= cplexConverter.graph->atomicEdges[cplexConverter.variables[vId].atomicEdgeId]->interest_rate * x[vId];
}
}
c.add(outFlow == cplexConverter.request);
// inFlow.end();
outFlow.end();
} else if(n == cplexConverter.dest){
// destination constraints
IloExpr inFlow(env);
// IloExpr outFlow(env);
for(auto &atoIn : n->atomicEdge_in){
int aeId = atoIn.second->atomicEdgeId;
for (int j = 0; j < cplexConverter.atomicIdToVarIdDict[aeId].size(); j++){
// var Id
int vId = cplexConverter.atomicIdToVarIdDict[aeId][j];
// outFlow += x[vId];
c.add(x[vId] == 0);
}
}
for (auto &atoOut : n->atomicEdge_out){
int aeId = atoOut.second->atomicEdgeId;
for (int j = 0; j < cplexConverter.atomicIdToVarIdDict[aeId].size(); j++){
// var Id
int vId = cplexConverter.atomicIdToVarIdDict[aeId][j];
inFlow += x[vId];
}
}
c.add(inFlow == cplexConverter.request);
inFlow.end();
// outFlow.end();
} else {
// Monotonicity Constraints
for (int i = 0; i < credNetConstants.totalIrs.size(); ++i){
IloExpr tempin(env);
IloExpr tempout(env);
for (auto &atoIn : n->atomicEdge_in){
int aeId = atoIn.second->atomicEdgeId;
for (int j = 0; j < cplexConverter.atomicIdToVarIdDict[aeId].size(); j++){
// var Id
int vId = cplexConverter.atomicIdToVarIdDict[aeId][j];
if (cplexConverter.variables[vId].interest_rate <= credNetConstants.totalIrs[i]){
tempout += x[vId];
}
}
}
for (auto &atoOut : n->atomicEdge_out){
int aeId = atoOut.second->atomicEdgeId;
for (int j = 0; j < cplexConverter.atomicIdToVarIdDict[aeId].size(); j++){
// var Id
int vId = cplexConverter.atomicIdToVarIdDict[aeId][j];
if (cplexConverter.variables[vId].interest_rate <= credNetConstants.totalIrs[i]){
tempin += x[vId];
}
}
}
c.add(tempout - tempin >= 0);
tempout.end();
tempin.end();
}
//Flow Constraints
IloExpr inFlow(env);
IloExpr outFlow(env);
for(auto &atoIn : n->atomicEdge_in){
int aeId = atoIn.second->atomicEdgeId;
for (int j = 0; j < cplexConverter.atomicIdToVarIdDict[aeId].size(); j++){
// var Id
int vId = cplexConverter.atomicIdToVarIdDict[aeId][j];
outFlow += x[vId];
}
}
for (auto &atoOut : n->atomicEdge_out){
int aeId = atoOut.second->atomicEdgeId;
for (int j = 0; j < cplexConverter.atomicIdToVarIdDict[aeId].size(); j++){
// var Id
int vId = cplexConverter.atomicIdToVarIdDict[aeId][j];
inFlow += x[vId];
}
}
c.add(inFlow - outFlow == 0);
inFlow.end();
outFlow.end();
}
}
model.add(c);
// model.add(IloMinimize(env, cost));
// model.add(IloMaximize(env,cost)); //option to minimize cost
// cost.end();
} // END populatebyrow
static void populatebyrow (IloModel model, IloNumVarArray x, IloRangeArray c) {
IloEnv env = model.getEnv();
IloNumArray costs(env);
IloNumArray time(env);
IloNumArray product(env);
int costs_array[] = {1,1,1,10,1,12,2,2,5,10};
int time_array[] = {10,1,7,3,2,3,2,3,7,1};
int product_array[] = {0,3,1,2,-2,0,0,0,0,0};
for(int i=0;i<10;i++)
costs.add(costs_array[i]);
for(int i=0;i<10;i++)
time.add(time_array[i]);
for(int i=0;i<10;i++)
product.add(product_array[i]);
x.add(IloBoolVar(env,"x12")); //0
x.add(IloBoolVar(env,"x24")); //1
x.add(IloBoolVar(env,"x46")); //2
x.add(IloBoolVar(env,"x13")); //3
x.add(IloBoolVar(env,"x32")); //4
x.add(IloBoolVar(env,"x35")); //5
x.add(IloBoolVar(env,"x56")); //6
x.add(IloBoolVar(env,"x25")); //7
x.add(IloBoolVar(env,"x34")); //8
x.add(IloBoolVar(env,"x45")); //9
x.add(IloNumVar(env,0,IloInfinity,ILOINT,"s2")); //10
x.add(IloNumVar(env,0,IloInfinity,ILOINT,"s3")); //11
x.add(IloNumVar(env,0,IloInfinity,ILOINT,"s4")); //12
x.add(IloNumVar(env,0,IloInfinity,ILOINT,"s5")); //13
x.add(IloNumVar(env,0,IloInfinity,ILOINT,"s1")); //14
x.add(IloNumVar(env,0,IloInfinity,ILOINT,"s6")); //15
x.add(IloNumVar(env,0,IloInfinity,ILOINT,"q2")); //16
x.add(IloNumVar(env,0,IloInfinity,ILOINT,"q3")); //17
x.add(IloNumVar(env,0,IloInfinity,ILOINT,"q4")); //18
x.add(IloNumVar(env,0,IloInfinity,ILOINT,"q5")); //19
x.add(IloNumVar(env,0,IloInfinity,ILOINT,"q1")); //20
x.add(IloNumVar(env,0,IloInfinity,ILOINT,"q6")); //21
model.add(IloMinimize(env, costs[0]*x[0] + costs[1]*x[1] + costs[2]*x[2] + costs[3]*x[3] + costs[4]*x[4] + costs[5]*x[5] + costs[6]*x[6] + costs[7]*x[7] + costs[8]*x[8] + costs[9]*x[9]));
c.add(x[0]+ x[3] == 1); // arcs sortant du noeud de depart
c.add(x[2]+ x[6] == 1); // arcs entrant au noeud d arrivee
c.add(x[1]+ x[7] - x[0] - x[4] == 0);
c.add(x[8]+ x[5] + x[4] - x[3] == 0);
c.add(x[9]+ x[2] - x[1] - x[8] == 0);
c.add(x[6]- x[7] - x[5] - x[9] == 0);
c.add(time[0]*x[0] + time[1]*x[1] + time[2]*x[2] + time[3]*x[3] + time[4]*x[4] + time[5]*x[5] + time[6]*x[6] + time[7]*x[7] + time[8]*x[8] + time[9]*x[9] <= 14);
//c.add(product[0]*x[0] + product[1]*x[1] + product[2]*x[2] + product[3]*x[3] + product[4]*x[4] + product[5]*x[5] + product[6]*x[6] + product[7]*x[7] + product[8]*x[8] + product[9]*x[9] <= 4);
c.add(x[14]+time[0]-1000*(1-x[0]) - x[10]<= 0);
c.add(x[20]+product[0]-1000*(1-x[0]) - x[16]<= 0);
c.add(x[10]+time[1]-1000*(1-x[1]) - x[12]<= 0);
c.add(x[16]+product[1]-1000*(1-x[1]) - x[18]<= 0);
c.add(x[18]-product[1]-1000*(1-x[1]) - x[16]<= 0);
c.add(x[12]+time[2]-1000*(1-x[2]) - x[15]<= 0);
c.add(x[18]+product[2]-1000*(1-x[2]) - x[21]<= 0);
c.add(x[21]-product[2]-1000*(1-x[2]) - x[18]<= 0);
c.add(x[14]+time[3]-1000*(1-x[3]) - x[11]<= 0);
c.add(x[20]+product[3]-1000*(1-x[3]) - x[17]<= 0);
c.add(x[17]-product[3]-1000*(1-x[3]) - x[20]<= 0);
c.add(x[13]+time[6]-1000*(1-x[6]) - x[15]<= 0);
c.add(x[19]+product[6]-1000*(1-x[6]) - x[21]<= 0);
c.add(x[10]+time[7]-1000*(1-x[7]) - x[13]<= 0);
c.add(x[16]+product[7]-1000*(1-x[7]) - x[19]<= 0);
c.add(x[12]+time[9]-1000*(1-x[9]) - x[13]<= 0);
c.add(x[18]+product[9]-1000*(1-x[9]) - x[19]<= 0);
c.add(x[11]+time[4]-1000*(1-x[4]) - x[10] <=0);
c.add(x[17]+product[4]-1000*(1-x[4]) - x[16]<= 0);
c.add(x[16]-product[4]-1000*(1-x[4]) - x[17]<= 0);
c.add(x[11]+time[8]-1000*(1-x[8]) - x[12]<= 0);
c.add(x[17]+product[8]-1000*(1-x[8]) - x[18]<= 0);
c.add(x[11]+time[5]-1000*(1-x[5]) - x[13]<= 0);
c.add(x[17]+product[5]-1000*(1-x[5]) - x[19]<= 0);
c.add(5 <= x[10] <= 7);
c.add(2 <= x[11] <= 5);
c.add(5 <= x[12] <= 9);
c.add(0 <= x[13] <= 20);
c.add(0 <= x[14] <= 0);
c.add(0 <= x[15] <= 14);
/*
c.add(2 <= x[17] <= 4);
c.add(0 <= x[16] <= 2);
c.add(3 <= x[18] <= 4);
c.add(0 <= x[19] <= 1000);
//c.add(0 <= x[20] <= 0);*/
c.add(3 <= x[21] <= 4);
c.add( x[20] == 1);
model.add(c);
}
/** Create a new worker.
* The constructor mainly does the following:
* - create an IloCplex instance that refers to a remote worker,
* - load the model in <code>modelfile</code>,
* - setup parameters depending on this worker's index,
* - start an asynchronous solve.
* If anything fails then an exception will be thrown.
* @param env The environment used for instantiating Ilo* objects.
* @param i The index of the worker to be created. This also
* determines the parameter settings to use in this worker.
* @param s A pointer to the global solve state.
* @param transport The transport name for the IloCplex constructor.
* @param argc The argument count for the IloCplex constructor.
* @param argv The array of transport arguments for the IloCplex
* constructor.
* @param modelfile Name of the model to be loaded into the worker.
* @param output The output mode.
* @param objdiff The minimal difference between so that two
* consecutive objective function values are considered
* different.
*/
Worker(IloEnv env, int i, SolveState *s, char const *transport,
int argc, char const **argv, char const *modelfile,
OUTPUT output, double objdiff)
: idx(i), state(s), model(env), cplex(0), handle(0),
primal(IloInfinity), dual(-IloInfinity),
obj(env), x(env), rng(env), infoHandler(this), outb(idx), outs(&outb)
{
try {
// Create remote object, setup output and load the model.
cplex = IloCplex(model, transport, argc, argv);
switch (output) {
case OUTPUT_SILENT:
// Disable output on the output and warning stream.
cplex.setOut(env.getNullStream());
cplex.setWarning(env.getNullStream());
break;
case OUTPUT_PREFIXED:
// Redirect output to our custom stream.
cplex.setOut(outs);
cplex.setWarning(outs);
break;
case OUTPUT_LOG:
// Nothing to do here. By default output is enabled.
break;
}
cplex.importModel(model, modelfile, obj, x, rng);
if ( obj.getSense() == IloObjective::Minimize ) {
primal = -IloInfinity;
dual = IloInfinity;
}
// We set the thread count for each solver to 1 so that we do not
// run into problems if multiple solves are performed on the same
// machine.
cplex.setParam(IloCplex::Param::Threads, 1);
// Each worker runs with a different random seed. This way we
// get different paths through the tree even if the other
// parameter settings are the same.
cplex.setParam(IloCplex::Param::RandomSeed, idx);
// Apply parameter settings.
for (class ParamValue const *vals = settings[idx % NUMSETTINGS].values;
vals->isValid(); ++vals)
vals->apply(cplex);
// Install callback and set objective change.
int status = cplex.userfunction (USERACTION_ADDCALLBACK,
0, NULL, 0, 0, NULL);
if ( status )
throw status;
IloCplex::Serializer s;
s.add(objdiff);
status = cplex.userfunction (USERACTION_CHANGEOBJDIFF,
s.getRawLength(), s.getRawData(),
0, 0, NULL);
if ( status )
throw status;
// Register the handler that will process info messages sent
// from the worker.
cplex.setRemoteInfoHandler(&infoHandler);
// Everything is setup. Launch the asynchronous solve.
handle = cplex.solve(true);
} catch (...) {
// In case of an exception we need to take some special
// cleanup actions. Note that if we get here then the
// solve cannot have been started and we don't need to
// kill or join the asynchronous solve.
if ( cplex.getImpl() )
cplex.end();
rng.end();
x.end();
obj.end();
model.end();
throw;
}
}
// Test KKT conditions on the solution.
// The function returns true if the tested KKT conditions are satisfied
// and false otherwise.
// The function assumes that the model currently extracted to CPLEX is fully
// described by obj, vars and rngs.
static bool
checkkkt (IloCplex& cplex, IloObjective const& obj, IloNumVarArray const& vars,
IloRangeArray const& rngs, IloIntArray const& cone, double tol)
{
IloEnv env = cplex.getEnv();
IloModel model = cplex.getModel();
IloNumArray x(env), dslack(env);
IloNumArray pi(env, rngs.getSize()), slack(env);
// Read primal and dual solution information.
cplex.getValues(x, vars);
cplex.getSlacks(slack, rngs);
// pi for second order cone constraints.
getsocpconstrmultipliers(cplex, vars, rngs, pi, dslack);
// pi for linear constraints.
for (IloInt i = 0; i < rngs.getSize(); ++i) {
IloRange r = rngs[i];
if ( !r.getQuadIterator().ok() )
pi[idx(r)] = cplex.getDual(r);
}
// Print out the data we just fetched.
streamsize oprec = env.out().precision(3);
ios_base::fmtflags oflags = env.out().setf(ios::fixed | ios::showpos);
env.out() << "x = [";
for (IloInt i = 0; i < x.getSize(); ++i)
env.out() << " " << x[i];
env.out() << " ]" << endl;
env.out() << "dslack = [";
for (IloInt i = 0; i < dslack.getSize(); ++i)
env.out() << " " << dslack[i];
env.out() << " ]" << endl;
env.out() << "pi = [";
for (IloInt i = 0; i < rngs.getSize(); ++i)
env.out() << " " << pi[i];
env.out() << " ]" << endl;
env.out() << "slack = [";
for (IloInt i = 0; i < rngs.getSize(); ++i)
env.out() << " " << slack[i];
env.out() << " ]" << endl;
env.out().precision(oprec);
env.out().flags(oflags);
// Test primal feasibility.
// This example illustrates the use of dual vectors returned by CPLEX
// to verify dual feasibility, so we do not test primal feasibility
// here.
// Test dual feasibility.
// We must have
// - for all <= constraints the respective pi value is non-negative,
// - for all >= constraints the respective pi value is non-positive,
// - the dslack value for all non-cone variables must be non-negative.
// Note that we do not support ranged constraints here.
for (IloInt i = 0; i < vars.getSize(); ++i) {
IloNumVar v = vars[i];
if ( cone[i] == NOT_IN_CONE && dslack[i] < -tol ) {
env.error() << "Dual multiplier for " << v << " is not feasible: "
<< dslack[i] << endl;
return false;
}
}
for (IloInt i = 0; i < rngs.getSize(); ++i) {
IloRange r = rngs[i];
if ( fabs (r.getLB() - r.getUB()) <= tol ) {
// Nothing to check for equality constraints.
}
else if ( r.getLB() > -IloInfinity && pi[i] > tol ) {
env.error() << "Dual multiplier " << pi[i] << " for >= constraint"
<< endl << r << endl
<< "not feasible"
<< endl;
return false;
}
else if ( r.getUB() < IloInfinity && pi[i] < -tol ) {
env.error() << "Dual multiplier " << pi[i] << " for <= constraint"
<< endl << r << endl
<< "not feasible"
<< endl;
return false;
}
}
// Test complementary slackness.
// For each constraint either the constraint must have zero slack or
// the dual multiplier for the constraint must be 0. We must also
// consider the special case in which a variable is not explicitly
// contained in a second order cone constraint.
for (IloInt i = 0; i < vars.getSize(); ++i) {
if ( cone[i] == NOT_IN_CONE ) {
if ( fabs(x[i]) > tol && dslack[i] > tol ) {
env.error() << "Invalid complementary slackness for " << vars[i]
<< ":" << endl
<< " " << x[i] << " and " << dslack[i]
<< endl;
return false;
}
}
}
for (IloInt i = 0; i < rngs.getSize(); ++i) {
if ( fabs(slack[i]) > tol && fabs(pi[i]) > tol ) {
env.error() << "Invalid complementary slackness for "
<< endl << rngs[i] << ":" << endl
<< " " << slack[i] << " and " << pi[i]
<< endl;
return false;
}
}
// Test stationarity.
// We must have
// c - g[i]'(X)*pi[i] = 0
// where c is the objective function, g[i] is the i-th constraint of the
// problem, g[i]'(x) is the derivate of g[i] with respect to x and X is the
// optimal solution.
// We need to distinguish the following cases:
// - linear constraints g(x) = ax - b. The derivative of such a
// constraint is g'(x) = a.
// - second order constraints g(x[1],...,x[n]) = -x[1] + |(x[2],...,x[n])|
// the derivative of such a constraint is
// g'(x) = (-1, x[2]/|(x[2],...,x[n])|, ..., x[n]/|(x[2],...,x[n])|
// (here |.| denotes the Euclidean norm).
// - bound constraints g(x) = -x for variables that are not explicitly
// contained in any second order cone constraint. The derivative for
// such a constraint is g'(x) = -1.
// Note that it may happen that the derivative of a second order cone
// constraint is not defined at the optimal solution X (this happens if
// X=0). In this case we just skip the stationarity test.
IloNumArray sum(env, vars.getSize());
for (IloExpr::LinearIterator it = obj.getLinearIterator(); it.ok(); ++it)
sum[idx(it.getVar())] = it.getCoef();
for (IloInt i = 0; i < vars.getSize(); ++i) {
IloNumVar v = vars[i];
if ( cone[i] == NOT_IN_CONE )
sum[i] -= dslack[i];
}
for (IloInt i = 0; i < rngs.getSize(); ++i) {
IloRange r = rngs[i];
if ( r.getQuadIterator().ok() ) {
// Quadratic (second order cone) constraint.
IloNum norm = 0.0;
for (IloExpr::QuadIterator q = r.getQuadIterator(); q.ok(); ++q) {
if ( q.getCoef() > 0 )
norm += x[idx(q.getVar1())] * x[idx(q.getVar1())];
}
norm = sqrt(norm);
if ( fabs(norm) <= tol ) {
// Derivative is not defined. Skip test.
env.warning() << "Cannot test stationarity at non-differentiable point."
<< endl;
return true;
}
else {
for (IloExpr::QuadIterator q = r.getQuadIterator(); q.ok(); ++q) {
if ( q.getCoef() < 0 )
sum[idx(q.getVar1())] -= pi[i];
else
sum[idx(q.getVar1())] += pi[i] * x[idx(q.getVar1())] / norm;
}
}
}
else {
// Linear constraint.
for (IloExpr::LinearIterator l = r.getLinearIterator(); l.ok(); ++l)
sum[idx(l.getVar())] -= pi[i] * l.getCoef();
}
}
// Now test that all elements in sum[] are 0.
for (IloInt i = 0; i < vars.getSize(); ++i) {
if ( fabs(sum[i]) > tol ) {
env.error() << "Invalid stationarity " << sum[i] << " for "
<< vars[i] << endl;
return false;
}
}
return true;
}
/** Extract the master block from <code>problem</code>.
* The constructor also sets up the solver for the newly created master
* block. The master block can only be extracted if all sub-blocks have
* already been extracted.
* @param problem The problem from which to extract the master.
* @param blocks The sub blocks that have already been extracted.
*/
BendersOpt::Block::Block(Problem const *problem, BlockVector const &blocks)
: env(), number(-1), vars(0), rows(0), cplex(0), cb(0)
{
IloNumVarArray problemVars = problem->getVariables();
IloRangeArray problemRanges = problem->getRows();
IloExpr masterObj(env);
IloNumVarArray masterVars(env);
IloRangeArray masterRows(env);
// Find columns that do not intersect block variables and
// copy them to the master block.
IdxMap idxMap;
RowSet rowSet;
for (IloInt j = 0; j < problemVars.getSize(); ++j) {
IloNumVar x = problemVars[j];
if ( problem->getBlock(x) < 0 ) {
// Column is not in a block. Copy it to the master.
IloNumVar v(env, x.getLB(), x.getUB(), x.getType(), x.getName());
varMap.insert(VarMap::value_type(v, x));
masterObj += problem->getObjCoef(x) * v;
idxMap[x] = masterVars.getSize();
masterVars.add(v);
}
else {
// Column is in a block. Collect all rows that intersect
// this column.
RowSet const &intersected = problem->getIntersectedRows(x);
for (RowSet::const_iterator it = intersected.begin();
it != intersected.end(); ++it)
rowSet.insert(*it);
idxMap[x] = -1;
}
}
// Pick up the rows that we need to copy.
// These are the rows that are only intersected by master variables,
// that is, the rows that are not in any block's rowset.
for (IloInt i = 0; i < problemRanges.getSize(); ++i) {
IloRange r = problemRanges[i];
if ( rowSet.find(r) == rowSet.end() ) {
IloRange masterRow(env, r.getLB(), r.getUB(), r.getName());
IloExpr lhs(env);
for (IloExpr::LinearIterator it = r.getLinearIterator(); it.ok(); ++it)
{
lhs += it.getCoef() * masterVars[idxMap[it.getVar()]];
}
masterRow.setExpr(lhs);
masterRows.add(masterRow);
}
}
// Adjust variable indices in blocks so that reference to variables
// in the original problem become references to variables in the master.
for (BlockVector::const_iterator b = blocks.begin(); b != blocks.end(); ++b) {
for (std::vector<FixData>::iterator it = (*b)->fixed.begin(); it != (*b)->fixed.end(); ++it)
it->col = idxMap[problemVars[it->col]];
}
// Create the eta variables, one for each block.
// See the comments at the top of this file for details about the
// eta variables.
IloInt const firsteta = masterVars.getSize();
for (BlockVector::size_type i = 0; i < blocks.size(); ++i) {
std::stringstream s;
s << "_eta" << i;
IloNumVar eta(env, 0.0, IloInfinity, s.str().c_str());
masterObj += eta;
masterVars.add(eta);
}
// Create model and solver instance
vars = masterVars;
rows = masterRows;
IloModel model(env);
model.add(obj = IloObjective(env, masterObj, problem->getObjSense()));
model.add(vars);
model.add(rows);
cplex = IloCplex(model);
cplex.use(cb = new (env) LazyConstraintCallback(env, this, blocks,
firsteta));
for (IloExpr::LinearIterator it = obj.getLinearIterator(); it.ok(); ++it)
objMap.insert(ObjMap::value_type(it.getVar(), it.getCoef()));
}
ILOSTLBEGIN
void makeCuts(IloRangeArray cuts, const IloNumVarArray& vars) {
IloNumVar x11, x12, x13, x14, x15;
IloNumVar x21, x22, x23, x24, x25;
IloNumVar x31, x32, x33, x34, x35;
IloNumVar x41, x42, x43, x44, x45;
IloNumVar x51, x52, x53, x54, x55;
IloNumVar w11, w12, w13, w14, w15;
IloNumVar w21, w22, w23, w24, w25;
IloNumVar w31, w32, w33, w34, w35;
IloNumVar w41, w42, w43, w44, w45;
IloNumVar w51, w52, w53, w54, w55;
IloInt num = vars.getSize();
for (IloInt i = 0; i < num; i++) {
if ( strcmp(vars[i].getName(), "X11") == 0 ) x11 = vars[i];
else if ( strcmp(vars[i].getName(), "X12") == 0 ) x12 = vars[i];
else if ( strcmp(vars[i].getName(), "X13") == 0 ) x13 = vars[i];
else if ( strcmp(vars[i].getName(), "X14") == 0 ) x14 = vars[i];
else if ( strcmp(vars[i].getName(), "X15") == 0 ) x15 = vars[i];
else if ( strcmp(vars[i].getName(), "X21") == 0 ) x21 = vars[i];
else if ( strcmp(vars[i].getName(), "X22") == 0 ) x22 = vars[i];
else if ( strcmp(vars[i].getName(), "X23") == 0 ) x23 = vars[i];
else if ( strcmp(vars[i].getName(), "X24") == 0 ) x24 = vars[i];
else if ( strcmp(vars[i].getName(), "X25") == 0 ) x25 = vars[i];
else if ( strcmp(vars[i].getName(), "X31") == 0 ) x31 = vars[i];
else if ( strcmp(vars[i].getName(), "X32") == 0 ) x32 = vars[i];
else if ( strcmp(vars[i].getName(), "X33") == 0 ) x33 = vars[i];
else if ( strcmp(vars[i].getName(), "X34") == 0 ) x34 = vars[i];
else if ( strcmp(vars[i].getName(), "X35") == 0 ) x35 = vars[i];
else if ( strcmp(vars[i].getName(), "X41") == 0 ) x41 = vars[i];
else if ( strcmp(vars[i].getName(), "X42") == 0 ) x42 = vars[i];
else if ( strcmp(vars[i].getName(), "X43") == 0 ) x43 = vars[i];
else if ( strcmp(vars[i].getName(), "X44") == 0 ) x44 = vars[i];
else if ( strcmp(vars[i].getName(), "X45") == 0 ) x45 = vars[i];
else if ( strcmp(vars[i].getName(), "X51") == 0 ) x51 = vars[i];
else if ( strcmp(vars[i].getName(), "X52") == 0 ) x52 = vars[i];
else if ( strcmp(vars[i].getName(), "X53") == 0 ) x53 = vars[i];
else if ( strcmp(vars[i].getName(), "X54") == 0 ) x54 = vars[i];
else if ( strcmp(vars[i].getName(), "X55") == 0 ) x55 = vars[i];
else if ( strcmp(vars[i].getName(), "W11") == 0 ) w11 = vars[i];
else if ( strcmp(vars[i].getName(), "W12") == 0 ) w12 = vars[i];
else if ( strcmp(vars[i].getName(), "W13") == 0 ) w13 = vars[i];
else if ( strcmp(vars[i].getName(), "W14") == 0 ) w14 = vars[i];
else if ( strcmp(vars[i].getName(), "W15") == 0 ) w15 = vars[i];
else if ( strcmp(vars[i].getName(), "W21") == 0 ) w21 = vars[i];
else if ( strcmp(vars[i].getName(), "W22") == 0 ) w22 = vars[i];
else if ( strcmp(vars[i].getName(), "W23") == 0 ) w23 = vars[i];
else if ( strcmp(vars[i].getName(), "W24") == 0 ) w24 = vars[i];
else if ( strcmp(vars[i].getName(), "W25") == 0 ) w25 = vars[i];
else if ( strcmp(vars[i].getName(), "W31") == 0 ) w31 = vars[i];
else if ( strcmp(vars[i].getName(), "W32") == 0 ) w32 = vars[i];
else if ( strcmp(vars[i].getName(), "W33") == 0 ) w33 = vars[i];
else if ( strcmp(vars[i].getName(), "W34") == 0 ) w34 = vars[i];
else if ( strcmp(vars[i].getName(), "W35") == 0 ) w35 = vars[i];
else if ( strcmp(vars[i].getName(), "W41") == 0 ) w41 = vars[i];
else if ( strcmp(vars[i].getName(), "W42") == 0 ) w42 = vars[i];
else if ( strcmp(vars[i].getName(), "W43") == 0 ) w43 = vars[i];
else if ( strcmp(vars[i].getName(), "W44") == 0 ) w44 = vars[i];
else if ( strcmp(vars[i].getName(), "W45") == 0 ) w45 = vars[i];
else if ( strcmp(vars[i].getName(), "W51") == 0 ) w51 = vars[i];
else if ( strcmp(vars[i].getName(), "W52") == 0 ) w52 = vars[i];
else if ( strcmp(vars[i].getName(), "W53") == 0 ) w53 = vars[i];
else if ( strcmp(vars[i].getName(), "W54") == 0 ) w54 = vars[i];
else if ( strcmp(vars[i].getName(), "W55") == 0 ) w55 = vars[i];
}
cuts.add(x21 - x22 <= 0);
cuts[0].setName("cut0");
cuts.add(x22 - x23 <= 0);
cuts.add(x23 - x24 <= 0);
cuts.add(2.08*x11 + 2.98*x21 + 3.47*x31 + 2.24*x41 + 2.08*x51 + 0.25*w11 +
0.25*w21 + 0.25*w31 + 0.25*w41 + 0.25*w51 <= 20.25);
cuts.add(2.08*x12 + 2.98*x22 + 3.47*x32 + 2.24*x42 + 2.08*x52 + 0.25*w12 +
0.25*w22 + 0.25*w32 + 0.25*w42 + 0.25*w52 <= 20.25);
cuts.add(2.08*x13 + 2.98*x23 + 3.4722*x33 + 2.24*x43 + 2.08*x53 + 0.25*w13
+ 0.25*w23 + 0.25*w33 + 0.25*w43 + 0.25*w53 <= 20.25);
cuts.add(2.08*x14 + 2.98*x24 + 3.47*x34 + 2.24*x44 + 2.08*x54 + 0.25*w14 +
0.25*w24 + 0.25*w34 + 0.25*w44 + 0.25*w54 <= 20.25);
cuts.add(2.08*x15 + 2.98*x25 + 3.47*x35 + 2.24*x45 + 2.08*x55 + 0.25*w15 +
0.25*w25 + 0.25*w35 + 0.25*w45 + 0.25*w55 <= 16.25);
}
/** Create the dual of a linear program.
* The function can only dualize programs of the form
* <code>Ax <= b, x >= 0</code>. The data in <code>primalVars</code> and
* <code>dualRows</code> as well as in <code>primalRows</code> and
* <code>dualVars</code> is in 1-to-1-correspondence.
* @param primalObj Objective function of primal problem.
* @param primalVars Variables in primal problem.
* @param primalRows Rows in primal problem.
* @param dualObj Objective function of dual will be stored here.
* @param dualVars All dual variables will be stored here.
* @param dualRows All dual rows will be stored here.
*/
void BendersOpt::makeDual(IloObjective const &primalObj,
IloNumVarArray const &primalVars,
IloRangeArray const &primalRows,
IloObjective *dualObj,
IloNumVarArray *dualVars,
IloRangeArray *dualRows)
{
// To keep the code simple we only support problems
// of the form Ax <= b, b >= 0 here. We leave it as a reader's
// exercise to extend the function to something that can handle
// any kind of linear model.
for (IloInt j = 0; j < primalVars.getSize(); ++j)
if ( primalVars[j].getLB() != 0 ||
primalVars[j].getUB() < IloInfinity )
{
std::stringstream s;
s << "Cannot dualize variable " << primalVars[j];
throw s.str();
}
for (IloInt i = 0; i < primalRows.getSize(); ++i)
if ( primalRows[i].getLB() > -IloInfinity ||
primalRows[i].getUB() >= IloInfinity )
{
std::stringstream s;
s << "Cannot dualize constraint " << primalRows[i];
std::cerr << s.str() << std::endl;
throw s.str();
}
// The dual of
// min/max c^T x
// Ax <= b
// x >= 0
// is
// max/min y^T b
// y^T A <= c
// y <= 0
// We scale y by -1 to get >= 0
IloEnv env = primalVars.getEnv();
IloObjective obj(env, 0.0,
primalObj.getSense() == IloObjective::Minimize ?
IloObjective::Maximize : IloObjective::Minimize);
IloRangeArray rows(env);
IloNumVarArray y(env);
std::map<IloNumVar,IloInt,ExtractableLess<IloNumVar> > v2i;
for (IloInt j = 0; j < primalVars.getSize(); ++j) {
IloNumVar x = primalVars[j];
v2i.insert(std::map<IloNumVar,IloInt,ExtractableLess<IloNumVar> >::value_type(x, j));
rows.add(IloRange(env, -IloInfinity, 0, x.getName()));
}
for (IloExpr::LinearIterator it = primalObj.getLinearIterator(); it.ok(); ++it)
rows[v2i[it.getVar()]].setUB(it.getCoef());
for (IloInt i = 0; i < primalRows.getSize(); ++i) {
IloRange r = primalRows[i];
IloNumColumn col(env);
col += obj(-r.getUB());
for (IloExpr::LinearIterator it = r.getLinearIterator(); it.ok(); ++it)
col += rows[v2i[it.getVar()]](-it.getCoef());
y.add(IloNumVar(col, 0, IloInfinity, IloNumVar::Float, r.getName()));
}
*dualObj = obj;
*dualVars = y;
*dualRows = rows;
}
/** Solve the <code>problem</code> using a distributed implementation of
* Benders' decomposition.
*/
bool
BendersOpt::solve (Problem const *problem,
int argc, char const *const *argv,
std::vector<char const *> const &machines)
{
IloEnv env = problem->getModel().getEnv();
std::vector<Block *> blocks;
Block *master = 0;
bool result = false;
try {
// Extract blocks and master problem.
std::cout << "Extracting " << problem->getNBlocks() << " blocks."
<< std::endl;
for (IloInt b = 0; b < problem->getNBlocks(); ++b)
blocks.push_back(new Block(problem, b, argc, argv, machines));
master = new Block(problem, blocks);
// Write out the master and all blocks (for debugging).
master->cplex.exportModel("master.lp");
for (BlockVector::size_type b = 0; b < blocks.size(); ++b) {
std::stringstream s;
s << "block" << b << ".lp";
blocks[b]->cplex.exportModel(s.str().c_str());
}
// Solve the master.
// If we find a feasible solution then perform a final solve
// with the original problem so that we get solution values for
// all variables in the original problem.
if ( master->cplex.solve() ) {
// Perform a last solve to get the solution values.
IloNumArray vals(env), startVals(env);
IloNumVarArray startVars(env);
master->cplex.getValues(vals, master->vars);
IloCplex cplex(problem->getModel());
// Fix integral variables to their value in the master solution
// and add them as MIP start.
for (IloInt i = 0; i < vals.getSize(); ++i)
if ( master->vars[i].getType() != IloNumVar::Float ) {
double const v = IloRound(vals[i]);
IloNumVar x = master->varMap[master->vars[i]];
startVals.add(v);
startVars.add(x);
// We add lazy constraints so as to make sure that
// - we don't modify the original model
// - the problem has only the unique optimal solution
// we are interested in
cplex.addLazyConstraint(x == v);
}
cplex.addMIPStart(startVars, startVals);
cplex.solve();
// Report the results.
std::cout << "#### Problem solved (" << cplex.getObjValue() << ", "
<< cplex.getCplexStatus() << ")." << std::endl;
IloNumVarArray vars = problem->getVariables();
for (IloInt i = 0; i < vars.getSize(); ++i)
std::cout << "#### \tx[" << i << "] = " << cplex.getValue(vars[i])
<< std::endl;
cplex.end();
result = true;
}
} catch (...) {
if ( master )
delete master;
while ( blocks.size() > 0 ) {
delete blocks.back();
blocks.pop_back();
}
throw;
}
delete master;
while ( blocks.size() > 0 ) {
delete blocks.back();
blocks.pop_back();
}
return result;
}
ILOBRANCHCALLBACK1(SOSbranch, IloSOS1Array, sos) {
IloNumArray x;
IloNumVarArray var;
try {
IloInt i;
x = IloNumArray(getEnv());
var = IloNumVarArray(getEnv());
IloNum bestx = EPS;
IloInt besti = -1;
IloInt bestj = -1;
IloInt num = sos.getSize();
for (i = 0; i < num; i++) {
if ( getFeasibility(sos[i]) == Infeasible ) {
var.clear();
sos[i].getVariables(var);
getValues(x, var);
IloInt n = var.getSize();
for (IloInt j = 0; j < n; j++) {
IloNum inf = IloAbs(x[j] - IloRound(x[j]));
if ( inf > bestx ) {
bestx = inf;
besti = i;
bestj = j;
}
}
}
}
if ( besti >= 0 ) {
IloCplex::BranchDirectionArray dir;
IloNumArray val;
try {
dir = IloCplex::BranchDirectionArray(getEnv());
val = IloNumArray(getEnv());
var.clear();
sos[besti].getVariables(var);
IloInt n = var.getSize();
for (IloInt j = 0; j < n; j++) {
if ( j != bestj ) {
dir.add(IloCplex::BranchDown);
val.add(0.0);
} else {
dir.add(IloCplex::BranchUp);
val.add(1.0);
}
}
makeBranch(var, val, dir, getObjValue());
makeBranch(var[bestj], 0.0, IloCplex::BranchDown, getObjValue());
}
catch (...) {
dir.end();
val.end();
throw;
}
dir.end();
val.end();
}
}
catch (...) {
var.end();
x.end();
throw;
}
var.end();
x.end();
}