int main(int narg, char* args[]) {
Kokkos::initialize(narg,args);
// Produce some 3D random data (see Algorithms/01_random_numbers for more info)
Kokkos::View<int***,Kokkos::LayoutRight> data("Data",512,512,32);
Kokkos::Random_XorShift64_Pool<> rand_pool64(5374857);
Kokkos::fill_random(data,rand_pool64,100);
// A global value to put the result in
Kokkos::View<int> gsum("Sum");
// Each team handles a slice of the data
// Set up TeamPolicy with 512 teams with maximum number of threads per team and 16 vector lanes.
// The team_size_max function will determine the maximum number of threads taking into account
// shared memory requirements of the Functor.
// The maximum vector length is hardware dependent but can always be smaller than the hardware allows.
// The vector length must be a power of 2.
const Kokkos::TeamPolicy<> policy( 512 , Kokkos::TeamPolicy<>::team_size_max(SomeCorrelation(data,gsum)) , 16);
Kokkos::parallel_for( policy , SomeCorrelation(data,gsum) );
Kokkos::fence();
// Copy result value back
int sum = 0;
Kokkos::deep_copy(sum,gsum);
printf("Result %i\n",sum);
Kokkos::finalize();
}
SEXP gmean(SEXP x, SEXP narm)
{
SEXP ans;
int i, protecti=0, thisgrp, n;
//clock_t start = clock();
if (!isLogical(narm) || LENGTH(narm)!=1 || LOGICAL(narm)[0]==NA_LOGICAL) error("na.rm must be TRUE or FALSE");
if (!isVectorAtomic(x)) error("GForce mean can only be applied to columns, not .SD or similar. Likely you're looking for 'DT[,lapply(.SD,mean),by=,.SDcols=]'. See ?data.table.");
if (!LOGICAL(narm)[0]) {
ans = PROTECT(gsum(x,narm)); protecti++;
switch(TYPEOF(ans)) {
case LGLSXP: case INTSXP:
ans = PROTECT(coerceVector(ans, REALSXP)); protecti++;
case REALSXP:
for (i=0; i<ngrp; i++) REAL(ans)[i] /= grpsize[i]; // let NA propogate
break;
default :
error("Internal error: gsum returned type '%s'. typeof(x) is '%s'", type2char(TYPEOF(ans)), type2char(TYPEOF(x)));
}
UNPROTECT(protecti);
return(ans);
}
// na.rm=TRUE. Similar to gsum, but we need to count the non-NA as well for the divisor
n = LENGTH(x);
if (grpn != n) error("grpn [%d] != length(x) [%d] in gsum", grpn, length(x));
long double *s = malloc(ngrp * sizeof(long double));
if (!s) error("Unable to allocate %d * %d bytes for sum in gmean na.rm=TRUE", ngrp, sizeof(long double));
memset(s, 0, ngrp * sizeof(long double)); // all-0 bits == (long double)0, checked in init.c
int *c = malloc(ngrp * sizeof(int));
if (!c) error("Unable to allocate %d * %d bytes for counts in gmean na.rm=TRUE", ngrp, sizeof(int));
memset(c, 0, ngrp * sizeof(int)); // all-0 bits == (int)0, checked in init.c
switch(TYPEOF(x)) {
case LGLSXP: case INTSXP:
for (i=0; i<n; i++) {
thisgrp = grp[i];
if(INTEGER(x)[i] == NA_INTEGER) continue;
s[thisgrp] += INTEGER(x)[i]; // no under/overflow here, s is long double
c[thisgrp]++;
}
break;
case REALSXP:
for (i=0; i<n; i++) {
thisgrp = grp[i];
if (ISNAN(REAL(x)[i])) continue;
s[thisgrp] += REAL(x)[i];
c[thisgrp]++;
}
break;
default:
free(s); free(c);
error("Type '%s' not supported by GForce mean (gmean) na.rm=TRUE. Either add the prefix base::mean(.) or turn off GForce optimization using options(datatable.optimize=1)", type2char(TYPEOF(x)));
}
ans = PROTECT(allocVector(REALSXP, ngrp));
for (i=0; i<ngrp; i++) {
if (c[i]==0) { REAL(ans)[i] = R_NaN; continue; } // NaN to follow base::mean
s[i] /= c[i];
if (s[i] > DBL_MAX) REAL(ans)[i] = R_PosInf;
else if (s[i] < -DBL_MAX) REAL(ans)[i] = R_NegInf;
else REAL(ans)[i] = (double)s[i];
}
free(s); free(c);
UNPROTECT(1);
// Rprintf("this gmean na.rm=TRUE took %8.3f\n", 1.0*(clock()-start)/CLOCKS_PER_SEC);
return(ans);
}
int main(int argc, char *argv[]) {
// feenableexcept(FE_ALL_EXCEPT & ~FE_INEXACT);
// This little trick lets us print to std::cout only if a (dummy) command-line argument is provided.
int iprint = argc - 1;
Teuchos::RCP<std::ostream> outStream;
Teuchos::oblackholestream bhs; // outputs nothing
/*** Initialize communicator. ***/
Teuchos::GlobalMPISession mpiSession (&argc, &argv, &bhs);
Teuchos::RCP<const Teuchos::Comm<int> > comm = Tpetra::DefaultPlatform::getDefaultPlatform().getComm();
const int myRank = comm->getRank();
if ((iprint > 0) && (myRank == 0)) {
outStream = Teuchos::rcp(&std::cout, false);
}
else {
outStream = Teuchos::rcp(&bhs, false);
}
int errorFlag = 0;
// *** Example body.
try {
/*** Read in XML input ***/
std::string filename = "input.xml";
Teuchos::RCP<Teuchos::ParameterList> parlist = Teuchos::rcp( new Teuchos::ParameterList() );
Teuchos::updateParametersFromXmlFile( filename, parlist.ptr() );
std::string stoch_filename = "stochastic.xml";
Teuchos::RCP<Teuchos::ParameterList> stoch_parlist = Teuchos::rcp( new Teuchos::ParameterList() );
Teuchos::updateParametersFromXmlFile( stoch_filename, stoch_parlist.ptr() );
/*** Initialize main data structure. ***/
Teuchos::RCP<const Teuchos::Comm<int> > serial_comm = Teuchos::rcp(new Teuchos::SerialComm<int>());
Teuchos::RCP<ElasticitySIMPOperators<RealT> > data
= Teuchos::rcp(new ElasticitySIMPOperators<RealT>(serial_comm, parlist, outStream));
/*** Initialize density filter. ***/
Teuchos::RCP<DensityFilter<RealT> > filter
= Teuchos::rcp(new DensityFilter<RealT>(serial_comm, parlist, outStream));
/*** Build vectors and dress them up as ROL vectors. ***/
Teuchos::RCP<const Tpetra::Map<> > vecmap_u = data->getDomainMapA();
Teuchos::RCP<const Tpetra::Map<> > vecmap_z = data->getCellMap();
Teuchos::RCP<Tpetra::MultiVector<> > u_rcp = createTpetraVector(vecmap_u);
Teuchos::RCP<Tpetra::MultiVector<> > z_rcp = createTpetraVector(vecmap_z);
Teuchos::RCP<Tpetra::MultiVector<> > du_rcp = createTpetraVector(vecmap_u);
Teuchos::RCP<Tpetra::MultiVector<> > dw_rcp = createTpetraVector(vecmap_u);
Teuchos::RCP<Tpetra::MultiVector<> > dz_rcp = createTpetraVector(vecmap_z);
Teuchos::RCP<Tpetra::MultiVector<> > dz2_rcp = createTpetraVector(vecmap_z);
Teuchos::RCP<std::vector<RealT> > vc_rcp = Teuchos::rcp(new std::vector<RealT>(1, 0));
Teuchos::RCP<std::vector<RealT> > vc_lam_rcp = Teuchos::rcp(new std::vector<RealT>(1, 0));
Teuchos::RCP<std::vector<RealT> > vscale_rcp = Teuchos::rcp(new std::vector<RealT>(1, 0));
// Set all values to 1 in u, z.
RealT one(1), two(2);
u_rcp->putScalar(one);
// Set z to gray solution.
RealT volFrac = parlist->sublist("ElasticityTopoOpt").get<RealT>("Volume Fraction");
z_rcp->putScalar(volFrac);
// Set scaling vector for constraint
RealT W = parlist->sublist("Geometry").get<RealT>("Width");
RealT H = parlist->sublist("Geometry").get<RealT>("Height");
(*vscale_rcp)[0] = one/std::pow(W*H*(one-volFrac),two);
// Set Scaling vector for density
bool useZscale = parlist->sublist("Problem").get<bool>("Use Scaled Density Vectors");
RealT densityScaling = parlist->sublist("Problem").get<RealT>("Density Scaling");
Teuchos::RCP<Tpetra::MultiVector<> > scaleVec = createTpetraVector(vecmap_z);
scaleVec->putScalar(densityScaling);
if ( !useZscale ) {
scaleVec->putScalar(one);
}
Teuchos::RCP<const Tpetra::Vector<> > zscale_rcp = scaleVec->getVector(0);
// Randomize d vectors.
du_rcp->randomize(); //du_rcp->scale(0);
dw_rcp->randomize();
dz_rcp->randomize(); //dz_rcp->scale(0);
dz2_rcp->randomize();
// Create ROL::TpetraMultiVectors.
Teuchos::RCP<ROL::Vector<RealT> > up
= Teuchos::rcp(new ROL::TpetraMultiVector<RealT>(u_rcp));
Teuchos::RCP<ROL::Vector<RealT> > dup
= Teuchos::rcp(new ROL::TpetraMultiVector<RealT>(du_rcp));
Teuchos::RCP<ROL::Vector<RealT> > dwp
= Teuchos::rcp(new ROL::TpetraMultiVector<RealT>(dw_rcp));
Teuchos::RCP<ROL::Vector<RealT> > zp
= Teuchos::rcp(new ROL::PrimalScaledTpetraMultiVector<RealT>(z_rcp,zscale_rcp));
Teuchos::RCP<ROL::Vector<RealT> > dzp
= Teuchos::rcp(new ROL::PrimalScaledTpetraMultiVector<RealT>(dz_rcp,zscale_rcp));
Teuchos::RCP<ROL::Vector<RealT> > dz2p
= Teuchos::rcp(new ROL::PrimalScaledTpetraMultiVector<RealT>(dz2_rcp,zscale_rcp));
Teuchos::RCP<ROL::Vector<RealT> > vcp
= Teuchos::rcp(new ROL::PrimalScaledStdVector<RealT>(vc_rcp,vscale_rcp));
Teuchos::RCP<ROL::Vector<RealT> > vc_lamp
= Teuchos::rcp(new ROL::DualScaledStdVector<RealT>(vc_lam_rcp,vscale_rcp));
// Create ROL SimOpt vectors.
ROL::Vector_SimOpt<RealT> x(up,zp);
ROL::Vector_SimOpt<RealT> d(dup,dzp);
ROL::Vector_SimOpt<RealT> d2(dwp,dz2p);
/*** Build sampler. ***/
BuildSampler<RealT> buildSampler(comm,*stoch_parlist,*parlist);
buildSampler.print("samples");
/*** Compute compliance objective function scaling. ***/
RealT min(ROL::ROL_INF<RealT>()), gmin(0), max(0), gmax(0), sum(0), gsum(0), tmp(0);
Teuchos::Array<RealT> dotF(1, 0);
RealT minDensity = parlist->sublist("ElasticitySIMP").get<RealT>("Minimum Density");
for (int i = 0; i < buildSampler.get()->numMySamples(); ++i) {
data->updateF(buildSampler.get()->getMyPoint(i));
(data->getVecF())->dot(*(data->getVecF()),dotF.view(0,1));
tmp = minDensity/dotF[0];
min = ((min < tmp) ? min : tmp);
max = ((max > tmp) ? max : tmp);
sum += buildSampler.get()->getMyWeight(i) * tmp;
}
ROL::Elementwise::ReductionMin<RealT> ROLmin;
buildSampler.getBatchManager()->reduceAll(&min,&gmin,ROLmin);
ROL::Elementwise::ReductionMax<RealT> ROLmax;
buildSampler.getBatchManager()->reduceAll(&max,&gmax,ROLmax);
buildSampler.getBatchManager()->sumAll(&sum,&gsum,1);
bool useExpValScale = stoch_parlist->sublist("Problem").get("Use Expected Value Scaling",false);
RealT scale = (useExpValScale ? gsum : gmin);
/*** Build objective function, constraint and reduced objective function. ***/
Teuchos::RCP<ROL::Objective_SimOpt<RealT> > obj
= Teuchos::rcp(new ParametrizedObjective_PDEOPT_ElasticitySIMP<RealT>(data, filter, parlist,scale));
Teuchos::RCP<ROL::EqualityConstraint_SimOpt<RealT> > con
= Teuchos::rcp(new ParametrizedEqualityConstraint_PDEOPT_ElasticitySIMP<RealT>(data, filter, parlist));
Teuchos::RCP<ROL::Reduced_Objective_SimOpt<RealT> > objReduced
= Teuchos::rcp(new ROL::Reduced_Objective_SimOpt<RealT>(obj, con, up, zp, dwp));
Teuchos::RCP<ROL::EqualityConstraint<RealT> > volcon
= Teuchos::rcp(new EqualityConstraint_PDEOPT_ElasticitySIMP_Volume<RealT>(data, parlist));
/*** Build bound constraint ***/
Teuchos::RCP<Tpetra::MultiVector<> > lo_rcp = Teuchos::rcp(new Tpetra::MultiVector<>(vecmap_z, 1, true));
Teuchos::RCP<Tpetra::MultiVector<> > hi_rcp = Teuchos::rcp(new Tpetra::MultiVector<>(vecmap_z, 1, true));
lo_rcp->putScalar(0.0);
hi_rcp->putScalar(1.0);
Teuchos::RCP<ROL::Vector<RealT> > lop
= Teuchos::rcp(new ROL::PrimalScaledTpetraMultiVector<RealT>(lo_rcp, zscale_rcp));
Teuchos::RCP<ROL::Vector<RealT> > hip
= Teuchos::rcp(new ROL::PrimalScaledTpetraMultiVector<RealT>(hi_rcp, zscale_rcp));
Teuchos::RCP<ROL::BoundConstraint<RealT> > bnd = Teuchos::rcp(new ROL::BoundConstraint<RealT>(lop,hip));
/*** Build Stochastic Functionality. ***/
ROL::StochasticProblem<RealT> opt(*stoch_parlist,objReduced,buildSampler.get(),zp,bnd);
/*** Check functional interface. ***/
bool checkDeriv = parlist->sublist("Problem").get("Derivative Check",false);
if ( checkDeriv ) {
// *outStream << "Checking Objective:" << "\n";
// obj->checkGradient(x,d,true,*outStream);
// obj->checkHessVec(x,d,true,*outStream);
// obj->checkHessSym(x,d,d2,true,*outStream);
// *outStream << "Checking Constraint:" << "\n";
// con->checkAdjointConsistencyJacobian(*dup,d,x,true,*outStream);
// con->checkAdjointConsistencyJacobian_1(*dwp, *dup, *up, *zp, true, *outStream);
// con->checkAdjointConsistencyJacobian_2(*dwp, *dzp, *up, *zp, true, *outStream);
// con->checkInverseJacobian_1(*up,*up,*up,*zp,true,*outStream);
// con->checkInverseAdjointJacobian_1(*up,*up,*up,*zp,true,*outStream);
// con->checkApplyJacobian(x,d,*up,true,*outStream);
// con->checkApplyAdjointHessian(x,*dup,d,x,true,*outStream);
*outStream << "Checking Reduced Objective:" << "\n";
opt.checkObjectiveGradient(*dzp,true,*outStream);
opt.checkObjectiveHessVec(*dzp,true,*outStream);
*outStream << "Checking Volume Constraint:" << "\n";
volcon->checkAdjointConsistencyJacobian(*vcp,*dzp,*zp,true,*outStream);
volcon->checkApplyJacobian(*zp,*dzp,*vcp,true,*outStream);
volcon->checkApplyAdjointHessian(*zp,*vcp,*dzp,*zp,true,*outStream);
}
/*** Run optimization ***/
ROL::AugmentedLagrangian<RealT> augLag(opt.getObjective(),volcon,*vc_lamp,1.0,
*opt.getSolutionVector(),*vcp,*parlist);
ROL::Algorithm<RealT> algo("Augmented Lagrangian",*parlist,false);
std::clock_t timer = std::clock();
algo.run(*opt.getSolutionVector(),*vc_lamp,augLag,*volcon,*opt.getBoundConstraint(),true,*outStream);
*outStream << "Optimization time: "
<< static_cast<RealT>(std::clock()-timer)/static_cast<RealT>(CLOCKS_PER_SEC)
<< " seconds." << std::endl;
data->outputTpetraVector(z_rcp, "density.txt");
data->outputTpetraVector(u_rcp, "state.txt");
data->outputTpetraVector(zscale_rcp, "weights.txt");
// Build objective function distribution
RealT val(0);
int nSamp = stoch_parlist->sublist("Problem").get("Number of Output Samples",10);
stoch_parlist->sublist("Problem").set("Number of Samples",nSamp);
BuildSampler<RealT> buildSampler_dist(comm,*stoch_parlist,*parlist);
std::stringstream name;
name << "samples_" << buildSampler_dist.getBatchManager()->batchID() << ".txt";
std::ofstream file;
file.open(name.str());
std::vector<RealT> sample;
RealT tol = 1.e-8;
std::clock_t timer_print = std::clock();
for (int i = 0; i < buildSampler_dist.get()->numMySamples(); ++i) {
sample = buildSampler_dist.get()->getMyPoint(i);
objReduced->setParameter(sample);
val = objReduced->value(*zp,tol);
for (int j = 0; j < static_cast<int>(sample.size()); ++j) {
file << sample[j] << " ";
}
file << val << "\n";
}
file.close();
*outStream << "Output time: "
<< static_cast<RealT>(std::clock()-timer_print)/static_cast<RealT>(CLOCKS_PER_SEC)
<< " seconds." << std::endl;
}
catch (std::logic_error err) {
*outStream << err.what() << "\n";
errorFlag = -1000;
}; // end try
if (errorFlag != 0)
std::cout << "End Result: TEST FAILED\n";
else
std::cout << "End Result: TEST PASSED\n";
return 0;
}
