void Utility::Yuv422FileSaver::savePlanar() {
for(std::size_t k=0; k<video_->getNumberOfFrames()&&isRunning_; k++) {
QVector<unsigned char> uBuffer;
QVector<unsigned char> vBuffer;
auto frame=video_->getFrame(k);
for(int i=0; i<width_*height_; i+=2) {
int y1=i/width_;
int x1=i%width_;
int y2=(i+1)/width_;
int x2=(i+1)%width_;
if(!frame->valid(x1,y1)||!frame->valid(x2,y2)) {
qDebug()<<"Wrong pixel coordinates";
continue;
}
auto vec=Rgb888ToYuv422(frame->pixel(x1,y1),frame->pixel(x2,y2));
dataStream_<<vec.getY1()<<vec.getY2();
uBuffer.push_back(vec.getU());
vBuffer.push_back(vec.getV());
}
while (!uBuffer.isEmpty()) {
dataStream_<<uBuffer.takeFirst();
}
while (!vBuffer.isEmpty()) {
dataStream_<<vBuffer.takeFirst();
}
}
}
void TestYuv444FileSaver::testRgb888ToYuv444() {
auto vector = Utility::Yuv444FileSaver::Rgb888ToYuv444(pixel);
QVERIFY(vector.getY() == 16);
QVERIFY(vector.getU() == 128);
QVERIFY(vector.getV() == 128);
}
void transformNormals(const Mat4& transformation){
const Vec4 u = transformation * getU();
const Vec4 v = transformation * getV();
const Vec4 n = transformation * getN();
setOrientation(u,v,n);
setPosition(position);
}
int GMSH_CutParametricPlugin::fillXYZ()
{
std::vector<std::string> expressions(3), variables(2);
for(int i = 0; i < 3; i++)
expressions[i] = CutParametricOptions_String[i].def;
variables[0] = "u";
variables[1] = "v";
mathEvaluator f(expressions, variables);
if(expressions.empty()) return 0;
int nbU = (int)CutParametricOptions_Number[2].def;
int nbV = (int)CutParametricOptions_Number[5].def;
x.resize(nbU * nbV);
y.resize(nbU * nbV);
z.resize(nbU * nbV);
std::vector<double> val(2), res(3);
for(int i = 0; i < nbU; ++i){
val[0] = getU(i);
for(int j = 0; j < nbV; ++j){
val[1] = getV(j);
if(f.eval(val, res)){
x[i * nbV + j] = res[0];
y[i * nbV + j] = res[1];
z[i * nbV + j] = res[2];
}
}
}
return 1;
}
void setPosition(const Vec4& pos){
position = pos;
const Vec4 eyeVector = position - Vec4::origin();
matrix.cArray[12] = -eyeVector * getU();
matrix.cArray[13] = -eyeVector * getV();
matrix.cArray[14] = -eyeVector * getN();
}
void computeU(const double *x, int xLen, double *u, int uLen)
{
assert(xLen == xDim && uDim == uLen);
setX(x, xLen);
computeU();
getU(u, uLen);
}
float BlockHighPass::update(float input)
{
float b = 2 * float(M_PI) * getFCut() * getDt();
float a = 1 / (1 + b);
setY(a * (getY() + input - getU()));
setU(input);
return getY();
}
float BlockIntegralTrap::update(float input)
{
// trapezoidal integration
setY(_limit.update(getY() +
(getU() + input) / 2.0f * getDt()));
setU(input);
return getY();
}
void TestYuv411FileSaver::testRgb888ToYuv411() {
auto vector = Utility::Yuv411FileSaver::Rgb888ToYuv411(pixel1,pixel2,pixel3,pixel4);
QVERIFY(vector.getY1() == 16);
QVERIFY(vector.getU() == 128);
QVERIFY(vector.getV() == 128);
QVERIFY(vector.getY2() == 16);
QVERIFY(vector.getY3() == 16);
QVERIFY(vector.getY4() == 16);
}
peano::applications::faxen::records::RegularGridCell peano::applications::faxen::records::RegularGridCellPacked::convert() const{
return RegularGridCell(
getP(),
getU(),
getV(),
getF(),
getG(),
getRhs(),
getRes(),
getIsInside()
);
}
bool SymBandSVDiv<T>::checkDecomp(
const BaseMatrix<T>& m, std::ostream* fout) const
{
Matrix<T> mm = m;
if (fout) {
*fout << "SymBandSVDiv:\n";
*fout << "M = "<<mm<<std::endl;
*fout << "U = "<<getU()<<std::endl;
*fout << "S = "<<getS()<<std::endl;
*fout << "Vt = "<<getVt()<<std::endl;
}
Matrix<T> usv = getU()*getS()*getVt();
RT nm = Norm(usv-mm);
nm /= Norm(getU())*Norm(getS())*Norm(getVt());
RT cond = condition();
if (fout) {
*fout << "USVt = "<<usv<<std::endl;
*fout << "Norm(M-USVt)/Norm(USVt) = "<<nm;
*fout <<" "<<cond<<" * "<<TMV_Epsilon<T>()<<std::endl;
}
return nm < cond*RT(mm.colsize())*TMV_Epsilon<T>();
}
bool LUDiv<T>::checkDecomp(const BaseMatrix<T>& m, std::ostream* fout) const
{
Matrix<T> mm = m;
if (fout) {
*fout << "LUDiv:\n";
*fout << "M = "<<
(pimpl->istrans?mm.transpose():mm.view())<<std::endl;
*fout << "L = "<<getL()<<std::endl;
*fout << "U = "<<getU()<<std::endl;
*fout << "P = "<<getP()<<std::endl;
*fout << " or by interchanges: ";
for(ptrdiff_t i=0;i<getP().size();i++)
*fout<<(getP().getValues())[i]<<" ";
}
Matrix<T> lu = getP()*getL()*getU();
RT nm = Norm(lu-(pimpl->istrans ? mm.transpose() : mm.view()));
nm /= Norm(getL())*Norm(getU());
if (fout) {
*fout << "PLU = "<<lu<<std::endl;
*fout << "Norm(M-PLU)/Norm(PLU) = "<<nm<<std::endl;
}
return nm < mm.doCondition()*RT(mm.colsize())*TMV_Epsilon<T>();
}
void gameEnv::checkIdenticals()
{
int i = 0;
int j = 1;
for(i;i<(n-1);i++)
{
for(j;j<n;j++)
{
int uToTest = 0;
int errCount = 0;
for(uToTest; uToTest < numU; uToTest++)
{
if(getU(i,uToTest) == getU(j,uToTest))
{
errCount++;
}
if(errCount == (numU-1))
{
//ERROR WE FOUND IDENTICAL NODES
std::cout << "ERROR!!!! IDENTICAL NODES!\n";
std::cout << "ERROR!!!! IDENTICAL NODES!\n";
std::cout << "ERROR!!!! IDENTICAL NODES!\n";
std::cout << "ERROR!!!! IDENTICAL NODES!\n";
std::cout << "ERROR!!!! IDENTICAL NODES!\n";
std::cout << "ERROR!!!! IDENTICAL NODES!\n";
std::cout << "ERROR!!!! IDENTICAL NODES!\n";
std::cout << "ERROR!!!! IDENTICAL NODES!\n";
std::cout << "ERROR!!!! IDENTICAL NODES!\n";
}
}
}
}
return;
}
returnValue Controller::feedbackStep( double currentTime,
const Vector& _y,
const VariablesGrid& _yRef
)
{
realClock.reset( );
if ( controlLaw == 0 )
return ACADOERROR( RET_NO_CONTROLLAW_SPECIFIED );
/* Do nothing if controller is disabled */
if ( isEnabled == BT_FALSE )
return SUCCESSFUL_RETURN;
// start real runtime measurement
realClock.start( );
Vector xEst, pEst;
/* 1) Call Estimator */
if ( obtainEstimates( currentTime,_y,xEst,pEst ) != SUCCESSFUL_RETURN )
return ACADOERROR( RET_CONTROLLER_STEP_FAILED );
/* 2) Evaluate reference trajectory */
VariablesGrid yRef( _yRef );
getCurrentReference( currentTime,yRef );
#ifdef SIM_DEBUG
yRef.print( "yRef feedback" );
#endif
controlLawClock.reset();
controlLawClock.start();
/* 3) Perform feedback step of control law */
if ( controlLaw->feedbackStep( currentTime,xEst,pEst,yRef ) != SUCCESSFUL_RETURN )
return ACADOERROR( RET_CONTROLLER_STEP_FAILED );
controlLawClock.stop();
realClock.stop( );
#ifdef SIM_DEBUG
Vector uTmp;
getU( uTmp );
uTmp.print("u(0) after feedbackStep");
#endif
return SUCCESSFUL_RETURN;
}
void Utility::Yuv444FileSaver::savePacked() {
for(std::size_t k=0; k<video_->getNumberOfFrames()&&isRunning_; k++) {
auto frame_=video_->getFrame(k);
for(int i=0; i<width_*height_; i++) {
int y1=i/width_;
int x1=i%width_;
if(!frame_->valid(x1,y1)) {
qDebug()<<"Wrong pixel coordinates";
continue;
}
auto vec=Rgb888ToYuv444(frame_->pixel(x1,y1));
dataStream_<<vec.getY()<<vec.getU()<<vec.getV();
}
}
}
int Pricer::put_american(double &price, double S0, double K, double T, double R, double vol, int N)
{
if (S0 <= 0 || K <= 0 || vol <= 0 || T <= 0) {
//std::cout << "the asset price, the exchange rate, the maturity, the strike and the volatilities must be positive " << std::endl;
return 10;
}
double *PutPayoffs;
//calcul des valeurs que l'on aura besoin plus tard pour le pricing
//pas de temps
const double h = T / (float)N;
const double rh = R * h;
//facteurs d'actualisation et de capitalisation
const double If = exp(rh);
const double Df = exp(-rh);
//pseudo-probabilites
const double u = getU(T, vol, N);
const double d = getD(T, vol, N);
const double pu = (If - d) / (u - d);
const double pd = 1.0 - pu;
const double puByDf = pu * Df;
const double pdByDf = pd * Df;
double x;
double s;
double payoff;
int i, j;
//calcul des payoffs a maturite
PutPayoffs = expiryPutValues(S0, K, N, u, d);
//descente dans l'arbre
for (i = N - 1; i >= 0; i--) {
for (j = 0; j <= i; j++) {
x = puByDf * PutPayoffs[j] + pdByDf * PutPayoffs[j + 1];
s = S0 * pow(u, i - j) * pow(d, j);
payoff = K - s;
payoff = (payoff>0) ? payoff : 0;
PutPayoffs[j] = (payoff>x) ? payoff : x;
}
}
//le prix en 0 est stocke a la premiere position du tableau
price = (double)PutPayoffs[0];
free(PutPayoffs);
return 0;
}
void peano::applications::faxen::records::RegularGridCellPacked::toString (std::ostream& out) const {
out << "(";
out << "P:" << getP();
out << ",";
out << "U:" << getU();
out << ",";
out << "V:" << getV();
out << ",";
out << "F:" << getF();
out << ",";
out << "G:" << getG();
out << ",";
out << "rhs:" << getRhs();
out << ",";
out << "res:" << getRes();
out << ",";
out << "isInside:" << getIsInside();
out << ")";
}
float BlockDerivative::update(float input)
{
float output;
if (_initialized) {
output = _lowPass.update((input - getU()) / getDt());
} else {
// if this is the first call to update
// we have no valid derivative
// and so we use the assumption the
// input value is not changing much,
// which is the best we can do here.
output = 0.0f;
_initialized = true;
}
setU(input);
return output;
}
returnValue Controller::preparationStep( double nextTime,
const VariablesGrid& _yRef
)
{
if ( controlLaw == 0 )
return ACADOERROR( RET_NO_CONTROLLAW_SPECIFIED );
/* Do nothing if controller is disabled */
if ( isEnabled == BT_FALSE )
return SUCCESSFUL_RETURN;
realClock.start();
controlLawClock.start();
/* 1) Evaluate reference trajectory */
VariablesGrid yRef( _yRef );
getCurrentReference( nextTime,yRef );
#ifdef SIM_DEBUG
yRef.print( "yRef preparation" );
#endif
/* 2) Perform preparation step of control law */
if ( controlLaw->preparationStep( nextTime,yRef ) != SUCCESSFUL_RETURN )
return ACADOERROR( RET_CONTROLLER_STEP_FAILED );
controlLawClock.stop();
logCollection.setLast( LOG_TIME_CONTROL_LAW,controlLawClock.getTime() );
// stop real runtime measurement
realClock.stop();
logCollection.setLast( LOG_TIME_CONTROLLER,realClock.getTime() );
#ifdef SIM_DEBUG
Vector uTmp;
getU( uTmp );
uTmp.print("u(0) after preparationStep");
#endif
return SUCCESSFUL_RETURN;
}
void right(float distance){
setPosition(position + distance * getU());
}
void pitch(Angle angle){
const Mat4 rotation = Mat4::rotation(getU().normalize(), angle);
transformNormals(rotation);
}
/// Returns the Debye-Waller factor, using an isotropic atomic displacement and
/// the stored unit cell.
double IsotropicAtomBraggScatterer::getDebyeWallerFactor(const V3D &hkl) const {
V3D dstar = getCell().getB() * hkl;
return exp(-2.0 * M_PI * M_PI * getU() * dstar.norm2());
}
//! Returns the number of nonzero entries in the global graph.
int getGlobalNumEntries() const {return(getL().getGlobalNumEntries()+getU().getGlobalNumEntries());}
peanoclaw::records::DataPacked peanoclaw::records::Data::convert() const{
return DataPacked(
getU()
);
}
void peanoclaw::records::Data::toString (std::ostream& out) const {
out << "(";
out << "u:" << getU();
out << ")";
}
//---------------------------------------------------------------------------
BezierSurface::BezierSurface(const unsigned rows,
const unsigned cols,
const unsigned detalization,
const bool _gridHidden,
const bool _useFilling = true)
: gridHidden(_gridHidden),
useFilling(_useFilling),
ptsPerUnit(detalization),
lastCanvasSize(TSize(0, 0)),
zBuffer(NULL)
{
frontColor = clRed;
backColor = clBlack;
points_container knots = points_container(rows, vector<Vertice *>(cols, 0));
points_container points = points_container(ptsPerUnit + 1, vector<Vertice *>(ptsPerUnit + 1, 0));
if(useFilling) {
surfaceTriangles.reserve(2 * ptsPerUnit * ptsPerUnit);
}
else {
surfaceWire.reserve(2 * ptsPerUnit * ptsPerUnit);
}
randomize();
int x0 = irand(30);
int y0 = -random(50);
int z0 = irand(30);
int lastx, lasty = y0, lastz = z0;
for (unsigned i = 0; i < rows; ++i)
{
lastx = x0;
for (unsigned j = 0; j < cols; ++j)
{
lastx += (j) ? irand(15) + 10 : irand(50);
int x = lastx;
int y = lasty + irand(30) + 1;
int z = lastz + irand(60);
UnicodeString tag = knotTag(i, j);
Vertice *point = new Vertice (x, y, z, tag);
knots[i][j] = point;
}
lasty += 10 + random(20);
}
grid = new GraphicObject();
for(poIt i = knots.begin(); i != knots.end(); ++i) {
for(pIt j = (*i).begin(); j != (*i).end(); ++j) {
grid->addVertice(*j);
}
}
for (unsigned i = 0; i < rows; ++i)
{
for (unsigned j = 0; j < cols - 1; ++j)
{
Edge *edge = new Edge(knotTag(i, j), knotTag(i, j + 1));
edge->setPen(clBlue, 2, psSolid);
grid->addEdge(edge);
}
}
for (unsigned j = 0; j < cols; ++j)
{
for (unsigned i = 0; i < rows - 1; ++i)
{
Edge *edge = new Edge(knotTag(i, j), knotTag(i + 1, j));
edge->setPen(clBlue, 2, psSolid);
grid->addEdge(edge);
}
}
// Surfacing
Matrix *U, *W, *N, *M, *B, *UN, *MW;
N = BezierSurface::getN(rows - 1);
M = BezierSurface::getN(cols - 1);
int row, col = 0;
for(float u = 0; u <= 1; u += 1.0/(float)ptsPerUnit, ++col) {
U = getU(u, rows - 1);
UN = new Matrix(*U * *N);
row = 0;
for(float w = 0; w <= 1; w += 1.0/(float)ptsPerUnit, ++row) {
W = getW(w, cols - 1);
MW = new Matrix(*M * *W);
B = new Matrix(rows, cols);
double x;
double y;
double z;
for(unsigned i = 0; i < rows; ++i) {
for(unsigned j = 0; j < cols; ++j) {
B->values[i][j] = knots[i][j]->getX();
}
}
Matrix Q1 = *UN * *B * *MW;
x = Q1.values[0][0];
for(unsigned i = 0; i < rows; ++i) {
for(unsigned j = 0; j < cols; ++j) {
B->values[i][j] = knots[i][j]->getY();
}
}
Matrix Q2 = *UN * *B * *MW;
y = Q2.values[0][0];
for(unsigned i = 0; i < rows; ++i) {
for(unsigned j = 0; j < cols; ++j) {
B->values[i][j] = knots[i][j]->getZ();
}
}
Matrix Q3 = *U * *N * *B * *M * *W;
z = Q3.values[0][0];
UnicodeString tag = pointTag(row, col);
Vertice *pt = new Vertice(x, y, z, tag);
points[row][col] = pt;
delete W;
delete B;
}
delete U;
delete UN;
}
delete N;
delete M;
for(unsigned i = 0; i < ptsPerUnit; ++i) {
for(unsigned j = 0; j < ptsPerUnit; ++j) {
Vertice *vert1 = points[i][j];
Vertice *vert2 = points[i][j + 1];
Vertice *vert3 = points[i + 1][j];
Vertice *vert4 = points[i + 1][j + 1];
if(useFilling) {
Triangle *triangle1 = new Triangle;
Vertice *a, *b, *c;
Vertice *verts[3] = {vert1, vert2, vert3};
a = verts[0];
b = verts[1];
c = verts[2];
triangle1->av = new Vertice(*a);
triangle1->bv = new Vertice(*b);
triangle1->cv = new Vertice(*c);
surfaceTriangles.push_back(triangle1);
Triangle *triangle2 = new Triangle;
Vertice *verts2[3] = {vert2, vert3, vert4};
a = verts2[1];
b = verts2[0];
c = verts2[2];
triangle2->av = new Vertice(*a);
triangle2->bv = new Vertice(*b);
triangle2->cv = new Vertice(*c);
surfaceTriangles.push_back(triangle2);
}
}
}
}
// topdir = 1..nd
// sidedir = -nd..nd
// toplinknum,sidelinknum = 0..nin-1
void
QOP_staples_deriv(QOP_info_t *info, int nout, int nin,
QDP_ColorMatrix *deriv[], QDP_ColorMatrix *chain[],
QDP_ColorMatrix *in[],
int nstaples[], int *topdir[], int *sidedir[],
int *toplinknum[], int *sidelinknum[], QLA_Real *coef[])
{
#define NC QDP_get_nc(in[0])
double dtime = QOP_time();
double nflops = 0;
int nd = QDP_ndim();
QDP_ColorMatrix *ftmps[nin][nd], *t1, *t2, *t3, *t4, *tc, *bt2[nd], *bt3[nd], *ctmps[nd];
int ctn[nd];
for(int i=0; i<nin; i++)
for(int j=0; j<nd; j++)
ftmps[i][j] = NULL;
for(int i=0; i<nd; i++) bt2[i] = bt3[i] = ctmps[i] = NULL;
t1 = QDP_create_M();
t2 = QDP_create_M();
t3 = QDP_create_M();
t4 = QDP_create_M();
tc = QDP_create_M();
// process in reverse in case calculated staples used as input for others
for(int io=nout-1; io>=0; io--) {
for(int i=0; i<nd; i++) {
if(ctmps[i]) QDP_discard_M(ctmps[i]);
ctn[i] = 0;
}
QDP_M_eq_M(tc, chain[io], QDP_all);
for(int s=0; s<nstaples[io]; s++) {
QLA_Real c = coef[io][s];
int tn = toplinknum[io][s];
int sdir = sidedir[io][s];
//QOP_printf0("io: %i s: %i sdir: %i tn: %i c: %g\n", io, s, sdir, tn, c);
if(sdir==0) {
if(c==1) {
QDP_M_peq_M(deriv[tn], tc, QDP_all);
nflops += PEQM;
} else {
QDP_M_peq_r_times_M(deriv[tn], &c, tc, QDP_all);
nflops += 2*PEQM;
}
} else if(sdir>0) {
int nu = sdir-1;
int mu = topdir[io][s]-1;
int sn = sidelinknum[io][s];
//QOP_printf0(" mu: %i nu: %i sn: %i\n", mu, nu, sn);
QDP_ColorMatrix *Umunu = getU(tn, mu, nu);
QDP_ColorMatrix *Unumu = getU(sn, nu, mu);
QDP_M_eq_M_times_M(t1, in[sn], Umunu, QDP_all);
QDP_M_eq_Ma_times_M(t2, tc, t1, QDP_all);
QDP_ColorMatrix *tb2 = shiftb(t2, mu);
QDP_M_eq_M_times_M(t1, tc, Unumu, QDP_all);
QDP_M_eq_Ma_times_M(t3, in[sn], t1, QDP_all);
QDP_ColorMatrix *tb3 = shiftb(t3, nu);
if(c==1) {
QDP_M_peq_M_times_Ma(deriv[sn], t1, Umunu, QDP_all);
QDP_M_peq_M(deriv[sn], tb2, QDP_all);
QDP_M_peq_M(deriv[tn], tb3, QDP_all);
nflops += 4*EQMTM+PEQMTM+2*PEQM;
} else {
QDP_M_eq_M_times_Ma(t4, t1, Umunu, QDP_all);
QDP_M_peq_r_times_M(deriv[sn], &c, t4, QDP_all);
QDP_M_peq_r_times_M(deriv[sn], &c, tb2, QDP_all);
QDP_M_peq_r_times_M(deriv[tn], &c, tb3, QDP_all);
nflops += 5*EQMTM+6*PEQM;
}
QDP_discard_M(tb2);
QDP_discard_M(tb3);
} else {
int nu = -sdir-1;
int mu = topdir[io][s]-1;
int sn = sidelinknum[io][s];
QDP_ColorMatrix *Cmunu = getC(nu);
QDP_ColorMatrix *Unumu = getU(sn, nu, mu);
QDP_M_eq_M_times_M(t1, in[sn], Cmunu, QDP_all);
QDP_M_eq_Ma_times_M(t2, in[tn], t1, QDP_all);
QDP_ColorMatrix *tb2 = shiftb(t2, mu);
QDP_M_eq_M_times_M(t3, in[tn], Unumu, QDP_all);
if(c==1) {
QDP_M_peq_M_times_Ma(deriv[tn], t1, Unumu, QDP_all);
QDP_M_peq_M_times_Ma(deriv[sn], t3, Cmunu, QDP_all);
QDP_M_peq_M(deriv[sn], tb2, QDP_all);
nflops += 3*EQMTM+2*PEQMTM+PEQM;
} else {
QDP_M_eq_M_times_Ma(t4, t1, Unumu, QDP_all);
QDP_M_peq_r_times_M(deriv[tn], &c, t4, QDP_all);
QDP_M_eq_M_times_Ma(t4, t3, Cmunu, QDP_all);
QDP_M_peq_r_times_M(deriv[sn], &c, t4, QDP_all);
QDP_M_peq_r_times_M(deriv[sn], &c, tb2, QDP_all);
nflops += 5*EQMTM+6*PEQM;
}
QDP_discard_M(tb2);
}
}
}
for(int i=0; i<nin; i++)
for(int j=0; j<nd; j++)
if(ftmps[i][j]!=NULL) QDP_destroy_M(ftmps[i][j]);
for(int i=0; i<nd; i++) {
if(bt2[i]!=NULL) QDP_destroy_M(bt2[i]);
if(bt3[i]!=NULL) QDP_destroy_M(bt3[i]);
if(ctmps[i]!=NULL) QDP_destroy_M(ctmps[i]);
}
QDP_destroy_M(t1);
QDP_destroy_M(t2);
QDP_destroy_M(t3);
QDP_destroy_M(t4);
QDP_destroy_M(tc);
info->final_sec = QOP_time() - dtime;
info->final_flop = nflops*QDP_sites_on_node;
info->status = QOP_SUCCESS;
#undef NC
}
float BlockDerivative::update(float input)
{
float output = _lowPass.update((input - getU()) / getDt());
setU(input);
return output;
}
// topdir = 1..nd
// sidedir = -nd..nd
// toplinknum,sidelinknum = 0..nin-1
void
QOP_staples(QOP_info_t *info, int nout, int nin,
QDP_ColorMatrix *out[], QDP_ColorMatrix *in[],
int nstaples[], int *topdir[], int *sidedir[],
int *toplinknum[], int *sidelinknum[], QLA_Real *coef[])
{
#define NC QDP_get_nc(in[0])
double dtime = QOP_time();
double nflops = 0;
int nd = QDP_ndim();
QDP_ColorMatrix *ftmps[nin][nd], *t1, *t2, *bt2[nd];
for(int i=0; i<nin; i++)
for(int j=0; j<nd; j++)
ftmps[i][j] = NULL;
for(int i=0; i<nd; i++) bt2[i] = NULL;
t1 = QDP_create_M();
t2 = QDP_create_M();
for(int io=0; io<nout; io++) {
//QOP_printf0("%i: ns: %i\n", io, nstaples[io]);
for(int s=0; s<nstaples[io]; s++) {
QLA_Real c = coef[io][s];
int tn = toplinknum[io][s];
int sdir = sidedir[io][s];
//QOP_printf0(" %i: sdir: %i c: %g\n", s, sdir, c);
if(sdir==0) {
if(c==1) {
QDP_M_peq_M(out[io], in[tn], QDP_all);
nflops += PEQM;
} else {
QDP_M_peq_r_times_M(out[io], &c, in[tn], QDP_all);
nflops += 2*PEQM;
}
} else if(sdir>0) {
int nu = sdir-1;
int mu = topdir[io][s]-1;
int sn = sidelinknum[io][s];
QDP_ColorMatrix *Umunu = getU(tn, mu, nu);
QDP_ColorMatrix *Unumu = getU(sn, nu, mu);
QDP_M_eq_M_times_M(t1, in[sn], Umunu, QDP_all);
if(c==1) {
QDP_M_peq_M_times_Ma(out[io], t1, Unumu, QDP_all);
nflops += EQMTM+PEQMTM;
} else {
QDP_M_eq_M_times_Ma(t2, t1, Unumu, QDP_all);
QDP_M_peq_r_times_M(out[io], &c, t2, QDP_all);
nflops += 2*EQMTM+2*PEQM;
}
} else {
int nu = -sdir-1;
int mu = topdir[io][s]-1;
int sn = sidelinknum[io][s];
QDP_ColorMatrix *Unumu = getU(sn, nu, mu);
QDP_M_eq_M_times_M(t1, in[tn], Unumu, QDP_all);
QDP_M_eq_Ma_times_M(t2, in[sn], t1, QDP_all);
QDP_ColorMatrix *tb = shiftb(t2, nu);
if(c==1) {
QDP_M_peq_M(out[io], tb, QDP_all);
nflops += 2*EQMTM+PEQM;
} else {
QDP_M_peq_r_times_M(out[io], &c, tb, QDP_all);
nflops += 2*EQMTM+2*PEQM;
}
QDP_discard_M(tb);
}
}
}
for(int i=0; i<nin; i++)
for(int j=0; j<nd; j++)
if(ftmps[i][j]!=NULL) QDP_destroy_M(ftmps[i][j]);
for(int i=0; i<nd; i++) if(bt2[i]!=NULL) QDP_destroy_M(bt2[i]);
QDP_destroy_M(t1);
QDP_destroy_M(t2);
info->final_sec = QOP_time() - dtime;
info->final_flop = nflops*QDP_sites_on_node;
info->status = QOP_SUCCESS;
#undef NC
}
/**
* @return the number of users.
*/
size_t getNumberOfUsers(void){return getU().size1();}
