int main()
{
vector<vector<int> > mtx; // a headerben mégsem szerette annyira, de most ez nem érdekel minket annyira, ha mégis, után lehet nézni
mtx.resize(3);
mtx[0].push_back(3);
mtx[0].push_back(4);
mtx[0].push_back(4);
mtx[0].push_back(8);
mtx[1].push_back(9);
mtx[1].push_back(12);
mtx[1].push_back(10);
mtx[1].push_back(3);
mtx[2].push_back(20);
mtx[2].push_back(21);
mtx[2].push_back(22);
mtx[2].push_back(23);
if (sor(mtx)) {
cout << "Igaz. A matrixnak MINDEN sora tartalmaz legalabb egy primet!" << endl;
}
else
{
cout << "Hamis. A matrixnak NEM minden sora tartalmaz legalabb egy primet!" << endl;
}
system("pause");
return 0;
}
void
calc_phi
(
real * const phi,
const real * const u,
const real * const v,
const real dt,
const Mesh & mesh
){
make_initial_value_0( phi, mesh );
const real * const vel[2] = { u, v };
LHSofPoissonEQ lhs( mesh );
RHSofPoissonEQ rhs( mesh, vel, dt );
Neumann_plus neu_p_x( mesh, 0 );
Neumann_minus neu_m_x( mesh, 0 );
Neumann_plus neu_p_y( mesh, 1 );
Neumann_minus neu_m_y( mesh, 1 );
BCondition * bc_set[6];
bc_set[ 0] = &neu_p_x; bc_set[ 1] = &neu_m_x;
bc_set[ 2] = &neu_p_y; bc_set[ 3] = &neu_m_y;
bc_set[ 4] = NULL; bc_set[ 5] = NULL;
SOR<2> sor( phi, bc_set, lhs, rhs, mesh, 10000, 1.2 );
sor.solve(1e-9);
}
t_l *sort(t_l *list, char *flag, char *path)
{
t_l *base;
base = list;
sor(list);
if (isflag(flag, 't'))
list = sort_t(list, flag, base, path);
else if (isflag(flag, 'r'))
list = sort_r(list, base);
return (list);
}
void sor_test() {
int** m = matrix_instance(50, 50, 1);
int** ret = sor(m, 50, 50, 5);
int i,j;
for(i=0; i<3; i++) {
for (j=0; j<3; j++) {
assert(ret[i][j] == 1);
}
}
}
void CrankNicolsonSorUpdater::update( const std::vector<double> & uOld,
std::vector<double>* uNew,
const std::vector<double> & prem ) const
{
int N = uOld.size()-1;
double c = getAlpha();
// uOld restricted in the interior region
Eigen::VectorXd u = Eigen::VectorXd::Zero(N-1);
for (int i=0; i<N-1; i++) { u(i) = uOld[i+1]; }
// Construct b-vector
Eigen::VectorXd b = d_B*u;
b(0) += 0.5*c*((*uNew)[0]+uOld[0]);
b(N-2) += 0.5*c*((*uNew)[N]+uOld[N]);
// Use the state vector at the previous time step as initial estimate
Eigen::VectorXd x0 = Eigen::VectorXd::Zero(N-1);
for (int i=0; i<N-1; i++) { x0(i) = uOld[i+1]; }
// Construct early exercise premium vector
Eigen::VectorXd p = Eigen::VectorXd::Zero(0);
if ( prem.size()>0 ) {
p = Eigen::VectorXd::Zero(N-1);
for (int i=0; i<N-1; i++) {
p(i) = prem[i+1];
// If early exercise premium is given, the initial
// estimate x0 is set to early exercise premium.
// This is a reasonable choice because:
// (1) At the 1st time step, i.e., m=0,
// early_exercise_premium = K*exp(ax)*max(1-exp(x),0),
// which is exactly the same as the terminal condition;
// (2) At the successive time steps, i.e., m>0,
// early_exercise_premium serves as a good approximation
// of the value of state vectors.
x0(i) = prem[i+1];
}
}
// update
std::tuple<Eigen::VectorXd, int> res = sor(d_omega, d_A, b, d_tol, x0, p);
Eigen::VectorXd x = std::get<0>(res);
for (int i=0; i<N-1; i++) { (*uNew)[i+1] = x(i); }
}
int main() {
int n, m, N;
real a, b, c, d;
real omega, tol;
int iterMax;
real valor;
scanf("%d", &n);
scanf("%d", &m);
scanf("%lf", &a);
scanf("%lf", &b);
scanf("%lf", &c);
scanf("%lf", &d);
scanf("%lf", &omega);
scanf("%d", &iterMax);
scanf("%lf", &tol);
scanf("%lf", &valor);
N = n*m;
Coef *coef = criaCoef(a, b, c, d, n, m);
valorPrescrito(coef, valor, DOWN);
valorPrescrito(coef, valor, LEFT);
valorPrescrito(coef, valor, RIGHT);
valorPrescrito(coef, valor, UP);
// printCoef(coef);
real *x = sor(coef, omega, iterMax, tol);
// real *val = validacao(a, b, c, d, n, m);
// real *x = sorLivre(a, b, c, d, n, m, omega, iterMax, tol, valor);
printf("Solução do sistema linear: \n");
int I;
for(I = 0; I < N; I++)
printf("%lf\n", x[I]);
liberaCoef(coef);
free(x);
// free(val);
return 0;
}
int main(void)
{
int i, j;
for(i=0; i<5; i++)
{
for(j=0; j<3; j++)
{
printf("Enter the number:\n");
scanf(" %d", &massiv[i][j]);
disp();
}
}
sor();
disp();
return 0;
}
int main(int argc, char** argv){
cudaEvent_t start, stop;
float time;
cudaEventCreate(&start);
cudaEventCreate(&stop);
cudaEventRecord(start, 0);
sor(domain, SIZE, ITER);
cudaEventRecord(stop, 0);
cudaEventSynchronize(stop);
cudaEventElapsedTime(&time, start, stop);
printf("F=%0.1f Time=%0.2f; %.0f MFlop/s\n",domain[10*SIZE+10],time/1000.0, MFlops(time) );
return 0;
}
int main(int argc, char** args){
double Re, UI, VI, PI, GX, GY, t_end, xlength, ylength, dt, dx, dy, alpha, omg, tau, eps, dt_value, t, res, dp, nu;
double **U, **V, **P, **F, **G, **RS;
double **K, **E; /* turbulent kinetic energy k, dissipation rate epsilon*/
double Fu, Fv; /* force integration variables */
double KI, EI, cn, ce, c1, c2; /* K and E: Initial values for k and epsilon */
int n, it, imax, jmax, itermax, pb, boundaries[4];
int fluid_cells; /* Number of fluid cells in our geometry */
int **Flag; /* Flagflield matrix */
char vtkname[200];
char pgm[200];
char problem[10]; /* Problem name */
char fname[200];
if(argc<2){
printf("No parameter file specified. Terminating...\n");
exit(1);
}
sprintf(fname, "%s%s", CONFIGS_FOLDER, args[1]);
printf("%s\n",fname);
read_parameters(fname, &Re, &UI, &VI, &PI, &GX, &GY, &t_end, &xlength, &ylength, &dt, &dx, &dy, &imax, &jmax, &alpha, &omg, &itermax, &eps, boundaries, &dp, &pb, &KI, &EI, &cn, &ce, &c1, &c2, pgm, &nu, problem);
/* Allocate Flag matrix */
Flag = imatrix( 0, imax+1, 0, jmax+1 );
U = matrix ( 0 , imax+1 , 0 , jmax+1 );
V = matrix ( 0 , imax+1 , 0 , jmax+1 );
P = matrix ( 0 , imax+1 , 0 , jmax+1 );
F = matrix ( 0 , imax , 0 , jmax );
G = matrix ( 0 , imax , 0 , jmax );
RS = matrix ( 0 , imax , 0 , jmax );
K = matrix ( 0 , imax+1 , 0 , jmax+1 );
E = matrix ( 0 , imax+1 , 0 , jmax+1 );
/* Initialize values to the Flag, u, v and p */
init_flag(CONFIGS_FOLDER,pgm, imax, jmax, &fluid_cells, Flag );
init_uvp(UI, VI, PI, KI, EI, imax, jmax, U, V, P, K, E, Flag, problem );
printf("Problem: %s\n", problem );
printf( "xlength = %f, ylength = %f\n", xlength, ylength );
printf( "imax = %d, jmax = %d\n", imax, jmax );
printf( "dt = %f, dx = %f, dy = %f\n", dt, dx, dy);
printf( "Number of fluid cells = %d\n", fluid_cells );
printf( "Reynolds number: %f\n\n", Re);
t=.0;
n=0;
while( t <= t_end ){
boundaryvalues( imax, jmax, U, V, K, E, boundaries, Flag );
/* special inflow boundaries, including k and eps */
spec_boundary_val( problem, imax, jmax, U, V, K, E, Re, dp, cn, ylength);
/* calculate new values for k and eps */
comp_KAEP(Re, nu, cn, ce, c1, c2, alpha, dt, dx, dy, imax, jmax, U, V, K, E, GX, GY, Flag);
/* calculate new values for F and G */
calculate_fg( Re, GX, GY, alpha, dt, dx, dy, imax, jmax, U, V, F, G, K, E, nu, cn, Flag );
/* calculate right hand side */
calculate_rs( dt, dx, dy, imax, jmax, F, G, RS, Flag );
it = 0;
res = 10000.0;
while( it < itermax && fabs(res) > eps ){
sor( omg, dx, dy, imax, jmax, fluid_cells, P, RS, Flag, &res, problem, dp );
it++;
}
printf("[%5d: %f] dt: %f, sor iterations: %4d \n", n, t, dt, it);
/* calculate new values for u and v */
calculate_uv( dt, dx, dy, imax, jmax, U, V, F, G, P, Flag );
t += dt;
n++;
}
sprintf(vtkname, "%s%s", VISUA_FOLDER, args[1]);
write_vtkFile( vtkname, 1, xlength, ylength, imax, jmax, dx, dy, U, V, P, K, E, Flag);
comp_surface_force( Re, dx, dy, imax, jmax, U, V, P, Flag, &Fu, &Fv);
printf( "\nProblem: %s\n", problem );
printf( "xlength = %f, ylength = %f\n", xlength, ylength );
printf( "imax = %d, jmax = %d\n", imax, jmax );
printf( "dt = %f, dx = %f, dy = %f\n", dt, dx, dy);
printf( "Number of fluid cells = %d\n", fluid_cells );
printf( "Reynolds number: %f\n", Re);
printf( "Drag force = %f Lift force = %f\n", Fu, Fv);
/* free memory */
free_matrix(U,0,imax+1,0,jmax+1);
free_matrix(V,0,imax+1,0,jmax+1);
free_matrix(P,0,imax+1,0,jmax+1);
free_matrix(K,0,imax+1,0,jmax+1);
free_matrix(E,0,imax+1,0,jmax+1);
free_matrix(F,0,imax,0,jmax);
free_matrix(G,0,imax,0,jmax);
free_matrix(RS,0,imax,0,jmax);
free_imatrix(Flag,0,imax+1,0,jmax+1);
return 0;
}
SpanOrQueryPtr ExplanationsFixture::sor(const String& s, const String& m, const String& e)
{
return sor(st(s), st(m), st(e));
}
int main(int argn, char** args){
if (argn !=2 ) {
printf("When running the simulation, please give a valid scenario file name!\n");
return 1;
}
//set the scenario
char *filename = NULL;
filename = args[1];
//initialize variables
double t = 0; /*time start*/
int it, n = 0; /*iteration and time step counter*/
double res; /*residual for SOR*/
/*arrays*/
double **U, **V, **P;
double **RS, **F, **G;
int **Flag; //additional data structure for arbitrary geometry
/*those to be read in from the input file*/
double Re, UI, VI, PI, GX, GY, t_end, xlength, ylength, dt, dx, dy, alpha, omg, tau, eps, dt_value;
int imax, jmax, itermax;
double presLeft, presRight, presDelta; //for pressure stuff
int wl, wr, wt, wb;
char problem[32];
double vel; //in case of a given inflow or wall velocity
//read the parameters, using problem.dat, including wl, wr, wt, wb
read_parameters(filename, &Re, &UI, &VI, &PI, &GX, &GY, &t_end, &xlength, &ylength, &dt, &dx, &dy, &imax,
&jmax, &alpha, &omg, &tau, &itermax, &eps, &dt_value, &wl, &wr, &wt, &wb, problem, &presLeft, &presRight, &presDelta, &vel);
int pics = dt_value/dt; //just a helping variable for outputing vtk
//allocate memory, including Flag
U = matrix(0, imax+1, 0, jmax+1);
V = matrix(0, imax+1, 0, jmax+1);
P = matrix(0, imax+1, 0, jmax+1);
RS = matrix(1, imax, 1, jmax);
F = matrix(0, imax, 1, jmax);
G = matrix(1, imax, 0, jmax);
Flag = imatrix(0, imax+1, 0, jmax+1); // or Flag = imatrix(1, imax, 1, jmax);
int kmax = 20; //test no slip boundary value function
double ***U3d = (double ***) malloc((size_t)((imax+1)*(jmax+1)*(kmax+1) * sizeof(double*)) ); //test no slip boundary value function
double ***V3d = (double ***) malloc((size_t)((imax+1)*(jmax+1)*(kmax+1) * sizeof(double*)) ); //test no slip boundary value function
double ***W3d = (double ***) malloc((size_t)((imax+1)*(jmax+1)*(kmax+1) * sizeof(double*)) ); //test no slip boundary value function
int ***Flag3d = (int ***) malloc((size_t)((imax+1)*(jmax+1)*(kmax+1) * sizeof(int*)) ); //test no slip boundary value function
//initialisation, including **Flag
init_flag(problem, imax, jmax, presDelta, Flag);
init_uvp(UI, VI, PI, imax, jmax, U, V, P, problem);
//going through all time steps
while(t < t_end){
//adaptive time stepping
calculate_dt(Re, tau, &dt, dx, dy, imax, jmax, U, V);
//setting bound.values
boundaryvalues(imax, jmax, U, V, P, wl, wr, wt, wb, F, G, problem, Flag, vel); //including P, wl, wr, wt, wb, F, G, problem
//test no slip boundary value function
for(int i=1; i<=imax; i++){
for(int j=1; j<=jmax; j++){
for(int k=1; k<=kmax; k++){
boundaryvalues_no_slip(i, j, k, U3d, V3d, W3d, Flag3d); //test no slip boundary value function
}
}
}
//computing F, G and right hand side of pressue eq.
calculate_fg(Re, GX, GY, alpha, dt, dx, dy, imax, jmax, U, V, F, G, Flag);
calculate_rs(dt, dx, dy, imax, jmax, F, G, RS);
//iteration counter
it = 0;
do{
//
//perform SOR iteration, at same time set bound.values for P and new residual value
sor(omg, dx, dy, imax, jmax, P, RS, &res, Flag, presLeft, presRight);
it++;
}while(it<itermax && res>eps);
/* if (it == itermax) {
printf("Warning: sor while loop exits because it reaches the itermax. res = %f, time =%f\n", res, t);
}
*/ //calculate U and V of this time step
calculate_uv(dt, dx, dy, imax, jmax, U, V, F, G, P, Flag);
//indent time and number of time steps
n++;
t += dt;
//output of pics for animation
if (n%pics==0 ){
write_vtkFile(filename, n, xlength, ylength, imax, jmax, dx, dy, U, V, P);
}
}
//output of U, V, P at the end for visualization
//write_vtkFile("DrivenCavity", n, xlength, ylength, imax, jmax, dx, dy, U, V, P);
//free memory
free_matrix(U, 0, imax+1, 0, jmax+1);
free_matrix(V, 0, imax+1, 0, jmax+1);
free_matrix(P, 0, imax+1, 0, jmax+1);
free_matrix(RS, 1, imax, 1, jmax);
free_matrix(F, 0, imax, 1, jmax);
free_matrix(G, 1, imax, 0, jmax);
free_imatrix(Flag, 0, imax+1, 0, jmax+1);
free(U3d);
free(V3d);
free(W3d);
free(Flag3d);
return -1;
}
int main(int argn, char** args){
double **U, **V, **P, **F, **G, **RS;
const char *szFileName = "dam_break.dat";
double Re, UI, VI, PI, GX, GY, t_end, xlength, ylength, dt, dx, dy, alpha, omg, tau, eps, dt_value;
double res = 0, t = 0, n = 0;
int imax, jmax, itermax, it;
/* Additional data structures for VOF */
double **fluidFraction;
double **fluidFraction_alt;
int **flagField;
double **dFdx, **dFdy;
int **pic;
char output_dirname[60];
double epsilon = 1e-10;
/* Read the program configuration file using read_parameters() */
read_parameters(szFileName, &Re, &UI, &VI, &PI, &GX, &GY, &t_end, &xlength, &ylength, &dt, &dx, &dy, &imax, &jmax, &alpha, &omg, &tau, &itermax, &eps, &dt_value);
/* Set up the matrices (arrays) needed using the matrix() command */
U = matrix(0, imax , 0, jmax+1);
V = matrix(0, imax+1, 0, jmax );
P = matrix(0, imax+1, 0, jmax+1);
F = matrix(0, imax , 0, jmax+1);
G = matrix(0, imax+1, 0, jmax );
RS= matrix(0, imax+1, 0, jmax+1);
flagField = imatrix(0, imax+1, 0, jmax+1);
fluidFraction = matrix(0, imax+1, 0, jmax+1);
fluidFraction_alt = matrix(0, imax+1, 0, jmax+1);
dFdx = matrix(0, imax+1, 0, jmax+1);
dFdy = matrix(0, imax+1, 0, jmax+1);
/*create a directory*/
strcpy(output_dirname, "dam_break/dam_break");
mkdir("dam_break", 0777);
/* Read pgm file with domain and initial setting */
pic = read_pgm("dam_break.pgm");
init_fluidFraction(pic, fluidFraction,fluidFraction_alt, imax, jmax);
/* Assign initial values to u, v, p */
init_uvp(UI, VI, PI, imax, jmax, U, V, P);
/* Set valid values for the fluid fraction */
adjust_fluidFraction(fluidFraction, flagField, epsilon, imax, jmax);
while(t <= t_end){
/*Select δt*/
calculate_dt(Re, tau, &dt, dx, dy, imax, jmax, U, V);
/* Set boundary values for u and v */
/* boundaryvalues(imax, jmax, U, V, flagField, dx, dy);
*/
/* Compute F(n) and G(n) */
calculate_fg(Re, GX, GY, alpha, dt, dx, dy, imax, jmax, U, V, F, G, flagField);
/* Compute the right-hand side rs of the pressure equation */
calculate_rs(dt, dx, dy, imax, jmax, F, G, RS, flagField);
/* Determine the orientation of the free surfaces */
calculate_freeSurfaceOrientation(fluidFraction, flagField, dFdx, dFdy, dx, dy, imax, jmax);
/* Perform SOR iterations */
it=0;
res = 1e6;
while(it < itermax && res > eps){
sor(omg, dx, dy, imax, jmax, P, RS, fluidFraction, flagField, dFdx, dFdy, &res);
it++;
}
/* Compute u(n+1) and v(n+1) */
calculate_uv(dt, dx, dy, imax, jmax, U, V, F, G, P, flagField);
boundaryvalues(imax, jmax, U, V, flagField, dx, dy);
/* Compute fluidFraction(n+1) */
calculate_fluidFraction(fluidFraction,fluidFraction_alt, flagField, U, V, dFdx, dFdy, imax, jmax, dx, dy, dt);
boundaryvalues(imax, jmax, U, V, flagField, dx, dy);
/* Set valid values for the fluid fraction */
adjust_fluidFraction(fluidFraction, flagField, epsilon, imax, jmax);
boundaryvalues(imax, jmax, U, V, flagField, dx, dy);
if((int)n % 10 == 0) {
write_vtkFile(output_dirname, n, xlength, ylength, imax, jmax, dx, dy, U, V, P, fluidFraction, flagField);
}
/* Print out simulation time and whether SOR converged */
printf("Time: %.4f", t);
if(res > eps) {printf("\t*** Did not converge (res=%f, eps=%f)", res, eps);return -1;}
printf("\n");
t = t + dt;
n++;
}
free_matrix(U , 0, imax , 0, jmax+1);
free_matrix(V , 0, imax+1, 0, jmax );
free_matrix(P , 0, imax+1, 0, jmax+1);
free_matrix(F , 0, imax , 0, jmax+1);
free_matrix(G , 0, imax+1, 0, jmax );
free_matrix(RS, 0, imax+1, 0, jmax+1);
return -1;
}
PointViewSet StatisticalOutlierFilter::run(PointViewPtr input)
{
bool logOutput = log()->getLevel() > LogLevel::Debug1;
if (logOutput)
log()->floatPrecision(8);
log()->get(LogLevel::Debug2) << "Process StatisticalOutlierFilter...\n";
// convert PointView to PointXYZ
typedef pcl::PointCloud<pcl::PointXYZ> Cloud;
Cloud::Ptr cloud(new Cloud);
BOX3D bounds;
input->calculateBounds(bounds);
pclsupport::PDALtoPCD(input, *cloud, bounds);
// PCL should provide console output at similar verbosity level as PDAL
int level = log()->getLevel();
switch (level)
{
case 0:
pcl::console::setVerbosityLevel(pcl::console::L_ALWAYS);
break;
case 1:
pcl::console::setVerbosityLevel(pcl::console::L_ERROR);
break;
case 2:
pcl::console::setVerbosityLevel(pcl::console::L_WARN);
break;
case 3:
pcl::console::setVerbosityLevel(pcl::console::L_INFO);
break;
case 4:
pcl::console::setVerbosityLevel(pcl::console::L_DEBUG);
break;
default:
pcl::console::setVerbosityLevel(pcl::console::L_VERBOSE);
break;
}
// setup the outlier filter
pcl::StatisticalOutlierRemoval<pcl::PointXYZ> sor(true);
sor.setInputCloud(cloud);
sor.setMeanK(m_meanK);
sor.setStddevMulThresh(m_multiplier);
pcl::PointCloud<pcl::PointXYZ> output;
sor.setNegative(true);
sor.filter(output);
// filtered to return inliers
pcl::PointIndicesPtr inliers(new pcl::PointIndices);
sor.getRemovedIndices(*inliers);
log()->get(LogLevel::Debug2) << inliers->indices.size() << std::endl;
PointViewSet viewSet;
if (inliers->indices.empty())
{
log()->get(LogLevel::Warning) << "Requested filter would remove all points. Try increasing the multiplier.\n";
viewSet.insert(input);
return viewSet;
}
// inverse are the outliers
std::vector<int> outliers(input->size()-inliers->indices.size());
for (PointId i = 0, j = 0, k = 0; i < input->size(); ++i)
{
if (i == (PointId)inliers->indices[j])
{
j++;
continue;
}
outliers[k++] = i;
}
if (!outliers.empty() && (m_classify || m_extract))
{
if (m_classify)
{
log()->get(LogLevel::Debug2) << "Labeled " << outliers.size() << " outliers as noise!\n";
// set the classification label of outlier returns as 18
// (corresponding to ASPRS LAS specification for high noise)
for (const auto& i : outliers)
{
input->setField(Dimension::Id::Classification, i, 18);
}
viewSet.insert(input);
}
if (m_extract)
{
log()->get(LogLevel::Debug2) << "Extracted " << inliers->indices.size() << " inliers!\n";
// create new PointView containing only outliers
PointViewPtr output = input->makeNew();
for (const auto& i : inliers->indices)
{
output->appendPoint(*input, i);
}
viewSet.erase(input);
viewSet.insert(output);
}
}
else
{
if (outliers.empty())
log()->get(LogLevel::Warning) << "Filtered cloud has no outliers!\n";
if (!(m_classify || m_extract))
log()->get(LogLevel::Warning) << "Must choose --classify or --extract\n";
// return the input buffer unchanged
viewSet.insert(input);
}
return viewSet;
}
//=============================================================================
int main(int argc, char** argv)
{
///////////////////////////////////////////////
//////////////// set precision ////////////////
///////////////////////////////////////////////
std::cout.precision(9);
std::cout.setf(std::ios::scientific, std::ios::floatfield);
std::cout.setf(std::ios::showpos);
std::cerr.precision(3);
std::cerr.setf(std::ios::scientific, std::ios::floatfield);
std::cerr.setf(std::ios::showpos);
///////////////////////////////////////////////
//////////////////// A,b //////////////////////
///////////////////////////////////////////////
CPPL::dgsmatrix A;
CPPL::dcovector b;
bool file =false;
///////////////////////////
////////// read ///////////
///////////////////////////
if(file){
//A.read("A10.dge"); b.read("b10.dco");
}
///////////////////////////
///////// random //////////
///////////////////////////
else{//file==false
std::cerr << "# making random matrix" << std::endl;
const int size(1000);
A.resize(size,size);
b.resize(size);
srand(time(NULL));
for(int i=0; i<size; i++){
for(int j=0; j<size; j++){
if( rand()%5==0 ){
A(i,j) =(double(rand())/double(RAND_MAX))*2.0 -1.0;
}
}
A(i,i) +=1e-2*double(size);//generalize
b(i) =(double(rand())/double(RAND_MAX))*1. -0.5;
}
A.write("A.dge");
b.write("b.dco");
}
///////////////////////////////////////////////
////////////////// direct /////////////////////
///////////////////////////////////////////////
std::cerr << "# making solution with dgesv" << std::endl;
CPPL::dgematrix A2(A.to_dgematrix());
CPPL::dcovector b2(b);
A2.dgesv(b2);
b2.write("ans.dco");
///////////////////////////////////////////////
///////////////// iterative ///////////////////
///////////////////////////////////////////////
//////// initial x ////////
CPPL::dcovector x(b.l);
//x.read("x_ini8.dco");
x.zero();
//////// eps ////////
double eps(fabs(damax(b))*1e-6);
std::cerr << "eps=" << eps << std::endl;
//////// solve ////////
if( sor(A, b, x, eps) ){
std::cerr << "failed." << std::endl;
x.write("x.dco");
exit(1);
}
x.write("x.dco");
std::cerr << "fabs(damax(err))=" << fabs(damax(b2-x)) << std::endl;
return 0;
}
void STWarp<T>::warpingIteration( const WarpingField<T> &warpField,
const Video<T> &Bx,
const Video<T> &By,
const Video<T> &Bt,
const Video<T> &C,
WarpingField<T> &dWarpField) {
int height = dimensions[0];
int width = dimensions[1];
int nFrames = dimensions[2];
// int nChannels = dimensions[3];
if(params.verbosity > 1) {
fprintf(stderr, " - preparing linear system...");
}
// Differentiate (u,v,w)+d(u,v,w) in x,y,t
WarpingField<T> warpDX(warpField.size());
WarpingField<T> warpDY(warpField.size());
WarpingField<T> warpDT(warpField.size());
WarpingField<T> warpTotal(warpField);
warpTotal.add(dWarpField);
VideoProcessing::dx(warpTotal,warpDX);
VideoProcessing::dy(warpTotal,warpDY);
VideoProcessing::dt(warpTotal,warpDT);
// Compute smoothness cost
Video<T> smoothCost(height, width, nFrames, 9);
Video<T> lapl(warpField.size());
computeSmoothCost(warpDX, warpDY, warpDT, warpField, smoothCost, lapl);
computeOcclusion(warpTotal, C, occlusion);
// Compute data cost
Video<T> dataCost;
if(params.useFeatures){
dataCost = Video<T>(height,width,nFrames,2);
}else{
dataCost = Video<T>(height,width,nFrames,1);
}
computeDataCost(Bx, By, Bt, C, dWarpField, occlusion, dataCost);
// Prepare system
Video<T> CBx(height,width,nFrames,1);
Video<T> CBy(height,width,nFrames,1);
Video<T> CBt(height,width,nFrames,1);
Video<T> Bx2(height,width,nFrames,1);
Video<T> By2(height,width,nFrames,1);
Video<T> Bt2(height,width,nFrames,1);
Video<T> Bxy(height,width,nFrames,1);
Video<T> Bxt(height,width,nFrames,1);
Video<T> Byt(height,width,nFrames,1);
prepareLinearSystem(Bx, By, Bt, C, lapl, dataCost,
CBx, CBy, CBt, Bx2, By2, Bt2,
Bxy, Bxt, Byt);
if(params.verbosity > 1) {
fprintf(stderr, "done.\n");
}
if(params.verbosity > 1) {
fprintf(stderr, " - SOR...");
}
sor( dataCost, smoothCost, lapl,
CBx, CBy, CBt, Bx2,
By2, Bt2, Bxy, Bxt, Byt, dWarpField);
if(params.verbosity > 1) {
fprintf(stderr, "done.\n");
}
}
int main(int argc, char* argv[])
{
// Relaxation parameters
double omega_start = 0.95;
double omega_delta = 0.25;
int omega_count = 2;
// Tolerance factor
double tol=1e-6;
// Output precision
int p=9;
// Matrix dimension
int N=8;
// Band width
int band=3;
// Matrix specification
Eigen::ArrayXXd A = Eigen::ArrayXXd::Zero(N,2*band+1);
//diagonal
for (int i=0; i<N; i++) {A(i,band) = 8.0;}
//lower diagonal
for (int i=1; i<N; i++) {
//if (i>=2) A(i,band-2) = 3;
//if (i>=3) A(i,band-3) = 1;
A(i,band-2) = 2.0;
A(i,band-3) = 1.0;
}
//upper diagonal
for (int i=0; i<N-1; i++)
{
//if (i<=5) A(i,band+2) = -2;
//if (i<=4) A(i,band+3) = -1;
A(i,band+2) = -1.0;
A(i,band+3) = -2.0;
}
Eigen::MatrixXd b = Eigen::VectorXd::Zero(N);
for (int i=0; i<N; i++) {
b(i) = (2.0*i-3.0)/(2.0*i*i+1.0);
}
// Jacobi iterative solve
std::tuple<Eigen::VectorXd, int> res = jacobi(A, band, b, tol);
Eigen::VectorXd x = std::get<0>(res);
int ic = std::get<1>(res);
Eigen::VectorXd r = b - band_mult(A,band,band,x);
std::cout << std::fixed
<< std::setprecision(p)
<< "Jacobi: ,"
<< ", Iterations =, " << ic
<< ", residual =, " << r.norm()
<< std::endl;
std::cout << "x = " << std::endl;
for (int i=0; i<N; i++) {
std::cout << std::fixed << std::setprecision(p) << x(i) << std::endl;
}
// Gauss-Seidel iterative solve
res = gs(A, band, b, tol);
x = std::get<0>(res);
ic = std::get<1>(res);
r = b - band_mult(A,band,band,x);
std::cout << std::fixed
<< std::setprecision(p)
<< "Gauss-Seidel: ,"
<< ", Iterations =, " << ic
<< ", residual =, " << r.norm()
<< std::endl;
std::cout << "x = " << std::endl;
for (int i=0; i<N; i++) {
std::cout << std::fixed << std::setprecision(p) << x(i) << std::endl;
}
// SOR iterative solve for all omega values
double omega = omega_start;
for (int k=0; k<omega_count; k++) {
std::tuple<Eigen::VectorXd, int> res = sor(omega, A, band, b, tol);
Eigen::VectorXd x = std::get<0>(res);
int ic = std::get<1>(res);
Eigen::VectorXd r = b - band_mult(A,band,band,x);
std::cout << std::fixed
<< std::setprecision(p)
<< "Omega = ," << omega
<< ", Iterations =, " << ic
<< ", residual =, " << r.norm()
<< std::endl;
std::cout << "x = " << std::endl;
for (int i=0; i<N; i++) {
std::cout << std::fixed << std::setprecision(p) << x(i) << std::endl;
}
omega += omega_delta;
}
return 0;
}
int main(int argn, char** args){
double **U, **V, **P, **F, **G, **RS;
int **Flag;
char problem[60];
char parameters_filename[60];
char pgm[60];
char output_dirname[60];
double Re, UI, VI, PI, GX, GY, t_end, xlength, ylength, dt, dx, dy, alpha, omg, tau, eps, dt_value, dp;
double res = 0, t = 0, n = 0;
int imax, jmax, itermax, it;
int wl, wr, wt, wb;
int timestepsPerPlotting;
char old_output_filename[128];
struct dirent *old_outputfile;
DIR *output_dir;
/* Variables for parallel program */
int iproc, jproc, myrank, il, ir, jb, jt, rank_l, rank_r, rank_b, rank_t, omg_i, omg_j, num_proc;
double min_dt;
double *bufSend, *bufRecv;
double totalTime = 0;
struct timespec previousTime, currentTime;
MPI_Init(&argn, &args);
MPI_Comm_size(MPI_COMM_WORLD, &num_proc);
/* Read name of the problem from the command line arguments */
if(argn > 1) {
strcpy(problem, args[1]);
} else {
printf("\n=== ERROR: Please provide the name of the problem\n=== e.g. Run ./sim problem_name if there is a problem_name.dat file.\n\n");
MPI_Finalize();
return 1;
}
/* Generate input filename based on problem name */
strcpy(parameters_filename, problem);
strcat(parameters_filename, ".dat");
/* Read the program configuration file using read_parameters() */
read_parameters(parameters_filename, pgm, &Re, &UI, &VI, &PI, &GX, &GY, &t_end, &xlength, &ylength, &dt, &dx, &dy, &imax, &jmax, &alpha, &omg, &tau, &itermax, &eps, &dt_value, problem, &dp, &wl, &wr, &wt, &wb, ×tepsPerPlotting, &iproc, &jproc);
printf("%s\n", pgm);
/* Check if the number of processes is correct */
if(iproc * jproc != num_proc) {
printf("\n=== ERROR: Number of processes is incorrect (iproc=%d, jproc=%d, -np=%d) ===\n\n", iproc, jproc, num_proc);
MPI_Finalize();
return 1;
}
/* Create folder with the name of the problem */
strcpy(output_dirname, problem);
strcat(output_dirname, "/");
strcat(output_dirname, problem);
mkdir(problem, 0777);
output_dir = opendir(problem);
/* Delete existing files in output folder*/
while((old_outputfile = readdir(output_dir))) {
sprintf(old_output_filename, "%s/%s", problem, old_outputfile->d_name);
remove(old_output_filename);
}
/* Determine subdomain and neighbours for each process */
init_parallel(iproc, jproc, imax, jmax, &myrank, &il, &ir, &jb, &jt, &rank_l, &rank_r, &rank_b, &rank_t, &omg_i, &omg_j, num_proc);
/* Set up the matrices (arrays) needed using the matrix() command */
U = matrix(il-2, ir+1, jb-1, jt+1);
V = matrix(il-1, ir+1, jb-2, jt+1);
P = matrix(il-1, ir+1, jb-1, jt+1);
F = matrix(il-2, ir+1, jb-1, jt+1);
G = matrix(il-1, ir+1, jb-2, jt+1);
RS= matrix(il, ir, jb, jt);
Flag = imatrix(il-1, ir+1, jb-1, jt+1);
/* Assign initial values to u, v, p */
init_uvp(UI, VI, PI, il, ir, jb, jt, U, V, P);
/* Allocate memory for buffers */
bufSend = malloc(max(ir-il+3, jt-jb+3) * sizeof(double));
bufRecv = malloc(max(ir-il+3, jt-jb+3) * sizeof(double));
/* Initialize lower part of the domain with UI = 0 for the flow_over_step problem */
/* (this code might be moved to somewhere else later) */
if(strcmp(problem, "flow_over_step") == 0) {
init_matrix(U, il, ir, jb, min(jmax/2, jt), 0);
}
/* Initialization of flag field */
init_flag(pgm, imax, jmax, il, ir, jb, jt, Flag, dp);
if(myrank == 0) {
clock_gettime(CLOCK_MONOTONIC, ¤tTime);
}
while(t <= t_end){
/* Select δt */
calculate_dt(Re, tau, &dt, dx, dy, il, ir, jb, jt, U, V);
MPI_Allreduce(&dt, &min_dt, 1, MPI_DOUBLE, MPI_MIN, MPI_COMM_WORLD);
dt = min_dt;
/* Set boundary values for u and v */
boundaryvalues(il, ir, jb, jt, imax, jmax, U, V, wl, wr, wt, wb, Flag);
/* Set special boundary values */
spec_boundary_val(problem, il, ir, jb, jt, imax, jmax, U, V, P, Re, xlength, ylength, dp);
/* Compute F(n) and G(n) */
calculate_fg(Re, GX, GY, alpha, dt, dx, dy, il, ir, jb, jt, imax, jmax, U, V, F, G, Flag);
/* Compute the right-hand side rs of the pressure equation */
calculate_rs(dt, dx, dy, il, ir, jb, jt, imax, jmax, F, G, RS);
/* Perform SOR iterations */
it = 0;
res = 1e6;
while(it < itermax && res > eps){
sor(omg, dx, dy, dp, il, ir, jb, jt, imax, jmax, rank_l, rank_r, rank_b, rank_t, P, RS, &res, Flag, bufSend, bufRecv);
it++;
}
/* Compute u(n+1) and v(n+1) */
calculate_uv(dt, dx, dy, il, ir, jb, jt, imax, jmax, U, V, F, G, P, Flag);
/* Exchange velocity strips */
uv_com(U, V, il, ir, jb, jt, rank_l, rank_r, rank_b, rank_t, bufSend, bufRecv);
t = t + dt;
n++;
/* Generate snapshot for current timestep */
if((int) n % timestepsPerPlotting == 0) {
write_vtkFile(output_dirname, myrank, n, xlength, ylength, il, ir, jb, jt, imax, jmax, dx, dy, U, V, P);
}
/* Print out simulation time and whether the SOR converged */
if(myrank == 0) {
/* Print simulation time */
printf("Time: %.4f", t);
/* Print runtime */
previousTime = currentTime;
clock_gettime(CLOCK_MONOTONIC, ¤tTime);
totalTime += (double)currentTime.tv_sec + 1e-9 * currentTime.tv_nsec - (double)previousTime.tv_sec - 1e-9 * previousTime.tv_nsec;
printf("\tRuntime: %.3f s (avg runtime/step: %.3f s)", totalTime, totalTime/n);
if(res > eps) printf("\tDid not converge (res=%f, eps=%f)", res, eps);
printf("\n");
}
}
/* Close the output folder */
closedir(output_dir);
/* Tell user where to find the output */
if(myrank == 0) {
printf("Please find the output in the folder \"%s\".\n", problem);
}
/* Free allocated memory */
free_matrix(U, il-2, ir+1, jb-1, jt+1);
free_matrix(V, il-1, ir+1, jb-2, jt+1);
free_matrix(P, il-1, ir+1, jb-1, jt+1);
free_matrix(F, il-2, ir+1, jb-1, jt+1);
free_matrix(G, il-1, ir+1, jb-2, jt+1);
free_matrix(RS, il, ir, jb, jt);
free_imatrix(Flag, il-1, ir+1, jb-1, jt+1);
free(bufSend);
free(bufRecv);
MPI_Barrier(MPI_COMM_WORLD);
MPI_Finalize();
return 0;
}
/**
* The main operation reads the configuration file, initializes the scenario and
* contains the main loop. So here are the individual steps of the algorithm:
*
* - read the program configuration file using read_parameters()
* - set up the matrices (arrays) needed using the matrix() command
* - create the initial setup init_uvp(), init_flag(), output_uvp()
* - perform the main loop
* - trailer: destroy memory allocated and do some statistics
*
* The layout of the grid is decribed by the first figure below, the enumeration
* of the whole grid is given by the second figure. All the unknowns corresond
* to a two dimensional degree of freedom layout, so they are not stored in
* arrays, but in a matrix.
*
* @image html grid.jpg
*
* @image html whole-grid.jpg
*
* Within the main loop the following big steps are done (for some of the
* operations a definition is defined already within uvp.h):
*
* - calculate_dt() Determine the maximal time step size.
* - boundaryvalues() Set the boundary values for the next time step.
* - calculate_fg() Determine the values of F and G (diffusion and confection).
* This is the right hand side of the pressure equation and used later on for
* the time step transition.
* - calculate_rs()
* - Iterate the pressure poisson equation until the residual becomes smaller
* than eps or the maximal number of iterations is performed. Within the
* iteration loop the operation sor() is used.
* - calculate_uv() Calculate the velocity at the next time step.
*/
int main(int argn, char** args)
{
double **U, **V, **P, **R, **F, **G;
double t = 0.0;
double t_end, xlength, ylength, tau, omg, alpha, eps, Re, GX, GY, PI, UI, VI, dx, dy, dt, res, dt_value;
int n, imax, jmax, itermax, it, i, outputCount;
double nextOutput;
bool outputSpecified = false;
char* paraviewOutput = "./Paraview/cavity100";
char* inputFile = "./cavity100.dat";
/* statistics */
int sumIter, maxIter, minIter;
double minTimeStep, maxTimestep;
clock_t begin, end;
double time_spent;
sumIter = 0;
maxIter = 0;
maxTimestep = 0;
begin = clock();
/* Parse command line arguments */
if (argn > 2)
{
for (i = 1; i < argn; i++)
{
if (strcmp(args[i], "-i") == 0)
{
inputFile = args[i + 1];
if (!outputSpecified)
paraviewOutput = "./Paraview/sim";
}
if (strcmp(args[i], "-o") == 0)
{
paraviewOutput = args[i + 1];
outputSpecified = true;
}
}
}
else if (argn == 2)
{
/* input file is the single argument besides fctn-name */
inputFile = args[1];
}
/* Initialization */
read_parameters(inputFile, &Re, &UI, &VI, &PI, &GX, &GY, &t_end, &xlength, &ylength, &dt, &dx, &dy, &imax, &jmax, &alpha, &omg, &tau, &itermax, &eps, &dt_value);
U = matrix(0, imax, 0, jmax + 1);
V = matrix(0, imax + 1, 0, jmax);
P = matrix(0, imax + 1, 0, jmax + 1);
R = matrix(0, imax + 1, 0, jmax + 1);
F = matrix(0, imax, 0, jmax);
G = matrix(0, imax, 0, jmax);
n = 0;
nextOutput = dt_value;
init_uvp(UI, VI, PI, imax, jmax, U, V, P);
minIter = itermax;
minTimeStep = t_end;
write_vtkFile(paraviewOutput,0,xlength,ylength,imax,jmax,dx,dy,U,V,P);
outputCount = 1;
while (t < t_end)
{
calculate_dt(Re, tau, &dt, dx, dy, imax, jmax, U, V);
if (dt < minTimeStep)
minTimeStep = dt;
if (dt > maxTimestep)
maxTimestep = dt;
boundaryvalues(imax, jmax, U, V);
calculate_fg(Re, GX, GY, alpha, dt, dy, dy, imax, jmax, U, V, F, G);
calculate_rs(dt, dx, dy, imax, jmax, F, G, R);
it = 0;
while(++it < itermax)
{
sor(omg, dx, dy, imax, jmax, P, R, &res);
if (res <= eps)
break;
}
if (res > eps)
fprintf(stderr, "WARNING: SOR did not converge in timestep %d, residual: %f\n", n + 1, res);
if (it > maxIter)
maxIter = it;
if (it < minIter)
minIter = it;
sumIter += it;
calculate_uv(dt, dx, dy, imax, jmax, U, V, F, G, P);
t = t + dt;
n++;
if (t >= nextOutput)
{
write_vtkFile(paraviewOutput,outputCount,xlength,ylength,imax,jmax,dx,dy,U,V,P);
nextOutput += dt_value;
outputCount++;
}
}
end = clock();
time_spent = (double) (end - begin) / CLOCKS_PER_SEC;
/* print statistics */
printf("Simulation took %f seconds\n", time_spent);
printf("Number of timesteps: %d\n", n);
printf("Min. timestep: %f Max. timestep: %f Avg. timestep: %f\n", minTimeStep, maxTimestep, t_end/n);
printf("Min. # iterations: %d Max. # iterations: %d Avg: %f\n", minIter, maxIter, sumIter/(double) n);
/* Free memory */
free_matrix(U, 0, imax, 0, jmax + 1);
free_matrix(V, 0, imax + 1, 0, jmax);
free_matrix(P, 0, imax + 1, 0, jmax + 1);
free_matrix(R, 0, imax, 0 ,jmax);
free_matrix(F, 0, imax, 0, jmax);
free_matrix(G, 0, imax, 0, jmax);
return 0;
}
int main(int argn, char** args){
if (argn !=2 ) {
printf("When running the simulation, please give a valid geometry file name!\n");
return 1;
}
//set the geometry file
char *filename = NULL;
filename = args[1];
//initialize variables
double t = 0; /*time start*/
int it, n = 0; /*iteration and time step counter*/
double res; /*residual for SOR*/
/*arrays*/
double ***U, ***V, ***W, ***P;
double ***RS, ***F, ***G, ***H;
int ***Flag; //additional data structure for arbitrary geometry
int ***S; //additional data structure for arbitrary geometry
int matrix_output = 0;
/*those to be read in from the input file*/
double Re, UI, VI, WI, PI, GX, GY, GZ, t_end, xlength, ylength, zlength, dt, dx, dy, dz, alpha, omg, tau, eps, dt_value;
int imax, jmax, kmax, itermax;
//double presLeft, presRight, presDelta; //for pressure stuff ...TODO: not allowed for now
int wl, wr, wt, wb, wf, wh;
char problemGeometry[200];
//in case of a given inflow or wall velocity TODO: will we have this? needs to be a vector?
double velIN;
double velMW[3]; // the moving wall velocity is a vector
struct particleline *Partlines = (struct particleline *)malloc((unsigned)(1 * sizeof(struct particleline)));
//char szFileName[80];
//read the parameters, using problem.dat, including wl, wr, wt, wb
read_parameters(filename, &Re, &UI, &VI, &WI, &PI, &GX, &GY, &GZ, &t_end, &xlength, &ylength, &zlength, &dt, &dx, &dy, &dz, &imax,
&jmax, &kmax, &alpha, &omg, &tau, &itermax, &eps, &dt_value, &wl, &wr, &wf, &wh, &wt, &wb, problemGeometry, &velIN, &velMW[0]); //&presLeft, &presRight, &presDelta, &vel);
//printf("d: %f, %f, %f\n", dx,dy,dz);
//int pics = dt_value/dt; //just a helping variable for outputing vtk
double last_output_t = -dt_value;
//double every = t_end/10; //helping variable. Can be used for displaying info every tenth of the progress during simulation
//allocate memory, including Flag
U = matrix2(0, imax+1, 0, jmax+1, 0, kmax+1);
V = matrix2(0, imax+1, 0, jmax+1, 0, kmax+1);
W = matrix2(0, imax+1, 0, jmax+1, 0, kmax+1);
P = matrix2(0, imax+1, 0, jmax+1, 0, kmax+1);
RS = matrix2(1, imax, 1, jmax, 1, kmax);
F = matrix2(0, imax, 1, jmax, 1, kmax);
G = matrix2(1, imax, 0, jmax, 1, kmax);
H = matrix2(1, imax, 1, jmax, 0, kmax);
Flag = imatrix2(0, imax+1, 0, jmax+1, 0, kmax+1); // or Flag = imatrix(1, imax, 1, jmax);
S = imatrix2(0, imax+1, 0, jmax+1, 0, kmax+1);
//S = Flag -> adjust C_x Flags in helper.h
//initialisation, including **Flag
init_flag(problemGeometry, imax, jmax, kmax, Flag, wl, wr, wf, wh, wt, wb); //presDelta, Flag);
init_particles(Flag,dx,dy,dz,imax,jmax,kmax,3,Partlines);
printf("!!!!!!!!%d\n",binMatch(B_NO,B_N));
init_uvwp(UI, VI, WI, PI, Flag,imax, jmax, kmax, U, V, W, P, problemGeometry);
write_particles("particles_init",0, 1, Partlines);
write_imatrix2("Flag_start.txt",t,Flag, 0, imax, 1, jmax, 1, kmax);
//write_flag_imatrix("Flag.txt",t,Flag, 0, imax+1, 0, jmax+1, 0, kmax+1);
write_vtkFile(filename, -1, xlength, ylength, zlength, imax, jmax, kmax, dx, dy, dz, U, V, W, P,Flag);
//write_vtkFile("init", n, xlength, ylength, zlength, imax, jmax, kmax, dx, dy, dz, U, V, W, P);
//going through all time steps
/*
write_matrix2("P_start.txt",t,P, 0, imax+1, 0, jmax+1, 0, kmax+1);
write_matrix2("U_start.txt",t,U, 0, imax+1, 0, jmax+1, 0, kmax+1);
write_matrix2("V_start.txt",t,V, 0, imax+1, 0, jmax+1, 0, kmax+1);
write_matrix2("W_start.txt",t,W, 0, imax+1, 0, jmax+1, 0, kmax+1);
*/
//setting bound.values
boundaryvalues(imax, jmax, kmax, U, V, W, P, F, G, H, problemGeometry, Flag, velIN, velMW); //including P, wl, wr, wt, wb, F, G, problem
// printf("calc bc \n");
while(t < t_end){
/*if(t - every >= 0){
printf("Calculating time %f ... \n", t);
every += t;
}*/
//adaptive time stepping
calculate_dt(Re, tau, &dt, dx, dy, dz, imax, jmax, kmax, U, V, W);
// printf("calc dt \n");
/*
mark_cells(Flag,dx,dy,dz,imax,jmax,kmax,1,Partlines);
set_uvwp_surface(U,V,W,P,Flag,dx,dy,dz,imax,jmax,kmax,GX,GY,GZ,dt,Re);
*/
//computing F, G, H and right hand side of pressue eq.
calculate_fgh(Re, GX, GY, GZ, alpha, dt, dx, dy, dz, imax, jmax, kmax, U, V, W, F, G, H, Flag);
// printf("calc fgh \n");
calculate_rs(dt, dx, dy, dz, imax, jmax, kmax, F, G, H, RS,Flag);
// printf("calc rs \n");
//iteration counter
it = 0;
//write_matrix2("RS.txt",t,RS, 1, imax, 1, jmax, 1, kmax);
//write_matrix2("F.txt",t,F, 0, imax, 1, jmax, 1, kmax);
//write_matrix2("G.txt",t,G, 1, imax, 0, jmax, 1, kmax);
//write_matrix2("H.txt",t,H, 1, imax, 1, jmax, 0, kmax);
do{
// sprintf( szFileName, "P_%d.txt",it );
//write_matrix2(szFileName,1000+t,P, 0, imax+1, 0, jmax+1, 0, kmax+1);
//perform SOR iteration, at same time set bound.values for P and new residual value
sor(omg, dx, dy, dz, imax, jmax, kmax, P, RS, &res, Flag); //, presLeft, presRight);
it++;
}while(it<itermax && res>eps);
/*if (it == itermax) {
printf("Warning: sor while loop exits because it reaches the itermax. res = %f, time =%f\n", res, t);
} */
//write_matrix2("P.txt",t,P, 0, imax+1, 0, jmax+1, 0, kmax+1);
//calculate U, V and W of this time step
calculate_uvw(dt, dx, dy, dz, imax, jmax, kmax, U, V, W, F, G, H, P, Flag);
//write_matrix2("U.txt",t,U, 0, imax+1, 0, jmax+1, 0, kmax+1);
//write_matrix2("V.txt",t,V, 0, imax+1, 0, jmax+1, 0, kmax+1);
//write_matrix2("W.txt",t,W, 0, imax+1, 0, jmax+1, 0, kmax+1);
// printf("calc uvw \n");
//sprintf( szFileName, "simulation/%s.%i_debug.vtk", szProblem, timeStepNumber );
//setting bound.values
boundaryvalues(imax, jmax, kmax, U, V, W, P, F, G, H, problemGeometry, Flag, velIN, velMW); //including P, wl, wr, wt, wb, F, G, problem
/*
set_uvwp_surface(U,V,W,P,Flag,dx,dy,dz,imax,jmax,kmax,GX,GY,GZ,dt,Re);
printf("advance_particles!\n");
advance_particles(dx,dy, dz,imax,jmax,kmax, dt,U,V,W,1,Partlines);
*/
//indent time and number of time steps
printf("timer\n");
n++;
t += dt;
//output of pics for animation
if ( t-last_output_t >= dt_value ){ //n%pics==0 ){
printf("output\n!");
write_particles(filename,n, 1, Partlines);
write_vtkFile(filename, n, xlength, ylength, zlength, imax, jmax, kmax, dx, dy, dz, U, V, W, P,Flag);
printf("output vtk (%d)\n",n);
last_output_t = t;
matrix_output++;
}
printf("timestep: %f - next output: %f (dt: %f) \n",t,dt_value- (t-last_output_t),dt);
}
//output of U, V, P at the end for visualization
//write_vtkFile("DrivenCavity", n, xlength, ylength, imax, jmax, dx, dy, U, V, P);
//free memory
free_matrix2(U, 0, imax+1, 0, jmax+1, 0, kmax+1);
free_matrix2(V, 0, imax+1, 0, jmax+1, 0, kmax+1);
free_matrix2(W, 0, imax+1, 0, jmax+1, 0, kmax+1);
free_matrix2(P, 0, imax+1, 0, jmax+1, 0, kmax+1);
free_matrix2(RS, 1, imax, 1, jmax, 1, kmax);
free_matrix2(F, 0, imax, 1, jmax, 1, kmax);
free_matrix2(G, 1, imax, 0, jmax, 1, kmax);
free_matrix2(H, 1, imax, 1, jmax, 0, kmax);
free_imatrix2(Flag, 0, imax+1, 0, jmax+1, 0, kmax+1);
free_imatrix2(S, 0, imax+1, 0, jmax+1, 0, kmax+1);
//printf("\n-\n");
return -1;
}
/**
* The main operation reads the configuration file, initializes the scenario and
* contains the main loop. So here are the individual steps of the algorithm:
*
* - read the program configuration file using read_parameters()
* - set up the matrices (arrays) needed using the matrix() command
* - create the initial setup init_uvp(), init_flag(), output_uvp()
* - perform the main loop
* - trailer: destroy memory allocated and do some statistics
*
* The layout of the grid is described by the first figure below, the enumeration
* of the whole grid is given by the second figure. All the unknowns correspond
* to a two dimensional degree of freedom layout, so they are not stored in
* arrays, but in a matrix.
*
* @image html grid.jpg
*
* @image html whole-grid.jpg
*
* Within the main loop the following big steps are done (for some of the
* operations a definition is defined already within uvp.h):
*
* - calculate_dt() Determine the maximal time step size.
* - boundaryvalues() Set the boundary values for the next time step.
* - calculate_fg() Determine the values of F and G (diffusion and confection).
* This is the right hand side of the pressure equation and used later on for
* the time step transition.
* - calculate_rs()
* - Iterate the pressure poisson equation until the residual becomes smaller
* than eps or the maximal number of iterations is performed. Within the
* iteration loop the operation sor() is used.
* - calculate_uv() Calculate the velocity at the next time step.
*/
int main(int argn, char** args){
double Re,UI,VI,PI,GX,GY, /* problem dependent quantities */
xlength,ylength,dx,dy, /* geometry data */
t=0,t_end,dt,tau,dt_value, /* time stepping data */
alpha,omg,eps,res; /* pressure iteration data */
int it, itermax, n=0, imax, jmax; /* max iterations, iteration step and # of interior cells */
double **U, **V, **F, **G, **P, **RS;
/* Read the problem parameters */
read_parameters("data/cavity100-2.dat", &Re, &UI, &VI, &PI, &GX, &GY, &t_end, &xlength, &ylength, &dt, &dx, &dy,
&imax, &jmax, &alpha, &omg, &tau, &itermax, &eps, &dt_value);
/* Allocate space for matrices */
U = matrix(0, imax, 0, jmax+1);
F = matrix(0, imax, 0, jmax+1);
V = matrix(0, imax+1, 0, jmax);
G = matrix(0, imax+1, 0, jmax);
P = matrix(0, imax+1, 0, jmax+1);
RS= matrix(0, imax+1, 0, jmax+1);
/* Assign initial values to u, v, p */
init_uvp(UI, VI, PI, imax, jmax, U, V, P);
while (t < t_end){
/* Select δt according to (14) */
calculate_dt(Re, tau, &dt, dx, dy, imax, jmax, U, V);
/* Set boundary values for u and v according to (15),(16) */
boundaryvalues(imax, jmax, U, V);
/* Compute F (n) and G (n) according to (10),(11),(18) */
calculate_fg(Re, GX, GY, alpha, dt, dx, dy, imax, jmax, U, V, F, G);
/* Compute the right-hand side rs of the pressure equation (12) */
calculate_rs(dt, dx, dy, imax, jmax, F, G, RS);
it = 0; res=DBL_MAX;
while (it < itermax && res > eps){
/* Perform a SOR iteration according to (19) using the provided function and retrieve the residual res */
sor(omg, dx, dy, imax, jmax, P, RS, &res);
it++;
}
/* Compute u (n+1) and v (n+1) according to (8),(9) */
calculate_uv(dt,dx,dy,imax,jmax,U,V,F,G,P);
t+=dt; n++;
}
/* Output of u, v, p for visualization */
write_vtkFile("visualization/cavity", n, xlength, ylength, imax, jmax, dx, dy, U, V, P);
/* Freeing memory */
free_matrix(U, 0, imax, 0, jmax+1);
free_matrix(F, 0, imax, 0, jmax+1);
free_matrix(V, 0, imax+1, 0, jmax);
free_matrix(G, 0, imax+1, 0, jmax);
free_matrix(P, 0, imax+1, 0, jmax+1);
free_matrix(RS, 0, imax+1, 0, jmax+1);
return -1;
}
int
main(
int argc, /* arg count */
char * argv[] /* arg vector */
){
static char * context = "main(chain)";
char * stem = NULL; /* dump filename stem */
char * suffix = NULL; /* dump filename suffix */
char * suff2 = NULL; /* last half of suffix */
int nr, nc; /* integer matrix sizes */
int n; /* square matrix/vector size */
real base_x, base_y; /* base of Mandelbrot */
real ext_x, ext_y; /* extent of Mandelbrot */
int limit, seed; /* randmat controls */
real fraction; /* invperc/thresh filling */
int itersLife; /* life iterations */
int itersElastic, relax; /* elastic controls */
int2D i2D; /* integer matrix */
bool2D b2D; /* boolean matrix */
pt1D cities; /* cities point vector */
int n_cities; /* number of cities */
pt1D net; /* net point vector */
int n_net; /* number of net points */
real2D r2D_gauss; /* real matrix for Gaussian */
real2D r2D_sor; /* real matrix for SOR */
real1D r1D_gauss_v; /* real vector input for Gaussian */
real1D r1D_sor_v; /* real vector input for SOR */
real1D r1D_gauss_a; /* real vector answer for Gaussian */
real1D r1D_sor_a; /* real vector answer for SOR */
real1D r1D_gauss_c; /* real vector check for Gaussian */
real1D r1D_sor_c; /* real vector check for SOR */
real tol; /* SOR tolerance */
real realDiff; /* vector difference */
bool choicesSet = FALSE; /* path choices set? */
bool doMandel = TRUE; /* mandel vs. randmat */
bool doInvperc = TRUE; /* invperc vs. thresholding */
bool doDump = FALSE; /* dump intermediate results? */
int argd = 1; /* argument index */
/* arguments */
#if NUMA
MAIN_INITENV(,32000000)
#endif
while (argd < argc){
CHECK(argv[argd][0] == '-',
fail(context, "bad argument", "index", "%d", argd, NULL));
switch(argv[argd][1]){
case 'E' : /* elastic */
itersElastic = arg_int(context, argc, argv, argd+1, argv[argd]);
relax = arg_int(context, argc, argv, argd+2, argv[argd]);
argd += 3;
break;
case 'F' : /* fraction (invperc/thresh) */
fraction = arg_real(context, argc, argv, argd+1, argv[argd]);
argd += 2;
break;
case 'L' : /* life */
itersLife = arg_int(context, argc, argv, argd+1, argv[argd]);
argd += 2;
break;
case 'M' : /* mandel */
base_x = arg_real(context, argc, argv, argd+1, argv[argd]);
base_y = arg_real(context, argc, argv, argd+2, argv[argd]);
ext_x = arg_real(context, argc, argv, argd+3, argv[argd]);
ext_y = arg_real(context, argc, argv, argd+4, argv[argd]);
argd += 5;
break;
case 'N' : /* winnow */
n_cities = arg_int(context, argc, argv, argd+1, argv[argd]);
argd += 2;
break;
case 'R' : /* randmat */
limit = arg_int(context, argc, argv, argd+1, argv[argd]);
seed = arg_int(context, argc, argv, argd+2, argv[argd]);
argd += 3;
break;
case 'S' : /* matrix size */
nr = arg_int(context, argc, argv, argd+1, argv[argd]);
nc = arg_int(context, argc, argv, argd+2, argv[argd]);
argd += 3;
break;
case 'T' : /* SOR tolerance */
tol = arg_real(context, argc, argv, argd+1, argv[argd]);
argd += 2;
break;
case 'c' : /* choice */
CHECK(!choicesSet,
fail(context, "choices already set", NULL));
suffix = arg_str(context, argc, argv, argd+1, argv[argd]);
argd += 2;
switch(suffix[0]){
case 'i' : doInvperc = TRUE; break;
case 't' : doInvperc = FALSE; break;
default :
fail(context, "unknown choice(s)", "choice", "%s", suffix, NULL);
}
switch(suffix[1]){
case 'm' : doMandel = TRUE; break;
case 'r' : doMandel = FALSE; break;
default :
fail(context, "unknown choice(s)", "choice", "%s", suffix, NULL);
}
suff2 = suffix+1;
choicesSet = TRUE;
break;
case 'd' : /* dump */
doDump = TRUE;
argd += 1;
if ((argd < argc) && (argv[argd][0] != '-')){
stem = arg_str(context, argc, argv, argd, argv[argd-1]);
argd += 1;
}
break;
#if GRAPHICS
case 'g' :
gfx_open(app_chain, arg_gfxCtrl(context, argc, argv, argd+1, argv[argd]));
argd += 2;
break;
#endif
#if MIMD
case 'p' :
DataDist = arg_dataDist(context, argc, argv, argd+1, argv[argd]);
ParWidth = arg_int(context, argc, argv, argd+2, argv[argd]);
argd += 3;
break;
#endif
case 'u' :
io_init(FALSE);
argd += 1;
break;
default :
fail(context, "unknown flag", "flag", "%s", argv[argd], NULL);
break;
}
}
CHECK(choicesSet,
fail("context", "choices not set using -c flag", NULL));
/* initialize */
#if MIMD
sch_init(DataDist);
#endif
/* mandel vs. randmat */
if (doMandel){
mandel(i2D, nr, nc, base_x, base_y, ext_x, ext_y);
if (doDump) io_wrInt2D(context, mkfname(stem, NULL, suff2, "i2"), i2D, nr, nc);
} else {
randmat(i2D, nr, nc, limit, seed);
if (doDump) io_wrInt2D(context, mkfname(stem, NULL, suff2, "i2"), i2D, nr, nc);
}
/* half */
half(i2D, nr, nc);
if (doDump) io_wrInt2D(context, mkfname(stem, "h", suff2, "i2"), i2D, nr, nc);
/* invperc vs. thresh */
if (doInvperc){
invperc(i2D, b2D, nr, nc, fraction);
if (doDump) io_wrBool2D(context, mkfname(stem, NULL, suffix, "b2"), b2D, nr, nc);
} else {
thresh(i2D, b2D, nr, nc, fraction);
if (doDump) io_wrBool2D(context, mkfname(stem, NULL, suffix, "b2"), b2D, nr, nc);
}
/* life */
life(b2D, nr, nc, itersLife);
if (doDump) io_wrBool2D(context, mkfname(stem, "l", suffix, "b2"), b2D, nr, nc);
/* winnow */
winnow(i2D, b2D, nr, nc, cities, n_cities);
if (doDump) io_wrPt1D(context, mkfname(stem, "w", suffix, "p1"), cities, n_cities);
/* norm */
norm(cities, n_cities);
if (doDump) io_wrPt1D(context, mkfname(stem, "n", suffix, "p1"), cities, n_cities);
/* elastic */
n_net = (int)(ELASTIC_RATIO * n_cities);
CHECK(n_net <= MAXEXT,
fail(context, "too many net points required",
"number of net points", "%d", n_net, NULL));
elastic(cities, n_cities, net, n_net, itersElastic, relax);
if (doDump) io_wrPt1D(context, mkfname(stem, "e", suffix, "p1"), net, n_net);
/* outer */
n = n_net;
outer(net, r2D_gauss, r1D_gauss_v, n);
if (doDump){
io_wrReal2D(context, mkfname(stem, "o", suffix, "r2"), r2D_gauss, n, n);
io_wrReal1D(context, mkfname(stem, "o", suffix, "r1"), r1D_gauss_v, n);
}
cpReal2D(r2D_gauss, r2D_sor, n, n);
cpReal1D(r1D_gauss_v, r1D_sor_v, n);
/* gauss */
gauss(r2D_gauss, r1D_gauss_v, r1D_gauss_a, n);
if (doDump) io_wrReal1D(context, mkfname(stem, "g", suffix, "r1"), r1D_gauss_a, n);
/* product (gauss) */
product(r2D_gauss, r1D_gauss_a, r1D_gauss_c, n, n);
if (doDump) io_wrReal1D(context, mkfname(stem, "pg", suffix, "r1"), r1D_gauss_c, n);
/* sor */
sor(r2D_sor, r1D_sor_v, r1D_sor_a, n, tol);
if (doDump) io_wrReal1D(context, mkfname(stem, "s", suffix, "r1"), r1D_gauss_a, n);
/* product (sor) */
product(r2D_sor, r1D_sor_a, r1D_sor_c, n, n);
if (doDump) io_wrReal1D(context, mkfname(stem, "ps", suffix, "r1"), r1D_gauss_c, n);
/* difference */
vecdiff(r1D_gauss_a, r1D_sor_a, n, &realDiff);
if (doDump) io_wrReal0D(context, mkfname(stem, "v", suffix, "r0"), realDiff);
#if IEEE
ieee_retrospective(stderr);
#endif
#if NUMA
MAIN_END;
#endif
return 0;
}