Variables are 1D arrays; which maintain an order same as the cells/vertices/faces indices in a grid. In addition to values, variables also contain boundary condition information.
This example creates a 5x5 block using quads in a unit 2D domain. A new variable is defined and named as f; where boundary conditions on east and north are defined using Neumann and Dirichlet conditions. For south and west, default boundary condition, i.e. zero gradient, is used.
Typical loop over celllist is given in lines xx-xx, where c->id gives the array’s index.
Grid adaptation based on gradient of the function is used to refine base grid.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
double pi = 4*atan(1.0);
Block2* volgrid = new Block2({0,0,0}, {1,1,0}, 5, 5);
// add a new variable
volgrid->addVar("f");
auto f = volgrid->getVar("f"); // variable handle
f->setBC("east", "grad", 0); // This is the default
f->setBC("north", "val", 1); //
for (auto i=0; i < 4; ++i) {
for (auto c : volgrid->listCell) {
auto x = c->getCoord(); // cell-centers
f->set(c->id, sin(3*pi*x[0])*cos(2*pi*x[1]));
}
volgrid->solBasedAdapt2(volgrid->getError(f));
volgrid->adapt();
volgrid->writeVTK("field_");
}
delete(volgrid);
Above code produces the following result:
