Open ivan-pi opened 2 years ago
The code from Filip:
// initialize vortex for (auto i : domain.all()) { double eps = 0.5 * cs; double xc = 0.5; double yc = 0.5; double xx = domain.pos(i)[0] / LX; double yy = domain.pos(i)[1] / LX; double xxc = xx - xc; double yyc = yy - yc; double xxc2 = xxc * xxc; double yyc2 = yyc * yyc; double rh0 = 1.; u[i][0] = u0 - eps * (yyc / Rc) * exp(-(xxc2 + yyc2) / (2. * Rc * Rc)); u[i][1] = eps * (xxc / Rc) * exp(-(xxc2 + yyc2) / (2. * Rc * Rc)); rho[i] = rh0 * (1. - 0.5 * eps * eps / (cs * cs) * exp(-(xxc2 + yyc2) / (Rc * Rc))); }
The benchmark is explained in the paper:
Wissocq, G., Boussuge, J. F., & Sagaut, P. (2020). Consistent vortex initialization for the athermal lattice Boltzmann method. Physical Review E, 101(4), 043306. https://doi.org/10.1103/PhysRevE.101.043306
The code above uses the first-order Taylor expansion in density. The paper seems to recommend, the true barotropic initialization instead.
The code from Filip:
The benchmark is explained in the paper: