feltor-dev / feltor Goto Github PK

Multigrid should cancel when the solution is already the input vector

I'm running the feltorSesol code with the following input.json:
{
"n" : 3, //# x-y-polynomials
"Nx" : 224, //# grid points in x
"Ny" : 224, //# grid points in y
"dt" : 0.0125, //time step in units of c_s/rho_s
"n_out" : 3, //# x-y polynomials in output
"Nx_out" : 224, //# grid points in x in output fields
"Ny_out" : 224, //# grid points in y in output fields
"itstp" : 1000, //steps between outputs
"maxout" : 1000, //# outputs excluding first
"eps_pol" : 1e-5, //accuracy of polarisation solver
"eps_gamma" : 1e-6, //accuracy of Gamma
"eps_time" : 1e-12, //accuracy of implicit time-stepper
"mu_e" : -0.000272121, // -m_e/m_i zb (-0.000544617,-0.000272121,-0.000181372 )
"curvature" : 1.5e-4, //magnetic curvature
"tau" : 1.0, //Ti/Te
"nu_perp" : 2.5e-3, //pependicular viscosity
"D" : 5.0e-11, //coupling (D ~ curvature^2)
"C" : 2.5e-6, //Parallel Resistivity (D lx/d)
"l_para" : 1e6 , //parallel Length ( 2 pi q / mcv)
"amplitude" : 2.5e-2, //amplitude of the blob
"sigma" : 2.5, //blob variance in units of rho_s
"posX" : 0.3333, //blob x-position in units of lx
"posY" : 0.5, //blob y-position in units of ly
"prof_amp": 4.0, //Profile amplitude
"bg_prof_amp" : 1.0, //Background Prof amplitude
"lx" : 200, //lx in units of rho_s
"ly" : 200, //ly in units of rho_s
"bc_x" : "DIR_NEU", //boundary condition in x
"bc_y" : "PER", //boundary condition in y
"bc_x_pot": "DIR_NEU", //boundary condition in x for phi
"hwmode": 1, // ordinary/modified HW (0/1)
"ln": 100, //gradient length (in rho_s)
"SOL_b" : 0.3, //SOL boundary in units of lx (>1 no SOL)
"prof_source_rate" : 0.1 , //profile source rate in units c_s/rho_s
"source_b" : 0.1666666, //source dampingb in u of lx (<1 no Source)
"source_damping_width" : 0.5, //sink, sol and edge damping width
"prof_sink_rate" : 0.1 , //profile source rate in units c_s/rho_s
"sink_b" : 0.8, //sink dampingb in u of lx (<1 no Source)
"damping_width" : 0.5 //source, sink, sol and edge damping width
"fluxmode" : 0 // "gradient" driven (0), flux driven (1)
}

It throws an error:
-bash-4.1$ ./feltor_mpi input.json out.nc
terminate called after throwing an instance of 'dg::Ooops'
what(): No matching boundary condition!

Any ideas?

The runge_kutta_t code produces wrong results when compiled with -O3
When compiled with -O2 everything is fine

msvc reports an error

there seems to be an error in the display of the header navigation bar of the doxygen documentation