Castro (Compressible Astrophysics): An adaptive mesh, astrophysical compressible (radiation-, magneto-) hydrodynamics simulation code for massively parallel CPU and GPU architectures.
Lot's of code fixes from coverity and clang-tidy (#2736, #2734, #2735, #2731, #2732, #2733)
Fix the boundary condition logic at a wall for Detonation (#2722)
Reimplement the shock detection algorithm to account for sources and do a better job in multidimensions (#2711, #2710, #2709, #2704)
Start the process of moving the runtime parameters to structs (#2688)
An option for unlimited PPM reconstruction was added (#2670)
An option allowing for optional passive sources to the conserved state was added (#2678)
A script diag_parser.py
was added to allow for easy parsing of
the global diagnostic files output at runtime (#2666, #2667)
Problem GNUmakefiles have been standardized and now allow for the problem to be compiled elsewhere (#2640, #2641, #2642, #2643)
The true-SDC Newton solver has been made more robust and faster (#2586, #2602, #2605, #2606)
Several problems that required the initial model grid spacing to
be specified in the inputs file now automatically compute it as
needed, including flame_wave
(#2610), convective_flame
,
bubble_convergence
, and hse_convergence
(#2624), double_bubble
,
gamma_law_bubble
, and hse_convergence_general
(#2612)
Outflow boundary conditions for the 4th order solver have been changed to no longer use the one-sided stencil (#2607)
The ca_rad_source hook in Fortran has been removed. The existing problem_rad_source() hook in C++ can be used instead. (#2626)
The compile option USE_AUX_UPDATE has been removed. If you want to manually update the auxiliary parameters, you can use an external source term or you can use the problem post-timestep hook. (#2614)
The pressure is now always included in the x-momentum flux in 1-d Cartesian, and this fixes an issue at jumps in refinement with the pressure gradient (#2468)
A bug was fixed in the 4th order diffusion operator that was introduced when it was originally converted to C++ (#2592)
The 2nd order Radau integrator had the wrong quadrature weights (#2594)
True-SDC no longer evolves density as part of the reaction system and now uses the same ODE code path as simplified-SDC. This means we don't need our own custom VODE righthand side functions (#2559, #2560, #2567, #2578, #2580, #2584)
The true SDC runtime parameter sdc_solve_for_rhoe
has been
removed. (#2572)
The true SDC runtime parameters sdc_solver_tol_spec
,
sdc_solver_tol_ener
, sdc_solver_atol
have been removed.
Instead the Microphysics integration tolerance parameters should
be used. (#2571)
The true SDC runtime parameter sdc_newton_use_analytic_jac
has
been removed. Instead the Microphysics integrator jacobian
parameter should be used (#2573)