Makefile Parameters

Most of the configuration options in Cholla currently require turning on or off flags in a make.type file. This page contains most of the available options, along with a brief summary of their meanings.

Makefile Options

• CUDA: Turns on GPU version of Cholla. The non-GPU version is no longer supported.

• CUDA_ERROR_CHECK: Turns on optional GPU error checking. Can be helpful for debugging GPU platform errors.

• MPI_CHOLLA: Turns on MPI version of Cholla. Output files will be appended with the process number.

• BLOCK: Tells Cholla to use a block-based MPI decomposition. Only works for even numbers of processes, otherwise Cholla will default to a slab-based decomposition. Slab decompositions are not supported in the gravity version of Cholla.

• SET_MPI_GRID: Set the MPI process grid decomposition manually. Can be useful for scaling tests, etc.

• PRECISION = 1 or 2 : Run with single or double precision.

• N_STEPS_LIMIT = : Limit the total number of steps taken to a certain number (useful primarily for testing).

Output options

• OUTPUT: Output data files. If neither BINARY or HDF5 are turned on, will default to text outputs.
• BINARY: Output binary files. No longer supported, but code exists.
• HDF5: Output hdf5 files. Requires the use of the hdf5 library when compiling.
• N_OUTPUT_COMPLETE = : Output full grid every N files. Can also be accomplished with the "nfull" parameter in the input file.
• SLICES: For 3D simulations, output x-z and y-z slices of density, x momentum, y momentum, z momentum, total energy, thermal energy, and passive scalars. Only compatible with HDF5 outputs.
• PROJECTION: For 3D simulations, output x-z and y-z projections of density and density-weighted temperature. Only compatible with HDF5 outputs.
• ROTATED_PROJECTION: For 3D simulations, output rotated projections of density and density-weighted temperature. Only compatible with HDF5 outputs.

Spatial reconstruction methods

• PCM: Use piecewise constant reconstruction for hydro integrator. Very diffusive, useful primarily for testing.
• PLMP: Use piecewise linear reconstruction with limiting in the primitive variables. Follows the MUSCL scheme outlined in Toro to advance interface states in time (2006).
• PLMC: Use piecewise linear reconstruction with limiting in the characteristic variables. Follows the procedure outlined in Stone (2008).
• PPMP: Use piecewise parabolic reconstruction with limiting in the primitive variables. Follows the procedure outlined in Fryxell (2000). Contact discontinuity steeping and shock flattening are turned off by default, but can be turned on in the "ppmp_cuda.cu" source file.
• PPMC: Use piecewise parabolic reconstruction with limiting in the characteristic variables. Follows the procedure outlined in Stone (2008).

Riemann Solvers

• EXACT: Use an exact Riemann solver, as described in Toro (2006).
• ROE: Use a Roe Riemann solver, as described in Stone (2008).
• HLLC: Use an HLLC Riemann solver, as described in Stone (2008).

Hydro Integrators

• VL: Use the Van Leer predictor-corrector method, as described in Stone (2009)
• SIMPLE: Use a simple dimensionally-split PLM or PPM style integrator, as described in Schneider (2017)

Cooling functions

• COOLING_GPU: Use one of the cooling functions defined in cooling_cuda.cu. By default, a piecewise parabolic analytic fit to a CIE cooling curve that cuts off at 10^4 K.
• CLOUDY_COOL: Use a cooling / heating table created with Cloudy assuming solar metallicity and a HM2012 UV background to apply cooling / heating (see Schneider 2017).
• COOLING_GRACKLE:

Self Gravity

• GRAVITY: Turns on the gravity solver for gas self-gravity; requires choosing a solver (see below)
• GRAVITY_GPU: Gravity arrays reside in GPU memory only (are not copied over to the CPU)

One of the following two solvers must be selected:

• PARIS: A block based 3D FFT solver
• SOR: A successive over relaxation solver (currently not supported with GRAVITY_GPU)

One of the following two couplings must be selected:

• COUPLE_GRAVITATIONAL_WORK: Gravitational work is calculated directly and added to the total energy
• COUPLE_DELTA_E_KINETIC: Change in kinetic energy due to gravity term added to momentum is calculated, this is added to the total energy

Optional parameters:

• OUTPUT_POTENTIAL: Write out the potential to file
• GRAVITY_5_POINTS_GRADIENT: Use a 5 point approximation when estimating the potential at the cell center (by default 3 point is used)

Particle Gravity

Note: Particles can be used alone, but in order to use particles + hydro, FFT Gravity must also be turned on

• PARTICLES: Turn on particle solver
• PARTICLES_CPU: Use the CPU to evolve the particles
• PARTICLES_GPU: Use the GPU to evolve the particles
• ONLY_PARTICLES: Run an n-body simulation only (no hydro)
• SINGLE_PARTICLE_MASS: Set a single mass for all the particles (if not selected, particle masses must be set in initial conditions)
• PARTICLE_IDS: Assign IDs to particles

Additional options

• DE: Use the dual energy formalism to advect / correct estimates for internal energy, as described in Bryan (2014)

• SCALAR: Evolve additional scalar quantities - by default 1, but the number can be changed by changing NSCALARS in global.h

• STATIC_GRAV: Turn on static gravity. Currently gravitational forces are added by hand by modifying the relevant functions in "gravity_cuda.cu".

• DENSITY_FLOOR: Apply a density floor to all cells. The specific numerical value can be changed by changing DENS_FLOOR in global.h. Note the the associated physical value will depend on the code units (also defined in global.h).

• TEMPERATURE_FLOOR: Apply a temperature floor to all cells. The specific numerical value can be changed by changing TEMP_FLOOR in global.h

• AVERAGE_SLOW_CELLS: If a cell is causing the timestep to become very small relative to a set number, average the conserved quantities for that cell with its nearest neighbors. This requires "min_dt_slow" to be set, which is done by default only when GRAVITY is turned on.

• CPU_TIME: Measure / print out some timing statistics for a simulation as it runs.