smacc-parallel is a parallel MPI-based implementation of SMACC (Subhalo Mass-loss Analysis using Core Catalogs).
The code takes GenericIO Core Catalog (.coreproperties) files as input and evolves the mass of satellite cores at each simulation snapshot according to a subhalo mass loss model.
The Core Catalog with an added evolved core mass column is saved to disk in HDF5 format for each snapshot.
See the paper for detailed information on SMACC and the mass loss model.
To get started, clone the repository with
git clone https://github.com/isulta/smacc-parallel
Ensure that the required Dependencies are installed.
Next, set the Input parameters in input_params.yaml.
mpiexec -n $NRANKS python cc_generate_parallel.py
Note that the above command to start the parallel MPI processes may be different depending on the system.
Sample scripts to run smacc-parallel on various HPC systems are provided in job_submission_scripts/.
For each snapshot, smacc-parallel will save the output HDF5 Core Catalog as a .corepropertiesextend.hdf5 file with
- a limited set of columns from the input GenericIO Core Catalog (see Data Columns for fine-tuning which columns are written)
m_evolved_{A}_{zeta}: modeled satellite core mass column (for central cores, this column has value 0)
All parameters for smacc-parallel are defined in input_params.yaml:
| Simulation | |
|---|---|
SIMNAME |
Set to LJ for Last Journey, SV for Last Journey-SV, HM for Last Journey-HM, AQ for AlphaQ, or FP for Farpoint. (See Other Simulations if using a simulation not listed here.) |
| Mass Loss Model | |
AFID |
A mass model parameter |
ZETAFID |
ζ mass model parameter |
useLocalHost |
If True, smacc-parallel will use the immediate parent (sub)halo mass for M in the mass model. If False, smacc-parallel will use the host halo mass for M. |
DELTATFACTOR |
Fraction of the time between the snapshot at which infall is detected for a core and the preceding snapshot, over which to compute the initial mass evolution of the new satellite. |
| File Input/Output | |
cc_input_template |
File path of GenericIO Core Catalog (.coreproperties) header files. Use an asterisk in place of the snapshot number. smacc-parallel will automatically create a list of all snapshots. |
cc_output_dir |
Directory to save output Core Catalog files with m_evolved_{A}_{zeta} column (.corepropertiesextend.hdf5) |
writeOutputFlag |
If True, smacc-parallel will save output Core Catalog files at each snapshot to cc_output_dir. |
save_cc_prev |
If True, smacc-parallel will also save "resume-files" (.ccprev.hdf5) at each snapshot to cc_output_dir. This allows smacc-parallel to resume from a later snapshot instead of having to restart if a run gets interrupted (See Resume SMACC). |
| Data Columns | |
extra_columns |
List of extra columns to read/write, in addition to the default columns required by the mass model (see Data Columns). If parameter is set to [], only the default columns will read/written. |
extra_columns_dtypes |
List of the data types of the columns given in extra_columns (each element must be i 32, i 64, or f 32). Parameter must be set to [] if extra_columns is []. |
steps_to_read_extra_columns |
List of snapshots at which to read/write the columns given in extra_columns, in addition to the default columns. For snapshots not in steps_to_read_extra_columns, only the default columns will be read/written. If this parameter is set to all, the columns given in extra_columns will be read/written for every snapshot. Parameter must be set to [] if extra_columns is []. |
| Resume SMACC | |
resume_smacc |
If True, smacc-parallel will start from snapshot resume_step. If False, smacc-parallel will start from the first snapshot. |
resume_dir |
Location of resume-files N.*.ccprev.hdf5 for snapshot N which immediately precedes snapshot resume_step. The resume-files must have been written with the same number of ranks/machine topology as the current run. (Define if resume_smacc is set to True.) |
resume_step |
Snapshot at which to resume smacc-parallel. (Define if resume_smacc is set to True.) |
Python 3 is required (smacc-parallel has been tested for Python 3.8.5). Running in a conda environment is recommended.
The following packages are also required:
- itk
- pygio
- HDF5 and H5py (both must be built with parallel support enabled)
- Astropy
- Numpy
- Matplotlib
- mpi4py
- pyyaml
- natsort
- SciPy
- numba
By default, smacc-parallel read/writes a limited set of columns of the input GenericIO Core Catalog to minimize runtime. The following default columns are required for the mass model and are always read:
core_tag, tree_node_index, infall_tree_node_mass, central, host_core
The default columns are written for every snapshot.
Additional columns can be added to the output HDF5 Core Catalog by adding the column names to extra_columns, and adding the data types of the new columns to extra_columns_dtypes.
The columns in extra_columns will be written at only the snapshots that are given in steps_to_read_extra_columns (see Input parameters).
To use smacc-parallel with a simulation not listed above, assign a key and add the key and simulation/cosmological parameters to simulationParams.yaml in itk.
smacc-parallel will then import the parameters if SIMNAME is set to the new key.
Note that itk only supports simulations with a cosmology close to the current ΛCDM best-fit cosmology; itk assumes Ωr,0 = Ωk,0 = 0.
By default, smacc-parallel uses the step number-to-redshift formula given below, which is valid for the HACC simulations listed above (Last Journey, LJ-SV, LJ-HM, AlphaQ, and Farpoint).
If using a simulation with a different step-to-redshift formula, the function redshift in itk.py must be updated.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent