Database building

Overview:

GASSPY builds up a database of Cloudy models to cover all of the cells withing a supplied simulation snapshot. To reduce the amount of required models, each cell is not given a model, but rather cells are grouped in phase space, and for each group a model is run. Neighboring models are used to determine if the grouping is fine enough such that the deviation in output emission from a model and a given cell does not exceed user defined criterion. If a grouping is considered to course, a refinement strategy is used to divide the groups further in phase space.

The output is a HDF5 database containing the output emission and absorption spectra for each group, and the mapping from a cell to a group is saved using the Simulation_Reader class.

Running:

Template script can be found at GASSPY/gasspy/scripts/database_building/gasspy_build_database.py. The overall class that is used is the DatabaseCreator class. This needs to be coupled to a ModelRunner class that deals with running the actuall models. currently only one ModelRunner is available (gasspy.physics.sourcefunction_database.cloudy.CloudyModelRunner). Use the gasspy_build_database.py with the appropriate parameters or write your own as

import gasspy
from gasspy.physics.sourcefunction_database.cloudy import CloudyModelRunner

"""
    Initialize simulation readers here.
"""

# initialize the model_runner
model_runner = CloudyModelRunner(gasspy_config, args.rundir, fluxdef)

# initialize database creator
database_creator = gasspy.DatabaseCreator(gasspy_config, model_runner)

# Add snapshots
for sim_reader in sim_readers:
    database_creator.add_snapshot(sim_reader)

# Run models
database_creator.run_models() 

where gasspy_config is the gasspy configuration, either as a dictionary or as a path to a yaml file, the same with the radiation field definition file fluxdef.

gasspy_build_database.py is written to allow for mpi4py usage. To parallelize the running of the cloudy models execute using the command

mpirun -n $number_of_cores python3 /path/to/gasspy_build_database.py --args 

Parameters

The parameters taken from the gasspy_config yaml file are given below. For details of syntax and options see GASSPY/gasspy/templates/gasspy_config_all.yaml

- database_name         : name of the database

- gasspy_modeldir       : directory to save the database

- database_fields       : the fields of the simulation to use in the database

- refinement_fields     : the fields we refine on. Must be a subset of database_fields

- log10_field_limits    : limits of the fields in log space

- fields_lrefine        : mininum and maximum refinement levels of the fields

- convergence_criterions: Limits of the maximal difference between
                          neighboring models, that if exceeded causes refinement.

- line_labels           : specific Cloudy lines we want to save along with 
                          the emission/absorption spectra. 
                          Mainly used in convergence_criterions.

- populator_dump_time   : time intervalls at which the code saves the
                          cloudy models to the database. 

- est_model_time        : estimated model time in seconds. 
                          Used for memory management.

- max_walltime          : maximum wall time the code runs, 
                          after which it safely exits and allows for restarts.

- cloudy_ini            : path to the cloudy ini configuration file that
                          sets all the physical parameters common to all 
                          cells (included chemical models, dust physics etc.)

- cloudy_path : path to the cloudy main directory