Routines to run and process a grid of MIST models on the Odyssey computing cluster at Harvard.
- a customized version of MESA v9691 : http://mesa.sourceforge.net/ (http://waps.cfa.harvard.edu/MIST/data/tarballs_v1.0/MESA_files.tar.gz)
- mesasdk (for MESA installation) : http://www.astro.wisc.edu/~townsend/static.php?ref=mesasdk
- iso : https://github.com/dotbot2000/iso
- initial_xa_calculator : https://github.com/dotbot2000/initial_xa_calculator
The following environment variables must be set in your e.g.,
$ISO_DIR= path to the iso directory.
$MIST_CODE_DIR= path to this directory.
$XA_CALC_DIR= path to the abundance calculator directory.
$MIST_GRID_DIR= path to where the MIST models will be computed.
$STORE_DIR= path to where the processed tarballs will be stored.
$MESAWORK_DIR= path to the directory where
cleanworkdirwill be stored.
inlists is a directory where the MESA inlists that are generated for each MIST grid are stored. The inlists are then copied to individual mass directories for each MIST grid.
cleanworkdir is copy of the
mesa/star/work directory. This serves as the template directory for each mass directory.
The two high-level scripts are
reduce_jobs.py. Files related to MESA, such as the template inlists, can be found in the
scripts contains all the necessary helper functions.
Submitting a grid of models
./submit_jobs.py my_first_MIST_grid -1.0 0.0
This will run a grid of models with [Fe/H]=-1 and [a/Fe]=0.
It creates a directory named
my_first_MIST_grid directory contains a number of subdirectories
XXXXXM_dir for each mass, e.g.,
00100M_dir for a 1 Msun model. Each subdirectory is populated with everything, e.g., a customized
inlist, that is required to run a single model. The
initial_xa_calculator generates an input file of initial abundances for the user-supplied value of [Fe/H] and [a/Fe].
The choice of masses is set in
scripts/make_inlist_inputs.py, and the input physics can be modified in
mesafiles/run_star_extras.f, though doing so requires familiarity with MESA. The jobs are submitted using
SLURM, where the exact choices of queue name, number of nodes, etc. can be set in the template file
The MESA history files are written out to
$MIST_GRID_DIR/my_first_MIST_grid/XXXXXM_dir/LOGS. The MESA terminal output, which is useful for checking the computation progress in real-time, is found in
$MIST_GRID_DIR/my_first_MIST_grid/XXXXXM_dir/XXXXXM.o. Any error messages are written out to
Processing a grid of models
This will process the MESA history files and call Aaron Dotter's iso package to generate EEP tracks and isochrones.
This script goes through multiple steps.
- It first renames the grid by attaching a
- It creates a
my_first_MIST_grid, which contains the subdirectories
- It copies over the MESA history files and renames them to
XXXXXM.track. It cuts out any redundant model steps due to e.g., retries in the MESA computation and removes any features due to numerical noise that may occur during the post-AGB phase.
- It analyzes the
XXXXXM.efiles to generate a summary file,
tracks_summary.txt, of all the individual MESA runs.
- It copies over the individual inlists to
my_first_MIST_grid/inlistsand renames them to
- If there are any MESA photos or models saved during the computation to allow for convenient restarts, they are also saved to
models_photosafter being renamed
- It calls the
isopackage to make the EEP files and isochrones. There is an option to choose between
full. As the names suggest,
fullis just more comprehensive and contains more MESA columns. See
mesafiles/my_history_columns_full.listfor more information.
- It plots the individual EEPs and isochrones and saves them as pdfs (Optional).
- It saves the original MESA history data files separately as
my_first_MIST_grid_raware compressed and moved over to
$STORE_DIRfor permanent storage.
my_first_MIST_grid-related is removed from
scripts/README.md for more information.