These computer programs compute continuous probability densities in 3D observable space and corresponding likelihoods of cluster parameters, based on a set of stellar models in multi-dimensional space. Follow the instructions below to populate the data directory and thus re-produce the analysis for magnitude and vsini data in NGC 1846 in Lipatov, Brandt, and Gossage (2022). Each set of instructions in English corresponds to the immediately following Python code. The file data_populated.zip is a compressed version of the populated data directory.
Install git and Python, then complete the following preliminary steps to obtain the files that serve as input to the programs in this repository.
The MIST model library we utilize has ten distinct rotation rates. It formed the basis for earlier work in Gossage et al 2019.
To filter the requisite information from the MIST model library, first copy parse_mist_iso.py and mist_txt_npy.py into directory mist/, which should contain directories such as feh_m0.15_afe_p0.0_vvcrit0.0_TP. Then run these Python scripts, to obtain text file mist_isochrones.txt and a corresponding numpy array file mist_grid.npy. Move the numpy array into the data/ directory.
cp prelim/parse_mist_iso.py mist/
cp prelim/mist_txt_npy.py mist/
cd mist
python parse_mist_iso.py
python mist_txt_npy.py
mv mist_txt_npy.py ../data
Install PARS. Make sure paint_atmospheres/data contains the filter/ directory with the three filter files from the SVO Filter Profile Service, such as HST_ACS_WFC.F435W.dat (and the same for F555W and F814W). Also make sure that paint_atmospheres/data contains limbdark/ with the intensity files for different metallicities from Dr. R. Kurucz's website, such as im01k2.pck, ip00k2.pck19, and ip02k2.pck. Now run the program that produces data/ldlist.pkl, a list of corresponding limb darkening information files.
cd paint_atmospheres/pa/opt
python grid_limbdark.py
Next, run grid_magnitudes.py to compute data/pars_grid.pkl, the PARS magnitudes on a grid of τ, ω, inclination, γ, metallicity, AV reddening parameter, and filter. This can run a lot faster if one can increase the script's sockets variable, which corresponds to the number of cores on the computer. The output file is larger than it needs to be for a single-metallicity analysis. Move the output file from the PARS data directory to the main analysis data directory.
python grid_magnitudes.py
cd ../..
mv data/pars_grid.pkl ../calc_cluster/data/
Go to the web page with the release of this repository that produces the published analysis, download the source code as a tar.gz file, put the file in the directory where you want to un-compress it.
Un-compress the file and go to the directory with executables.
tar -xf calc_cluster-x.x.x.tar.gz
cd calc_cluster-x.x.x/cc/
File pseudo-code.txt contains the pseudo-code for scripts calc_obs.py, calc_dens.py, and calc_like.py, which appear below. Additionally, the pseudo-code file lists the runtimes and memory output requirements for each script.
File config.py contains variables that are constant throughout the analysis. File load_data.py loads the cluster data ../data/ngc1846*.txt and filters it. A number of scripts in this repository access variables in config.py and load_data.py.
From the above grid of PARS magnitudes, compute a smaller grid at one metallicity, e.g., ../data/pars_grid_ZMm0p45.pkl.
python mist_met_convert.py
CAUTION: run the following command only if you have 100 GB of space on the hard drive for the output.
Compute magnitude, color, and vsini, a.k.a. the observables, on the MIST model grid. Refine the grid to make observable spacing between the models comparable to minimum instrument error. This uses file lib/mist_util.py and places files such as obs_t9p0537.pkl into ../data/observables/. Re-run with ages = 2 in the script for the second half of the ages.
python calc_obs.py
Next, compute the probability densities. This places minimum-error densities across observable space, such as density_t9p0537.pkl, into ../data/densities/pkl/. The script also produces the individual data point densities, evaluated at each data point's observables, such as ../data/points/points_os060_005_015_t9p0537_9p2550.pkl.
python calc_dens.py
Finally, compute cluster likelihoods. This places the log-likelihoods on a grid of cluster parameters, such as ll_m0p45_os060_005_015.pkl, into ../data/likelihoods/pkl/. The script also places the likelihood factors for individual data points at maximum-likelihood parameters, such as lf_m0p45_os060_005_015.pkl, into the same directory.
python calc_like.py
Go to the directory with plot scripts.
cd plot
Each entry below consists of brief figure description, the figure's number in the published work, the script call that produces it, and the resulting figure file.
python 01_plot_omega.py → ../../data/omega_plot.pdf
python 02_plot_pars_diff.py → ../../data/pars_diff.pdf.
python 04_plot_mist.py → ../../data/model_grids/cvmd/mist_[cv]md_t9p1544.pdf.
python 05_plot_densities.py → ../../data/densities/cmd/density_t9p1594_*.pdf, ../../data/densities/vmd/density_t9p1594_*.pdf, and density_dist_*.pdf.
python 08_plot_dP.py → ../../data/normalization/(mag|col)/dP91594_om2_mul1.pdf.
python 09_plot_lf.py → ../../data/likelihoods/png/*_lf_m0p45_os060_005_015.pdf.
python 10_plot_prob.py → ../../data/likelihoods/png/ll_m0p45_os060_005_015_(age|rotation)_prob.pdf.
python 11_plot_diff.py → ../../data/model_grids/png/diff_vs_Mini_t9p1544_Zm0p45.pdf.
python 12_plot_diff_EEP.py → ../../data/diff_EEP.pdf & ../../data/delta_m_delta_t.pdf.
python 13_plot_interp_age.py → ../../data/plots/(L|R)_EEP_interp_age.pdf
python 14_plot_ll.py → ../../data/likelihoods/png/ll_m0p45_os060_005_015_(age|rotation).pdf.
Aaron Dotter helped us understand and work with the MIST stellar evolution models. Nate Bastian and Sebastian Kamann provided the magnitude and vsini data for NGC 1846 that are described in Kamann et al 2020.