A stellar abundance matching code
Project description
Python package for matching stellar abundance measurements against a database of model stellar explosions. Based on the old IDL code.
StarFit can match combined abundances of multiple models. For single stars and combinations of two stars, a complete search can be found. For three or more stars, the problem is extremely expensive, so a Genetic Algorithm is used to find an approximate solution.
An online interface (with a subset of functionality) is available at starfit.org.
Installation
Tested with Python 3.10
Optional: A working LaTeX installation and dvipng is required to create plots with LaTeX labels (ideal for publication). Otherwise, Matplotlib's default MathText is used, which may not render all symbols correctly.
From PyPI (recommended)
pip install starfit
The PyPI package includes the necessary data files.
Developer instructions
The data files are not included into the Git repo, and must first be downloaded from the web-server before installing from the Git repo.
git clone git@github.com:conradtchan/starfit.git
cd starfit
# Download data files
./download-data.sh
# Set environment variable to allow for editable installs
export SETUPTOOLS_ENABLE_FEATURES="legacy-editable"
# "-e" creates an editable install, "[testing]" installs additional dependencies for testing
pip install -e .[testing]
# Run all tests
python -m pytest
Usage
starfit.Single
fits an abundance pattern to a single model from the database.
Optional arguments:
combine
: a list of lists of element charge numbers to treat as combined abundances (e.g. combine the CNO elements)z_max
: highest element charge number to fitz_exclude
: element charge numbers to exclude from fitz_lolim
: elements that are model lower limits (effectively the same as observational upper limits)upper_lim
: include observational upper limits in data fittingcdf
: use the uncertainty of upper limits to calculate a cumulative distribution function when calculating error contribution (otherwise treat the upper limit as a simple one-sided chi-squared error)
import starfit
s = starfit.Single(
filename = 'HE1327-2326.dat',
db = 'znuc2012.S4.star.el.y.stardb.gz',
combine = [[6, 7, 8]],
z_max = 30,
z_exclude = [3, 24, 30],
z_lolim = [21, 29],
upper_lim = True,
cdf = True,
)
print(s)
The database indices of the best fitting models (sorted from best to worst) are given by:
s.sorted_stars['index']
The corresponding reduced chi^2 values are:
s.sorted_fitness
The physical properties of the models corresponding to these indices can be accessed using the database:
i_bestfit = s.sorted_stars['index'][0]
s.db.fielddata[i_bestfit]
The chemical yield of the models (and the respective element names) are:
list( zip(s.list_db, s.full_abudata[:,i_bestfit]) )
To make the same plots as the web version:
s.plot()
starfit.Double
fits an abundance pattern to a combination of two models from the database. This takes approximately 1 hour, depending on your machine.
Additional arguments:
fixed
: Use dilution factors based on the ejecta mass, rather than solving for the optimal dilution ratio of each explosion independently (decreases solve time)threads
: Number of threads to use. Default is t use CPU could (including hyprthreading)nice
: Nice level of background threads. Default is 19 (lowest priority on unix systems).
s = starfit.Double(
filename = 'HE1327-2326.dat',
db = 'znuc2012.S4.star.el.y.stardb.gz',
combine = [[6, 7, 8]],
z_max = 30,
z_exclude = [3, 24, 30],
z_lolim = [21, 29],
upper_lim = True,
cdf = True,
fixed = False,
)
starfit.Ga
fits an abundance pattern to a combination of two or more models from the database. The solution is approximate, but approaches the best solution with increased run time.
Additional arguments:
time_limit
: amount of time (in seconds) to search for solutionsol_size
: number of explosion models to find combinations ofpop_size
: GA parameter - number of solutions in the populationtour_size
: GA parameter - number of solutions per tournament selectionfrac_mating_pool
: GA parameter - fraction of solutions in the mating poolfrac_elite
: GA parameter - top fraction of elite solutionsmut_rate_index
: GA parameter - mutation rate of the database indexmut_rate_offset
: GA parameter - mutation rate of the dilution factormut_offset_magnitude
: GA parameter - size of the mutation of the dilution factorlocal_search
: GA parameter - solve for the best dilution factors rather than relying on the GA
The default GA parameters should be used unless you really know what you are doing.
s = starfit.Ga(
filename = 'HE1327-2326.dat',
db = 'znuc2012.S4.star.el.y.stardb.gz',
combine = [[6, 7, 8]],
z_max = 30,
z_exclude = [3, 24, 30],
z_lolim = [21, 29],
upper_lim = True,
cdf = True,
time_limit=20,
sol_size=3,
)
Custom data directory
Custom stellar data and model database files can always be used by providing a full path in the argument. However, users may optionally specify their own data directory using the environment variable STARFIT_DATA
for convenience:
export STARFIT_DATA='/your/custom/data'
The custom data directory should have the same structure as src/starfit/data
, i.e. it should contain the db
, ref
, and stars
directories:
❯ ls
db
ref
stars
Specifying a custom data directory will cause the default data directory to be ignored. This behaviour is intended to prevent ambiguity between file versions. Users opting to use a custom data directory are responsible for managing all of their own versions of data files.
If only one (or a subset) of data files needs to be specified, the default files should be copied from the default directory.
Contributing to StarFit
Contributions to the StarFit code are welcome. The master
branch is protected and cannot be committed to directly. Instead, please create a Pull Request with your proposed contributions. Two automated checks (on Github Actions) must be passed:
- Code formatting using pre-commit. To ensure your changes are compliant with this project's linters, we recommend installing pre-commit prior to making any commits locally.
pip install pre-commit
pre-commit install
If you have already made non-compliant commits prior to installing pre-commit, then the pre-commit check on Github will fail. To make the code compliant again, run
pre-commit run --all
and include these changes in a follow-up commit.
- Code tests using
pytest
. New tests can be added to thetests/
directory.
Adding new data files
Data files specified in the .hashlist
files in src/starfit/data/db,ref,stars
are downloaded from the web server. To add new data files:
- Add the new files to the web server hosting the data files at
/var/www/html/data
- Generate the hash using
shasum -a 256
(orsha256sum
) - Add an entry into the hash list
When adding new databases into data/db
, add corresponding labels into the file data/db/labels
. This label is used in the web application.
Publishing to PyPI
Github releases will automatically be published to https://pypi.org/project/starfit/
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.