Code developed to work with (largely) cool dwarf spectra from the WiFeS instrument on the ANU 2.3 m Telescope. Functionality:

• Data reduction/preparation--flux calibration, RV determination (see Žerjal+2021), normalisation, etc.
• Synthetic fitting for parameter determination (per Rains+2021) using MARCS model spectra (grid from Nordlander+2019) and photometry.
• Light curve fitting of TESS light curves (per Rains+2021).
• Data-driven fitting for parameter determination using the Cannon (per Rains+2024).

1D WiFeS spectra should be extracted as individual fits files using the functions extract_stellar_spectra_fits and extract_stellar_spectra_fits_all in pywifes_utils.py on the PyWiFeS repository. These fits files maintain the header information from the telescope, and contain at least the unfluxed 1D spectra in units of counts, and optionally the PyWiFeS flux and telluric corrected spectra.

The following two scripts import/consolidate the 1D WiFeS spectra into a single fits file, and fit synthetic templates to determine radial velocities. If spectra are not yet fluxed, they can be corrected for the mean atmospheric/telescope/instrument response function here. Generates bad pixel masks using the residuals from the best template fit.

• scripts_reduction/import_spectra.py
• scripts_reduction/determine_rvs.py

Performs up to a 4 term spectral fit in T_eff, logg, [Fe/H], and m_bol using synthetic MARCS spectra and bolometric corrections. MARCS spectra are interpolated using IDL routines run from python using the pidly library, and bolometric corrections from bolometric-corrections. Literature photometry required for the m_bol fit, with Gaia, 2MASS, and SkyMapper currently supported. logg can be fixed to the value determined from empirical mass and radius relations.

• scripts_synth/synth_fit.py

Can also generate synthetic spectra for the sample at the literature stellar parameters for the sample.

• scripts_synth/get_lit_param_synth.py

Note that the synthetic MARCS grid used for fitting is not currently publicly available.

Uses the Lightkurve package to download and work with TESS light curves, and the batman package to model the transits themselves. Given the known orbital period, and estimated M_★, we apply a prior for the semi-major axis during fitting. Limb darkening coefficients are chosen based on previously fitted stellar parameters, and fitted R_★ is used to determine the unscaled R_P and semi-major axis.

• scripts_lc/lightcurve_fit.py

The Cannon is implemented using pyStan 2, and there are currently three (T_eff, logg, [Fe/H]) and four (T_eff, logg, [Fe/H], [Ti/Fe]) label models. The core Cannon scripts (in order) are:

• scripts_cannon/prepare_stannon_training_sample.py
• scripts_cannon/train_stannon.py
• scripts_cannon/make_stannon_diagnostics.py
• scripts_cannon/compare_stannon_models.py
• scripts_cannon/run_stannon.py

With the various settings required by these scripts stored in scripts_cannon/cannon_settings.yml.

Note that the trained three and four label Cannon models themselves from Rains+2024 are available in the models folder, and a general use script for using these models to generate spectra and predict labels is available as scripts_cannon/example_cannon_script.py. This script is separate from the sequence outlined above, and is the thing to look at if you want to play around with the Cannon. Please get in touch if you have questions!

For full functionality, the non-anaconda dependencies are:

• pystan2
• M_-M_K-
• pIDLy
• Lightkurve
• batman
• dustmaps

However, if you just want to use the Cannon, pystan2 is technically (i.e. functionally) the only thing required.