ASCL.net

Astrophysics Source Code Library

Making codes discoverable since 1999

ASCL Code Record

[ascl:2101.011] Nigraha: Find and evaluate planet candidates from TESS light curves

Nigraha identifies and evaluates planet candidates from TESS light curves. Using a combination of high signal to noise ratio (SNR) shallow transits, supervised machine learning, and detailed vetting, the neural network-based pipeline identifies planet candidates missed by prior searches. The pipeline runs in four stages. It first performs period finding using the Transit Least Squares (TLS) package and runs sector by sector to build a per-sector catalog. It then transforms the flux values in .fits lightcurve files to global/local views and write out the output in .tfRecords files, builds a model on training data, and saves a checkpoint. Finally, it loads a previously saved model to generate predictions for new sectors. Nigraha provides helper scripts to generate candidates in new sectors, thus allowing others to perform their own analyses.

Code site:
https://github.com/ExoplanetML/Nigraha
Described in:
https://ui.adsabs.harvard.edu/abs/2021arXiv210109227R
Bibcode:
2021ascl.soft01011R

Views: 178

ascl:2101.011
Add this shield to your page
Copy the above HTML to add this shield to your code's website.