Open Source Program Office
Open Sourcing the Wisconsin Idea
Melinda Soares-Furtado, (mmsoares@wisc.edu)
This astronomy research project focuses on developing a machine learning tool to obtain reliable stellar ages from rotation periods. The project will create an open-source Python package called the “Machine Learning Gyrochronology Age Estimator” that automatically estimates stellar ages using gyrochronology—the method of dating stars by measuring how fast they rotate.
Stars slow their rotation as they age due to magnetic braking. By measuring a star’s rotation period and comparing it to calibration samples (star clusters with known ages), we can estimate stellar ages. This is critical for understanding planetary system evolution, stellar physics, and galactic history. Unlike simple interpolation methods, this tool will use sophisticated machine learning algorithms to capture complex relationships between rotation periods, stellar properties (color, mass, metallicity), and age while providing reliable uncertainty estimates.
Create an end-to-end automated pipeline that takes only a Gaia DR3 source ID (stellar identifier) as input and automatically: (1) queries stellar parameters and time-series photometry, (2) computes rotation periods from light curves, and (3) estimates ages with error bars using ML models trained on benchmark clusters spanning 50 million to 3+ billion years old.
Concept phase. This project would be developed from scratch using Python, leveraging existing astronomy packages (astroquery, lightkurve, astropy) and ML frameworks (scikit-learn, PyTorch, TensorFlow). Training datasets will be provided by the research team.
The complete project will be made publicly available on GitHub with comprehensive documentation, tutorials, and contribution guidelines. This will benefit the broader astronomy community and provide the student intern with a prominent portfolio piece and first-author publication opportunity.
The intern will contribute across the full development pipeline of this astronomy software project. Core responsibilities include implementing automated data retrieval systems in Python to query Gaia DR3, TESS, ZTF, and other astronomical archives with cross-matching algorithms; developing time-series analysis routines to detect stellar rotation periods from light curves using algorithms like Lomb-Scargle periodograms and Gaussian processes; building and optimizing machine learning models trained on provided cluster datasets to predict stellar ages with uncertainty quantification; and creating comprehensive documentation, tutorials, and a user-friendly command-line interface. The intern will also contribute to scientific analysis, figure preparation, and manuscript writing as first author (or second author if preferred) on a peer-reviewed journal article.
The intern must be fluent in Python programming and familiar with machine learning tools and frameworks such as scikit-learn, PyTorch, or TensorFlow, as all development will be conducted in Python. Essential skills include experience with data analysis and visualization libraries (numpy, pandas, matplotlib), understanding of ML concepts including training/validation/testing, overfitting, regularization, and cross-validation, and comfort with Git/GitHub for version control. Strong problem-solving abilities, independent work skills, and excellent written and verbal communication are required, along with genuine interest in astronomy and willingness to learn about stellar physics.
The ideal candidate is a sophomore or junior undergraduate student (preferred to enable multi-semester engagement) who is interested in continuing work over multiple semesters and pursuing first-author publication. No prior astronomy research experience is required.