Magdalena Siwek
Research Scientist | Simons Junior Fellow | NYU & Flatiron CCA
About
I build statistical and machine learning surrogate models for data from expensive simulations, and I am especially interested in quantifying the uncertainty in their predictions. Most of my work involves probabilistic emulators that generate realistic synthetic data, infer physical parameters and forecast time series. I closely work with transformers, writing architectures in jax to suit the model requirements. A current example is fates, an attention-based forecasting model for time-series data. I release my tools as open-source Python packages, such as calypso on PyPI.
I completed my PhD in astrophysics at Harvard in 2024, and have since been a Junior Fellow of the Simons Society of Fellows at New York University and the Flatiron Institute. I am a member of the NANOGrav collaboration and co-develop holodeck, the massive black hole binary population synthesis code used to model the gravitational wave background.
Selected Publications
- calypso: a Parameter-Conditioned Stochastic Surrogate Model for Circumbinary Accretion Time-Series
- Signatures of Circumbinary Disk Dynamics in Multi-Messenger Population Studies of Massive Black Hole Binaries
- Orbital Evolution of Binaries in Circumbinary Disks
- Preferential Accretion and Circumbinary Disk Precession in Eccentric Binary Systems
- The effect of differential accretion on the Gravitational Wave Background and the present day MBH Binary population
- Optical and X-ray luminosity of expanding nebulae around ultraluminous X-ray sources
Full list, including NANOGrav collaboration papers, on ADS.