Menu Close
 

SkAI Institute

SkAI Institute horizontal logo


The SkAI Institute is one of the National Artificial Intelligence Research Institutes funded by the National Science Foundation (NSF) and Simons Foundation, which will accelerate Astro-AI research and help educate a diverse Astro-AI workforce. CAPS is one of the lead partners of the SkAI Institute.

CAPS Work on the SkAI Institute Project:

CAPS Personnel Holding Leadership Roles on the SkAI Institute Project:
Armgard Haken: Project Development and Data Manager on the SkAI Operations Team
Gautham Narayan: Co-Principal Investigator; Deputy Director for Research (Astro)
Matthew Turk: Co-Chair of the Community Software, Data, & Hardware Resources Committee
Joaquin Vieira: Co-Chair of SkAI Institute Pillar 3 (Learning-Based Astrophysical Survey and Instrument Design)
Nicolas Yunes: Co-Chair of the Public Outreach & High School Program Committee

Current CAPS Personnel on the SkAI Institute Project:

Jay Alameda (CAPS Senior Technical Program Manager)
Fabián Araneda-Baltierra (CAPS Visiting Research Scientist)
Matias Carrasco Kind (CAPS affiliated faculty)
Siegfried Eggl (CAPS affiliated faculty)
Bill Gropp (NCSA Director)
Robert Gruendl (CAPS Senior Research Scientist)
Armgard Haken (CAPS Research Coordinator)
Matthew Krafczyk (CAPS Research Scientist)
Jennifer Li (CAPS Research Scientist)
Felipe Menanteau (CAPS Senior Research Scientist)
Billy Moses (CAPS affiliated faculty)
Gautham Narayan (CAPS Deputy Director)
Matthew Turk (CAPS affiliated faculty)
Joaquin Vieira (CAPS Director)
Amanda Wasserman (CAPS Graduate Student Fellow)
Nicolas Yunes (CAPS affiliated faculty)

SkAI-Funded Projects Involving CAPS Personnel:

Title: A Foundation AI Model to Infer the Physics of Transients
Co-Lead PI: Gautham Narayan
Co-PI: Matt Krafczyk
Collaborator: Amanda Wasserman
Project Summary: This project aims to build a foundational AI model that would form a core part of a community pipeline for Rubin to identify (including anomalies) and characterize transients and set priorities for follow-up observations. The foundation model will be based on a new custom variational autoencoder (VAE) architecture, optimized for the sparse, multimodal and irregular data, and both observational data and emulated data from (physical) transient models. In year 2, the VAE will be replaced by a transformer.

Title: Automating Bayesian Inference of Millimeter Source Association
Lead PIs: Joaquin Vieira, Billy Moses
Collaborators: Jennifer Li, Felipe Menanteau, Melanie Archipley (former CAPS Graduate Student Fellow), Kedar Phadke (former CAPS Graduate Student Fellow)
Project Summary: This project aims to develop an automated Bayesian inference tool for millimeter source association [cross-matching South Pole Telescope (SPT) sources with other surveys]. Correct associations to observations in different wavelengths are necessary to reconstruct spectral energy distributions (SEDs) and infer properties and redshifts of astrophysical objects. Bayesian inference is one of the methods to do this, but due to large numbers of objects that will be available from the current and future surveys, faster methods are necessary. This project aims to use machine learning (ML) to automate Bayesian inference, focusing on advancements in differentiable and probabilistic programming. Unlike existing differential programming problems, this project will incorporate the use of discrete variables, which are not handled well by existing automatic differentiation (AD) frameworks. The team plans to build a Zooniverse project to confirm source associations via visual inspection by volunteers.

Title: AI Accelerated Simulations with Multi-Scale Astrophysics using Physics-Based Deep Learning
Co-PI: Billy Moses
Collaborator: Matt Krafczyk
Project Summary: This project has the following objectives: (1) model stellar structure and evolution, (2) combine learned components (trained to emulate simulation) with numeric methods, (3) order-of-magnitude speedups, (4) enable modeling large populations of (single and binary) stars, and (5) incorporate uncertainty quantification. These objectives will be met using (1) physics-informed neural networks (PINNs) and physics-informed neural operators (PINOs), (2) hybrid combination of physics-based deep learning (PBDL) with classical methods, and (3) active learning/mesh refinement.

Title: A Universal Forecaster for Astronomical Light Curves, and Other Out-of-domain Time-Series Data
Co-PI: Gautham Narayan
Project Summary: This project seeks to predict nonuniformly sampled time-series data using artificial intelligence with uncertainty quantification. The main goal is astronomical time-series data, mainly from stars within the Zwicky Transient Facility (ZTF) survey, but ultimately applicable to Rubin/Legacy Survey of Space and Time (LSST) and even other domains.