KAAI Workshop

AI Methods for Turbulence
and Early-Universe Signals

A KAAI Workshop · Carnegie Mellon University

November 2 – 6  ·  McWilliams Center for Cosmology & Astrophysics, CMU

Overview

About the workshop

Turbulence appears across astrophysics and cosmology, from magnetized plasmas in galaxies and the interstellar medium to the primordial plasma of the early universe. Early-universe processes such as first-order phase transitions and primordial magnetic fields can drive turbulence, which may leave observable imprints in stochastic gravitational-wave backgrounds and in cosmic magnetic fields. Modeling these systems requires expensive simulations over a wide range of scales, and connecting them to data is a hard inverse problem. This workshop brings together researchers in turbulence, early-universe physics, and machine learning to discuss how AI methods can help with simulation, inference, and signal detection in this area.

  • AI methods for turbulence: surrogate models, emulators, and learned representations of hydrodynamic and magnetohydrodynamic turbulence across scales.
  • Early-universe signals: modeling and inference of signatures of early-universe physics, including gravitational-wave backgrounds and primordial magnetic fields.

This is a Keystone Astronomy & AI (KAAI) Fellows Program workshop, sponsored by the Simons Foundation Targeted Grant to Institutes.

Registration

Registration & online participation

Registration is closed. All sessions can be joined online via Zoom.

Join on Zoom

Meeting ID: 979 8889 7636

QR code for the workshop Slido

Q&A on Slido

Scan the code or open Slido to ask questions and vote on them during the talks.

Questions? Email mgurgeni@andrew.cmu.edu.

Location

Location & Directions

All workshop meetings will be held in Wean Hall 8330, Carnegie Mellon University.

Wean Hall
Carnegie Mellon University
5000 Forbes Avenue
Pittsburgh, PA 15213

Directions to Room 8330

  1. Enter Wean Hall through the main entrance on the 5th floor (view entrance location on Google Maps).
  2. Take the main elevators or central stairs up to the 8th floor.
  3. Follow the hallway signs to Room 8330 (Department of Physics / McWilliams Center).

Schedule

Schedule

Tentative schedule, subject to change. All times Eastern (EST).

Monday, November 2

9:30 – 10:00 amWelcome and introductions
10:00 – 10:45 amTalkTurbulence production of gravitational waves in the early UniverseAlberto Roper Pol
10:45 – 11:15 amCoffee break
11:15 am – 12:00 pmTalkEnergy Dissipation in Gravitational Wave SourcesArthur Kosowsky
12:00 – 1:00 pmLunch
1:00 – 1:45 pmTalkWhat Can We Learn from Pulsar Timing Arrays When AI Is on Board?Tina Kahniashvili
1:45 – 2:30 pmIntroduction of the proposed projects for turbulence
2:30 – 3:15 pmOpen discussion about projects
3:15 – 4:00 pmTalkSolving PDEs with ConstraintsMurman Gurgenidze
4:00 – 4:30 pmCoffee break
4:30 – 6:00 pmFree time to discuss / work on collaborative projects

Tuesday, November 3

9:30 – 10:15 amTalkCosmoAgent: AI Agent for Theoretical CosmologyWenrong Sun
10:15 – 10:45 amCrash course on numerical simulations / hands-on Pencil Code
10:45 – 11:15 amCoffee break
11:15 am – 12:00 pmCrash course on numerical simulations / hands-on Pencil Code
12:00 – 1:00 pmLunch
1:00 – 1:45 pmCrash course on numerical simulations / hands-on Pencil Code
1:45 – 4:00 pmProject proposal with Pencil Code / description of hackathon project
4:00 – 4:30 pmCoffee break
4:30 – 5:15 pmEthical concerns about use of AI
5:15 – 6:00 pmFree time to discuss / work on collaborative projects

Wednesday, November 4

9:30 – 10:00 amTalkIn-the-loop ML for LES and other applicationsTouko Puro / Shreyas Giridhar (online)
10:00 – 10:45 amTalkSimulations of first-order phase transitionsIsak Stomberg (online)
10:45 – 11:15 amCoffee break
11:15 am – 12:00 pmTalkGW detection prospects via astrometryDeyan Mihaylov
12:00 – 1:00 pmLunch
1:00 – 1:45 pmTalkPractical Scientific Constraints with Simulation-Based InferenceMatthew Ho
1:45 – 4:00 pmFree time to discuss / work on collaborative projects / hackathon projects / spontaneous talks
4:00 – 4:30 pmCoffee break
4:30 – 6:00 pmFree time to discuss / work on collaborative projects

Thursday, November 5

9:30 – 10:00 amTalkTesting Massive Gravity with Pulsar Timing ArraysChris Choi
10:00 – 10:30 amTalkIntegral Invariant Discovery via Bayesian Experimental DesignJames Wedgwood
10:45 – 11:15 amCoffee break
11:15 am – 12:00 pmTalkCoarse-Graining and the Closure Problem in Plasma TurbulenceMuni Zhou
12:00 – 1:00 pmLunch
1:00 – 4:00 pmFree time to discuss / work on collaborative projects / hackathon projects / spontaneous talks
4:00 – 4:30 pmCoffee break
4:30 – 5:15 pmOpen discussion about the future of science with AI
5:15 – 6:00 pmFree time to discuss / work on collaborative projects

Friday, November 6

9:30 – 10:45 amHackathon
10:45 – 11:15 amCoffee break
11:15 am – 12:00 pmHackathon
12:00 – 1:00 pmLunch
1:00 – 4:00 pmHackathon
4:00 – 4:30 pmCoffee break
4:30 – 6:00 pmClosing / hackathon winners

Slides

Slides

Slides are posted here as speakers share them.

  • Solving PDEs with Constraints

    Murman Gurgenidze · Monday, November 2

    PDF

Project descriptions

Descriptions of the proposed collaborative projects.

  • Solving PDEs with Constraints

    Project description · 5 pages

    PDF

  • Large-Scale Simulations and Subgrid Modeling of MHD

    Project description · 2 pages

    PDF

  • Gravitational Wave Production Using Simulation-Based Inference

    Project description · 2 pages

    PDF

  • Pencil Code Template for the Workshop

    Hands-on session: gravitational waves from MHD turbulence · PDF, 2 pages

    PDF

  • Source Files for the Simulations

    Datasets for “Numerical Simulations of Gravitational Waves from Early-Universe Turbulence” · Zenodo

    Link

Speakers

Invited speakers

Arthur Kosowsky

Arthur Kosowsky

Professor · University of Pittsburgh

Isak Stomberg

Isak Stomberg

Postdoctoral Researcher · IFIC (CSIC–Universitat de València)

Deyan Mihaylov

Deyan Mihaylov

NASA ADP Postdoctoral Fellow · Case Western Reserve University

Matthew Ho

Matthew Ho

Member · Institute for Advanced Study

Muni Zhou

Muni Zhou

Assistant Professor · Dartmouth College

Wenrong Sun

Wenrong Sun

Brown University · Brown Center for Theoretical Physics and Innovation

Committee

Organizing committee

Tina Kahniashvili

Carnegie Mellon University, Department of Physics

Barnabás Póczos

Carnegie Mellon University, Machine Learning Department

Alberto Roper Pol

University of Geneva, Department of Theoretical Physics

Murman Gurgenidze

Carnegie Mellon University, Department of Physics