20 papers 4 threads last build 2026.11.12 Toronto 43.661N 79.397W
Tri Nguyen

[ NSERC CPRA – Dunlap Fellow ] · Dunlap Institute for Astronomy and Astrophysics · University of Toronto

Tri V. Nguyen

I am a near-field cosmologist working at the interface of astrophysics, statistics and machine learning. I am interested in the particle nature of dark matter and how it shapes structure formation on the smallest scales, using the dwarf galaxies and stellar streams of the Milky Way as my laboratory. Some of the questions I work on:

  • Are the centers of dwarf galaxies cusped or cored, and what does that tell us about dark matter?
  • How many dark subhalos orbit the Milky Way, and can stellar streams detect them?
  • What are the shape, mass profile, and assembly history of the Milky Way’s dark matter halo?

On the statistics and machine learning side, I am interested in developing novel and robust inference techniques, and I work on making them reliable, e.g. calibrated posteriors, robustness to the mismatch between simulations and data, and principled model comparison.

Before Toronto I was a CIERA Postdoctoral Fellow at Northwestern University and a postdoctoral affiliate of the NSF–Simons SkAI Institute. I completed my Ph.D. in Physics at MIT in 2024 with Prof. Lina Necib, where I was also a member of the NSF AI Institute for Artificial Intelligence and Fundamental Interactions (IAIFI). I received my B.S. in Physics & Astronomy from the University of Rochester in 2019.

tri.nguyen [at] astro.utoronto.ca GitHub Google Scholar ADS ORCID CV

Field: Draco dSph · 1,335 member stars · Gaia DR3 · ESA/Gaia/DPAC · dashed: half-light radius

// News

Sep 08, 2026 Open SkAI 2026, the second annual SkAI astronomy and AI conference, took place in Chicago, September 8–11.
Sep 01, 2026 Started as an NSERC CPRA–Dunlap Fellow at the University of Toronto! 🇨🇦🔭
Jun 24, 2026 New paper out: Dark Matter in Draco and Boötes I: Hints of a Core in an Ultra-Faint Dwarf from Simulation-Based Inference arXiv:2606.26218
Dec 08, 2025 New paper out: Forecasting Dark Matter Subhalo Constraints from Stellar Streams using Implicit Likelihood Inference arXiv:2512.07960
Sep 15, 2025 The inaugural Open SkAI conference, the NSF–Simons SkAI Institute’s annual astronomy and AI meeting, took place in Chicago.

// Research

The Large Magellanic Cloud

Dwarf galaxies

Dwarf galaxies are the most dark-matter-dominated systems in the Universe, and how their dark matter is distributed is a direct test of dark matter physics. I am interested in developing novel, robust inference techniques that extract the dark matter content of a dwarf from the kinematics of a few dozen stars. I am also interested in tackling the systematics of spectroscopic surveys, from selection functions to membership selection, and actively work with observers to fold them into the inference and to apply it to new observations of the Milky Way's dwarfs.

Draco & Boötes ITrial by FIREGraphNPE
Artist's impression of the Milky Way surrounded by stellar streams, James Josephides and the S5 Collaboration

Stellar streams

Stellar streams are the tidal remnants of dwarf galaxies and globular clusters. As stellar streams orbit the Milky Way, they are perturbed by dark matter subhalos, which create visible features such as gaps and spurs in the stream that can be traced back to the subhalo. I am actively working on several projects that use stellar kinematics from large photometric and spectroscopic surveys to hunt for subhalos, using machine learning and simulation-based inference. I am especially interested in the model-selection side of the problem: whether a stream's features can be explained without a subhalo at all, and how to answer that question with a statistical framework that is robust to the choices made along the way.

Stream forecasts
A dark matter halo merger tree

Halo assembly

How dark matter halos grow through mergers shapes the properties of the galaxies they host. Fast, accurate models of assembly histories and full merger trees are challenging to build. Analytic methods miss the dependence on environment and internal structure, and N-body simulations are expensive. With the Galaxy Formation group at the CCA, I developed generative models that produce assembly histories and merger trees with the statistical fidelity of simulations at a fraction of the cost, as a backbone for semi-analytic galaxy models.

FLORAHFLORAH-Tree
Satellites around a host halo

Galaxy-halo connection

The connection between galaxies and their dark matter halos is central to cosmology and to testing dark matter models. Hydrodynamic simulations are among the leading methods to study the galaxy-halo connection, but their computational cost limits their ability to fully explore the dark matter and baryonic parameter space. I authored NeHOD, a variational diffusion model that paints galaxies and subhalos onto dark matter halos, which the DREAMS collaboration uses to turn a thousand simulations into predictions across the whole parameter space.

NeHOD

// Selected papers

Full list →
  1. Dark Matter in Draco and Boötes I: Hints of a Core in an Ultra-Faint Dwarf from Simulation-Based Inference
    Tri Nguyen, Lina Necib, Ting S. Li, and 8 more authors
    arXiv e-prints, Jun 2026
  1. Forecasting Dark Matter Subhalo Constraints from Stellar Streams using Implicit Likelihood Inference
    Tri Nguyen, Rutong Pei, Zhuofu Li, and 9 more authors
    arXiv e-prints, Dec 2025
  1. Trial by FIRE: probing the dark matter density profile of dwarf galaxies with GraphNPE
    Tri Nguyen, Justin Read, Lina Necib, and 4 more authors
    Monthly Notices of the Royal Astronomical Society, Aug 2025
  1. Emulating dark matter halo merger trees with graph generative models
    Tri Nguyen, Chirag Modi, Siddharth Mishra-Sharma, and 2 more authors
    Monthly Notices of the Royal Astronomical Society, Oct 2025
  1. How DREAMS are made: Emulating Satellite Galaxy and Subhalo Populations with Diffusion Models and Point Clouds
    Tri Nguyen, Francisco Villaescusa-Navarro, Siddharth Mishra-Sharma, and 16 more authors
    The Astrophysical Journal, Jan 2026

// Mentoring

Students I have worked closely with, at their level when we worked together.

Current

Andy Yu · undergraduate, University of Toronto SUDS, with Ting Li · 2026 –
Ella Marin · graduate, University of Washington with Nora Shipp · 2025 –

Former

Andreas Filipp · graduate, Université de Montréal with Yashar Hezaveh and Laurence Perreault-Levasseur · 2025 – 2026
Rutong Pei · undergraduate, Carnegie Mellon University with Nora Shipp and Scott Dodelson · 2023 – 2024
Hanna Chen · undergraduate, MIT UROP, with Lina Necib · Summer 2023
Anna Orgel · undergraduate, MIT UROP, with Lina Necib · Summer 2023
Hang Su · undergraduate, MIT MSRP, with Lina Necib · Summer 2022
Michael Huang · high school, Phillips Exeter Academy RSI at MIT, with Lina Necib · Summer 2022

Prospective students. I only take on students through official University of Toronto programs such as SUDS and SURP. If you are interested in working with me on dark matter, dwarf galaxies, stellar streams, or machine learning for astrophysics, please contact me via email.