Research

A simulated Milky Way-mass system: dark matter (blue) and the stars that formed in it (orange)
Credit: Zewei Wu

Galaxy formation

How did the smallest galaxies form?

  • What is the halo mass threshold for galaxy formation?
  • When and how does gas cool to form stars in ultra-faint dwarf galaxies?
  • How does our local reionization history impact dwarf galaxy evolution?
Same large-scale structure 10−22 eV keV GeV 100 TeV MPl 10 M☉ UltralightLightWIMPsCompositePBHs Different small-scale structure FuzzyWarmSelf-interactingCold
Adapted from T. Lin (2019)

Dark matter

What is the nature of dark matter?

  • Do completely dark (galaxy-free) halos exist?
  • What are the astrophysical signatures of dark matter physics beyond gravity?
  • Can we combine small-scale structure observables to detect a deviation from CDM?
Telescopes at Cerro Tololo under the Magellanic Clouds, with illustrated dark matter halos around the Milky Way's satellites
Credit: Anja von der Linden

Near-field cosmology

How can observations of the local universe inform cosmology?

  • Are there features in the small-scale linear power spectrum?
  • How can we efficiently and accurately model structure in the local universe?
  • What can near-field cosmology reveal about early universe dynamics and inflation?

Collaborations and funding

Targeted Simons Research Group on Dark Matter Dynamics

Simons Foundation · PI (UC San Diego node) · from January 2027

Connecting particle physics models of dark matter with new forces to predictions for galaxies, using simulations and semi-analytic models. UC San Diego is one of the group's nodes.

Collaborative Research: Reconstructing Primordial Density Fluctuations using Near-Field Cosmology

National Science Foundation · PI (NSF AAG) · 2024–

Reconstructing the primordial power spectrum with near-field cosmology by combining simulations, including COZMIC, and semi-analytic models with dwarf galaxy data. A collaboration between UC San Diego, USC, and Carnegie Observatories.

Simulations

Cosmological zoom-in simulations of small-scale structure. Click a suite for details and data.

Interactive compilations of self-interacting and beyond-CDM simulations from the literature are on the Visualizations page.

Dark matter constraints and predictions

Our simulations and models make predictions for small-scale structure; observations of dwarf galaxies and strong gravitational lenses turn them into constraints.

Predictions

Observational Constraints

Dwarf galaxies

Milky Way satellites from bright to faint: the Large Magellanic Cloud, the classical dwarf Fornax, and the ultra-faint dwarf Segue 1, with its spectroscopic member stars (Simon et al. 2011) circled
Credit: Zdeněk Bardon/ESO; ESO/Digitized Sky Survey 2; Legacy Surveys / D. Lang (Perimeter Institute) (CC BY 4.0)

In progress: hydrodynamic simulations of ultra-faint galaxies orbiting Milky Way analogs, to measure the galaxy formation threshold with Rubin.

Symphony

2023 · Stanford / KIPAC

Symphony is a compilation of 262 cosmological dark matter-only zoom-in simulations spanning four decades of host halo mass, including hosts similar to the LMC, the Milky Way, group-mass systems (e.g., strong gravitational lenses), low-mass galaxy clusters, and massive galaxy clusters.

Symphony hosts at five mass scales (rows), ordered by concentration (columns). Adapted from Nadler et al. 2023.

Symphony hosts at five mass scales (rows), ordered by concentration (columns). Adapted from Nadler et al. 2023.

Milky Way-est

2024 · Led by Deveshi Buch (Stanford / KIPAC)

Milky Way-est is a suite of 20 cosmological dark matter-only zoom-in simulations of systems that specifically resemble the Milky Way, including analogs of the LMC and Gaia–Sausage–Enceladus. It provides a realistic template for modeling our galaxy's substructure in CDM, including its satellite galaxy and stellar stream populations.

Milky Way-est host 004Milky Way-est host 113Milky Way-est host 222Milky Way-est host 659

Dark matter in four Milky Way-est simulations, each centered on a Milky Way-mass host; LMC analogs are visible near the center of each system. Adapted from Buch & Nadler et al. 2024.

COZMIC

2024–25 · COZMIC Collaboration: UC San Diego, USC, and Carnegie Observatories

COZMIC (COsmological ZooM-in simulations with Initial Conditions beyond CDM) includes hundreds of Milky Way simulations with initial conditions for warm, fuzzy, interacting, and mixed dark matter and for enhanced small-scale power, as well as simulations with dark matter self-interactions. COZMIC accurately captures how dark matter physics impacts Milky Way substructure.

The same Milky Way host forming in CDM (left) and with dark matter–proton scattering (right).

SIDM Concerto

2025 · With UC Riverside (D. Kong, D. Yang, H.-B. Yu)

SIDM Concerto is a compilation of cosmological zoom-in simulations of strong, velocity-dependent self-interacting dark matter (SIDM) spanning seven decades of subhalo mass, from LMC-mass hosts to galaxy clusters. These simulations provide a benchmark for testing SIDM predictions with dwarf galaxies, strong lenses, and stellar streams.

SIDM Concerto simulations of an LMC-mass, Milky Way-mass, group-mass, and low-mass-cluster host. Adapted from Nadler et al. 2025.

SIDM Concerto simulations of an LMC-mass, Milky Way-mass, group-mass, and low-mass-cluster host. Adapted from Nadler et al. 2025.

Galaxy formation threshold

Predictions for the mass spectrum of dark matter halos that form stars. By including molecular hydrogen cooling and relative streaming motion between dark matter and baryons, this model predicts a galaxy formation threshold significantly smaller than previous predictions based only on atomic hydrogen cooling and reionization.

The fraction of dark matter halos that form galaxies (left) and a sample of halo growth histories (right). Blue lines include molecular hydrogen cooling and dark matter–baryon streaming; orange lines only include atomic hydrogen cooling and reionization. Halos that pass above the thick lines in the right panel form stars. Adapted from Nadler 2025.

The fraction of dark matter halos that form galaxies (left) and a sample of halo growth histories (right). Blue lines include molecular hydrogen cooling and dark matter–baryon streaming; orange lines only include atomic hydrogen cooling and reionization. Halos that pass above the thick lines in the right panel form stars. Adapted from Nadler 2025.

Faint-end galaxy–halo connection

An empirical model and inference framework for the galaxy–halo connection that extends to the faintest dwarf galaxies. This framework includes a machine-learning model for satellite disruption due to central galaxies, an empirical model for the subhalo–satellite galaxy connection, and forecasts for future surveys.

Nadler et al. (2020) applied this framework to Milky Way satellites observed by the Dark Energy Survey and Pan-STARRS1 to reveal the impact of LMC satellites and constrain the faint-end galaxy–halo connection.

The inference framework of Nadler et al. (2020), from a notebook sketch to its Figure 1.

The fraction of low-mass halos that host galaxies (left) and the corresponding stellar mass–halo mass relation (right), derived by combining a galaxy–halo connection model with observations of the Milky Way satellite population. Blue regions are allowed by the data. Adapted from Nadler et al. 2020.

The fraction of low-mass halos that host galaxies (left) and the corresponding stellar mass–halo mass relation (right), derived by combining a galaxy–halo connection model with observations of the Milky Way satellite population. Blue regions are allowed by the data. Adapted from Nadler et al. 2020.

Constrained merger trees

An extension of the open-source structure and galaxy formation model Galacticus that efficiently and accurately generates constrained realizations of the Milky Way's subhalo and satellite galaxy population. Nadler et al. (2023) introduced the constrained merger tree technique to model high-redshift JWST galaxies; it produces Milky Way-like accretion histories about a thousand times faster than existing methods. The Milky Way model is being calibrated on COZMIC simulations.

Example of a Milky Way system's merger tree (left) and dark matter substructure (right) generated using Galacticus. The constrained merger tree technique guarantees that the Milky Way host (black) accretes an LMC analog (red).

Example of a Milky Way system's merger tree (left) and dark matter substructure (right) generated using Galacticus. The constrained merger tree technique guarantees that the Milky Way host (black) accretes an LMC analog (red).

Small-scale structure limits

Nadler et al. (2021a) used the population of Milky Way satellites observed by the Dark Energy Survey and Pan-STARRS1 to constrain dark matter's production mechanism, particle mass, and Standard Model interactions. Nadler et al. (2021b) combined these constraints with strong lensing flux ratio statistics to set the most stringent warm dark matter limit at the time.

Constraints on sterile neutrino warm dark matter (left) and dark matter–Standard Model interactions (right) derived using the population of ultra-faint dwarf galaxies near the Milky Way. Red and blue regions are excluded. Adapted from Nadler et al. 2021a.

Constraints on sterile neutrino warm dark matter (left) and dark matter–Standard Model interactions (right) derived using the population of ultra-faint dwarf galaxies near the Milky Way. Red and blue regions are excluded. Adapted from Nadler et al. 2021a.

Self-interacting dark matter predictions

Nadler, Yang, & Yu (2023) ran the first cosmological simulation with strong, velocity-dependent dark matter self-interactions in a strong lens analog. The predicted halo and subhalo populations explain the extremely dense substructure in the strong lens SDSS J0946+1006 and the rotation curves of isolated ultra-diffuse galaxies.