Research and Collaborations
Primary Research Directions
Dark Energy
Multiple independent lines of evidence have revealed that the expansion of the universe is accelerating. Observations indicate that roughly two-thirds of all energy in the universe is a mysterious component called "dark energy," which drives this accelerated expansion, influences the growth of structure, and determines the ultimate fate of our universe. However, its fundamental nature remains a mystery. Our group investigates dark energy by modeling and measuring its effects across time, especially on the large-scale distribution of galaxies and the growth of cosmic structure. Because the effects of dark energy are intertwined with other cosmological parameters, detailed simulations are essential for isolating its signatures in observational data. We then work with large cosmic surveys to directly measure these effects: galaxy clustering, weak gravitational lensing, and the abundance of the most massive clusters in our universe are highly sensitive to this physics.
Dark Matter
Dark matter makes up most of the matter in the universe. It plays a foundational role in the formation and growth of galaxies and has a strong influence on how the universe expands and forms structure on cosmic scales. However, the detailed physics of dark matter is still a mystery. The properties of dark matter influence how cosmic structure forms—including the nature of small dark matter structures and the faint galaxies they host. Our group searches for these signatures to try to test different dark matter models. Our approaches range from vast cosmological scales to the smallest accessible scales. At large scales, we study the statistical impact of dark matter physics on massive galaxy surveys like DESI, Rubin, and Roman. At small scales, we use techniques such as searching for gravitational lenses made by invisible dark matter structures, studying the tiniest galaxies in the universe, or looking for features in disrupting star clusters within the Milky Way.
Galaxy Formation
Galaxies are some of the most foundational and interesting systems in our Universe—their formation and evolution bring together myriad astrophysical processes, from star formation to the dynamics of dark matter. Even though we have observed and studied countless galaxies across cosmic time, there are open questions about how they form and about the interplay between and impact of these astrophysical processes. Our group’s focus spans from detailed studies of individual objects to analysis of large surveys of galaxies where internal galaxy physics is a chief systematic uncertainty. The wide range of expertise in our group allows us to approach these problems from both observational and theoretical perspectives. Our group pursues analyses that require collecting and understanding new data from a range of ground- and space-based instruments. We also push the limits of galaxy formation modeling, from detailed simulations to large data-driven approaches.
Milky Way
The Milky Way galaxy is our home, and the best-measured galaxy in the universe. These measurements let us extract more information about the physics and history of the Milky Way than is possible for any other galaxy, enabling unique tests of fundamental physics. Our group leverages observational and numerical approaches to understanding the Milky Way and the dark matter halo that surrounds it. We participate in large scientific surveys such as DES, DESI, Rubin/LSST, and Via that are mapping out the Milky Way, its satellite galaxies, and the structure of the stellar halo that surrounds it. These imaging and spectroscopic surveys can be used together to inform the nature of dark matter and the Milky Way’s formation history. We are actively developing tailored cosmological simulations and models that are needed to match the detailed picture provided by current and future surveys of the Milky Way, and new tools for joint inference between models and data.
Simulations
Simulations are a primary technique our group uses to answer complex questions about galaxies, dark matter, and cosmology, on scales ranging from star clusters to universes. Many astrophysical objects are so complex that simulations are the only route to reliable theoretical predictions. We leverage a broad spectrum of modeling techniques, from data-driven approaches designed to produce mock universes for large surveys to detailed hydrodynamical simulations designed to test models of galaxy formation, spanning scales from individual galaxies to the scale of the entire universe. We also develop field-leading simulation analysis tools. These tools range from clustering algorithms to post-processing physics models to emulators that can quickly generate results to match full simulations at high accuracy. We have a strong tradition in developing techniques that model and constrain the connection between dark matter halos and the galaxies within them.
AI and machine learning
AI and machine learning are transforming astrophysics and cosmology—enabling discovery in large datasets, uncovering hidden patterns, and accelerating modeling of diverse cosmic phenomena. We use computer vision techniques to identify faint structures such as dwarf galaxies and stellar streams that are difficult to detect with traditional methods, and apply clustering and deep learning to classify subtle features across survey data. To accelerate modeling, we build emulators that compress months of computation into minutes, enabling rapid exploration of theoretical models and their connection to survey data. We use techniques such as simulation-based inference to perform rigorous statistical comparisons between simulations and observations, extracting robust constraints on physical parameters. In collaboration with the Center for Decoding the Universe, we are exploring new uses of AI to accelerate our work and advance discovery across our science.
Survey Collaborations
Via
The Via Project—a collaboration between Stanford, Harvard, Yale, and Carnegie—is building twin high-resolution fiber-fed spectrographs for the 6.5-meter MMT and Magellan telescopes. Via will measure stellar velocities to ~100 m/s across the sky, targeting two powerful probes of dark matter physics. The first is the population of faint low-mass galaxies, whose internal motions trace the masses of their host dark matter halos. The second is stellar streams from disrupted globular clusters, which can reveal the gravitational signatures of dark halos too small to host any galaxy. The abundance of these tiny halos is a critical test of dark matter models: the standard cold dark matter model predicts they are plentiful, while alternative models predict far fewer. Our group is developing cosmologically grounded models of Milky Way-like hosts, their satellites, and streams to interpret Via data and place strong constraints on dark matter physics and the formation of the lowest-mass galaxies. We are also developing ideas for Via's Boombox instrument, which will provide fast, responsive transient followup.
DESI
The Dark Energy Spectroscopic Instrument (DESI) is a multi-object spectrograph that has operated since 2021 on the Mayall Telescope at Kitt Peak, Arizona. With the ability to take spectra for 5,000 objects simultaneously, DESI has already observed tens of millions of galaxies, stars, and quasars—more than ten times the total from all previous spectroscopic surveys. This three-dimensional map of cosmic structure, spanning nearly half the sky, enables precise measurements of the expansion history of the universe and stringent tests of dark energy models. Our group has been deeply involved in DESI from its inception, including founding leadership of the collaboration. Today, group members use DESI data to learn about the formation of structure and to study the connection between galaxies and dark matter on cosmological scales. We are also pursuing what DESI reveals about the local universe, including studies of vast numbers of small galaxies and of stars within the Milky Way. These data allow us to probe the nature of dark matter, the formation of low-mass galaxies, and the structure and formation history of our own galaxy.
SAGA
The Milky Way's satellite galaxies offer detailed insight into low-mass galaxy formation and the nature of dark matter, but placing this knowledge in context requires understanding how our galaxy compares to similar systems. Our group played a leading role in the SAGA Survey, which studied 101 Milky Way analogs to characterize their satellite populations. We identified 378 satellites around these hosts, measured their properties, and compared them to predictions from simulated Milky Way analogs—providing key insight into how the Milky Way and its satellites differ from other systems. SAGA also obtained spectra for many background galaxies, and we continue to analyze this rich dataset. Building on SAGA, we designed a DESI secondary target program that has collected spectra for a large sample of low-redshift galaxies across the sky, dramatically expanding our ability to study the nearby galaxy population. These datasets provide crucial context for interpreting Milky Way observations and testing models of satellite galaxy formation.
DES
The Dark Energy Survey (DES) mapped 300 million galaxies over one-eighth of the sky using the Blanco Telescope in Chile. DES measured the properties of dark energy through galaxy clustering, weak gravitational lensing, galaxy cluster abundances, and supernovae. Our group played leading roles in these measurements and their interpretation. We ran large suites of N-body simulations used to compare theory against observation, developed galaxy models to turn these simulations into mock measurements, and pioneered techniques for extracting cosmological information—including methods for combining cluster, lensing, and clustering probes. Beyond large-scale cosmology, our group led efforts to use DES imaging of resolved stars to discover faint dwarf galaxies orbiting the Milky Way, contributing to the discovery and interpretation of dozens of new satellites. These tiny objects are powerful probes of dark matter physics, and we continue to use them to constrain dark matter's mass and behavior.
Rubin
The Vera C. Rubin Observatory and its flagship Legacy Survey of Space and Time (LSST) will capture the largest-ever image of the sky every few nights for ten years—observing 20 billion galaxies and 20 billion stars and driving discoveries across cosmology, time-domain astrophysics, and Milky Way science. Our group is working to maximize the scientific return of this transformative dataset. For cosmology, we are building synthetic sky surveys to test analysis pipelines, developing methods for joint-survey cosmology with Rubin and other experiments, and creating tools to accelerate reproducible inference. For dark matter, we are developing techniques to detect ultra-faint dwarf galaxies, identify strong gravitational lenses produced by dark substructure, and extract dark matter constraints from small-scale structure. For the Milky Way, we are building models of the stellar halo and its substructure to interpret LSST's unprecedented stellar maps. We are most excited about the new discoveries that LSST will make, and we are designing new techniques for automated discovery in the Rubin data.
Roman
The Nancy Grace Roman Space Telescope is NASA's next flagship observatory, designed to study dark energy, galaxy evolution, and exoplanets. Roman's wide-field infrared instrument captures 100 times more sky per image than Hubble at comparable sensitivity, enabling a sweeping survey that will measure light from over a billion galaxies. These observations will allow us to reconstruct the history of structure formation under dark matter's influence and constrain dark energy models with unprecedented precision. Our group has contributed significantly to Roman's science definition and survey planning, shaping the High Latitude Wide Area Survey for precision cosmology and studies of stellar streams. We are providing essential infrastructure for the mission, including cosmological simulations that model Roman's galaxy redshift survey and models of the Milky Way and local universe relevant to Roman's wide-field imaging. We are excited about the broad science reach Roman will enable, from the structure of the Milky Way and resolved dwarf galaxies in the Local Volume to precision cosmology at the largest scales.