Research

Large-scale structure growth collage

Executive Summary

I work at the interface of cosmological theory, forward modeling, and data analysis. The main goal is to extract more information from large-scale structure data while keeping the inference physically controlled and statistically robust.

In practice, this means building models and analysis methods that can address a few central questions:

  • Is the late-time growth of cosmic structure fully consistent with the standard cosmological model?
  • How much information are we leaving on the table when we compress surveys into a small set of summary statistics?
  • Which observables beyond galaxy number counts can sharpen constraints on gravity, dark energy, and the initial conditions of the Universe?

At a Glance

  • Core research area: large-scale structure cosmology, with an emphasis on Bayesian forward modeling and field-level inference.
  • Theory toolkit: perturbation theory and the effective field theory of large-scale structure.
  • Data connections: galaxy clustering, galaxy shapes, cluster observables, and CMB cross-correlations.
  • Current theme: combine physically motivated models with information-optimal statistics to improve cosmological constraints and control systematics.

Priority Research Programs

1. Field-level Bayesian inference

This line of work asks whether we can infer cosmological parameters and primordial initial conditions directly from the observed large-scale structure field, rather than from a compressed set of low-order summary statistics.

Key results:

  • We developed a forward-modeling framework for field-level inference based on the effective field theory of large-scale structure and Bayesian inference.
  • We showed that field-level methods can recover unbiased cosmological parameters while substantially improving information extraction relative to standard analyses based on the power spectrum and bispectrum.
  • This program led to work published in JCAP and Physical Review Letters.

2. Growth of structure in the late Universe

Another major thread of my research is testing whether current CMB and large-scale structure data prefer a slower late-time growth history than the standard model predicts.

Key results:

  • Using current cosmological data, we found evidence for a suppressed growth rate of large-scale structure in the late Universe.
  • This result was published in Physical Review Letters and received broad coverage, including features in Scientific American and New Scientist.
  • More broadly, this work reflects my interest in identifying where tensions in cosmological data may point to new physics, improved modeling, or both.

3. New observables for cosmology

I am also interested in observables that go beyond standard galaxy number counts and can add information or break degeneracies in cosmological analyses.

This includes:

  • galaxy shapes and intrinsic alignments,
  • cluster gas and velocity-sensitive probes such as the kinematic Sunyaev-Zel’dovich effect,
  • galaxy sizes as complementary tracers of primordial physics.

The broader goal is to turn more aspects of the galaxy field into precision cosmological observables, not just positions and counts.

Selected Highlights

Field-level inference of large-scale structure

Our field-level inference program showed that one can jointly reconstruct initial conditions and infer cosmological parameters from the large-scale structure field itself. Relative to standard low-order summary-statistic analyses, the gain in constraining power can be substantial while preserving an interpretable physics-based model.

Relevant papers:

Suppressed growth of structure

By combining current CMB and large-scale structure data, we identified evidence that late-time structure growth may be lower than expected in the standard cosmological model. This is part of a wider effort to understand whether present-day data are revealing new physics or exposing limitations in the way we model late-time observables.

Relevant papers and coverage:

Galaxy sizes as tracers of local primordial non-Gaussianity

In recent work with Kazuyuki Akitsu and Atsushi Taruya, we study galaxy sizes as complementary tracers of local primordial non-Gaussianity. The main point is that size fluctuations can behave very differently from standard number-density fluctuations: for galaxy-mass halos, the response to late-time density can be close to zero while the response to local PNG remains sizable. That makes sizes an attractive companion observable in multi-tracer analyses.

For a DESI-like survey, the paper finds that combining galaxy numbers and sizes can improve local-PNG sensitivity by a factor of about 3.6, while the sign structure of the number-size cross spectrum can help diagnose systematics in the event of a detection.

Kinematic Sunyaev-Zel’dovich measurements

I developed a Bayesian framework for extracting the kSZ signal from CMB and cluster-velocity data while consistently propagating uncertainty in the reconstructed velocities. This work showed that velocity uncertainty is not a secondary detail: if ignored, it can materially bias the inferred signal.

Galaxy shapes and intrinsic alignments

I have also worked on measuring and modeling galaxy-shape correlations with the large-scale tidal field. In observational data, we detected the intrinsic-alignment signal at the field level and used it to probe how galaxy shapes trace the surrounding cosmic environment.