K-State Ph.D. student Hanyu Zhang showed that we can learn a lot about the way luminous red galaxies populate dark matter halos if we analyze their three-point functions.

The constraints on HOD parameters for the DESI like luminous red galaxies can be improved by almost a factor of two (left panel below, Figure 9 in the paper)! But you have to be careful with the binning effects that may turn out to be much more important than one naively expects (right panel below, Figure 11 in the paper).

We often use galaxy correlation functions as standard rulers to measure distances. It is not always clear where this information is encoded and using simple Fisher Information formulas may result in wrong forecasts.

In this controversial paper we argue that because of the tricky nature of the standard ruler tests, the information content of the higher order correlation functions may turn out to be richer than one would naively assume.

Cosmological fields are often more conveniently analyzed in spherical basis. The Fourier-Bessel decomposition that has been used in the past suffers from a serious problem. It is meant to be used for a field distributed in infinite space and provides more basis functions than needed for the real galaxy surveys in a finite volume.

In this paper I propose a much simpler algorithm for an orthogonal expansion of the observed field. Not only is the new approach more economical theoretically, it will also save you a lot of CPU time. My colleagues from Carnegie-Mellon University and Jet Propulsion Lab found this approach very useful in their work and cited it in their papers. I am considering sending it to a journal so that it is a proper peer reviewed publication.

The most computationally intensive part of computing 3-point functions is the triplet counts in random catalogs. For large galaxy samples this computation is prohibitively slow on a regular workstation.

In this work, we describe a purely analytic way of computing the expected triplet expectations from periodic boxes. The derivation is based on simple Euclidean geometry and a bit of calculus and can be completed in milliseconds on any computer.

This is our first attempt (and I believe the first attempt at the time) at detecting the BAO feature in galaxy bispectrum (Fourier space). Similar attempts have been made before in configuration space. Our goal is to improve the methodology and apply it to DESI data.

Bispectrum is a function of five variables and is very difficult to interpret. In this work we showed that bispectrum can be reduced to three multipoles without significant loss of information. We also developed a simple Fisher Information code for predicting cosmological constraints that are derivable from the bispectrum. More accurate methods that build on this and provide an improved treatment of nonlinear effects have been proposed since then.

Fourier transforming galaxy distribution is tricky for many reasons. One reason is that many physical effects tend to work along the line-of-sight from the observer, making Fast Fourier Transform methods difficult to apply. When I was a postdoc this problem of "plain-parallel approximation" was a hotly discussed topic. We have understood the problem since then and have good methods for working around it.

This paper was one of the first to study the ways of working around wide-angle effects. We showed how the magnitude of the systematic effect can be computed based on the size of the survey. The systematic effects turned out to be more significant than what was initially anticipated.

My best paper from the postdoctoral period (~250 citations). I used BOSS data to measure the growth rate of structure and the distance-redshift relationship.

My first highly cited paper (~200 citations). I used BOSS data to constrain parameters of Dark Energy and Gravity.