Broadly, my research is in the fields of cosmology and large-scale structure. I am interested in understanding how structure forms in the universe, which involves running and analyzing complex simulations that numerically solve equations of motion for cosmological systems. These simulations allow us to determine predictions for the nonlinear behavior of cosmological theories as well as search for obervational signatures that could potentially allow us to distinguish between competing cosmological models. In particular, the discovery of the accelerated expansion of the Universe has motivated a wealth of theories of dark energy and modified gravity. Simulations allow us to determine new ways of testing these theories.

Below, you will find information on some of the projects that I am involved in as well as work I did for my thesis. I have also written an overview of my current research intended for a general audience.

Unscreening Modified Gravity with Cosmic Voids

Posted on March 1, 2018

The Vainshtein mechanism is very efficient at screening dark matter halos, but not other elements of the cosmic web, suggesting voids as an effective test of many models of gravity. We computed density, velocity, and fifth force profiles of cosmic voids in cosmological N-body simulations, finding that voids are unscreened at all radii. Tangential velocity and velocity dispersion profiles of voids are excellent tracers of screening and may be fruitful probes of modified gravity.


Stacked profiles of the tangential velocity dispersions of halos around cosmic voids show that voids are unscreened at all radii.

Corner Modes in Initial Conditions

Posted on May 9, 2017

In view of future high-precision large-scale structure surveys such as Euclid, it is important to quantify the percent and subpercent level effects in cosmological N-body simulations. One such effect involves whether modes above the one-dimensional Nyquist frequency, the so-called “corner” modes, are zeroed in the initial conditions. We found that zeroing the corner modes affects modes even below the Nyquist frequency at high redshifts, though this effect is smaller than that of particle discreteness, and corner modes are repopulated as the simulation evolves, even if initially zeroed.


This figure shows the ratio of power spectra in simulations without corner modes to those with corner modes, as a function of redshift. The vertical line gives the one-dimensional Nyquist frequency.

Screening in the Cosmic Web

Posted on April 20, 2015

Theories that modify general relativity to produce accelerated expansion often require a screening mechanism to suppress scalar interactions on solar system scales. I have previously found that the Vainshtein screening mechanism depends on the cosmic web morphology of dark matter particles as determined by ORIGAMI: halo particles are screened while filament, wall, and void particles are unscreened, independent of density. Recently, I have showed that this is in contrast to the chameleon screening mechanism, which does not depend on the cosmic web but instead on the mass and environmental density of dark matter particles and halos.


In the above I plot histograms of ΔF, the difference between the ratio of the fifth force to gravitational force and the linear ratio, for dark matter particles according to their morphology. ΔF is -1 if screening is working and 0 if a particle is unscreened. In the Vainshtein mechanism (left), the screening depends on the cosmic web morphology with halo particles always screened, while there is no cosmic web dependence in the chameleon mechanism (right). The top, middle, and bottom panels reflect differences in the screening model parameters.

Single-stream Voids Percolate

Posted on November 12, 2014

My cosmic web identification routine ORIGAMI defines voids as dark matter particles that have not undergone any shell-crossing, thus they are in the single-stream regime. We found that single-stream voids are not surrounded on all sides by walls and filaments and instead percolate, filling the simulation volume. This percolation can even occur after restricting single-stream void particles to have large volumes/low densities, depending on the threshold. The paper has been accepted for publication in MNRAS.


In this thin, 2Mpc/h-thick simulation slice, void particles with normalized volumes above 8 are plotted in color over halo, filament, and wall particles in black. Connected void particles are given the same color, and the largest void in dark blue percolates.


Posted on October 12, 2012

We are developing a suite of large scale N-body cosmological simulations called Indra. We are running hundreds of gigaparsec simulations to capture the very large-scale modes of the matter power spectrum with an excellent handle on cosmic variance. Additionally, all particle data for each simulation run and each snapshot, as well as halo catalogs, will be made available to the public.

The full suite of simulations will amount to over a peta-byte of data, presenting tough challenges for data organization in a way that allows for fast analysis. We have explored developing an inverted index for the particle data, which will total 35 trillion particles, which was published in SSDBM.


Posted on October 12, 2012

ORIGAMI is a dynamical method of determining the cosmic web morphology of particles in a cosmological simulation by checking for whether, and in how many dimensions, a particle has undergone shell-crossing. For more details and to download the code, check out the main ORIGAMI page.

The Log-density Field

Posted on October 12, 2012


A logarithmic transform of the density field has been shown to restore information to the matter power spectrum in the nonlinear regime. We have found that this transform also improves the relation between the density and Lagrangian displacement fields, as shown in the plot above. For more details, check out our paper which has been published in ApJ.

Photometric Supernova Cosmology

Posted on October 12, 2012

Future supernova surveys will collect phometric light curves for thousands of supernovae, only a fraction of which will be followed up to obtain spectroscopic identification. I worked with Adam Riess and Renee Hlozek to develop and test statistical methods that are able to make use of the full set of unidentified supernova light curves without being biased by the presence of Core Collapse supernovae in the sample.

Previously, I spent two years working with Adam Riess and Chuck Bennett doing supernova simulations and analysis for the ADEPT concept study as part of NASA's JDEM program. For a primer on cosmology with Type Ia supernovae, check out my Second Year Seminar Presentation.