Projects

Yen-Chi Chen, Shirley Ho, Peter Freeman, Christopher Genovese, Larry Wasserman
A filament is a one-dimensional, smooth, connected structure embedded in a multi-dimensional space that characterizes the high density regions. Matter in the Universe tend to aggregate around these filaments which weave our Universe into an intricate network structure. We use ridges of density function (density ridges) as tracers for the filaments and apply our method to the SDSS DR12 to construct a filament catalogue.

Link to our publicly available catalogue: Cosmic Web Reconstruction


Searching for Voids in MassiveBlack II using Persistent Homology
Jisu Kim, Jessi Cisewski, Tiziana DiMatteo, Alessandro Rinaldo, Larry Wasserman 
This project proposes a new approach using persistent homology to find voids in the universe. The proposed method counts the number of statistically significant voids by considering the multi-scale topological structure of the data. Furthermore, the proposed method provides a finer classification of voids with respect to particular topological characteristics.


Exploring the Intergalactic Medium
Collin Eubanks, Jessi Cisewski, Rupert Croft, Doug Nychka, Larry Wasserman, Kolby Weisenburger
https://collineubanks.wordpress.com/research/
The intergalactic medium (IGM) is a diffuse, highly inhomogeneous gas that permeates intergalactic space and hosts a majority of the baryons in the Universe. This matter is too diffuse to be seen by the naked eye; however, its presence is clearly marked by absorption lines in the spectra of luminous, high redshift quasars. We explore nonparametric methods for constructing a 3D map of the neutral hydrogen density fluctuations of the IGM and apply them to the BOSS DR12 data of SDSS-III.


Evolution of Galaxy Morphology 
Jining Qin, Peter Freeman, Rafael Izbicki, Ann Lee

How galaxies appear on the sky evolves with time. In this project, we attempt to quantify the evolution of morphology by analyzing galaxy images. Such images are intrinsically high-dimensional data, so we reduce the dimensionality of the problem by using ~10 summary statistics that describe the images...and then we track the evolution of this space with redshift. Statistical methods that we bring to bear on this problem include density ratio estimation, conditional density estimation, spectral series estimation, etc.


Combining Photometric Redshift Estimators 
James Eby, Dritan Kodra, Peter Freeman, Ann LeeJeff Newman, Chad Schafer

There are a number of publicly available photometric redshift estimation algorithms, and it is an empirically established fact that given the same input, these algorithms can (and usually will) generate different estimates for the photometric redshift error function, p(z). How should these estimates be combined?


Approximate Bayesian Computing: Inferring Stellar Multiplicity 
Eric Alpert, Carles Badenes, Peter Freeman, Chad Schafer

Type IA supernovae occur in binary star systems, after one star transfers sufficient mass to its white dwarf companion. To determine the rate of Type IA supernovae in a given galaxy, it is necessary to determine what fraction of stars lie in multiple-star systems. In this project, we use the method of approximate Bayesian computation, or ABC, to attempt to infer stellar multiplicity by comparing output from a simulation code to data from the SDSS SEGUE and APOGEE projects.