My Journey in Astronomy

I had broad interests in astronomy research and outreach. My primary research interests focused on large-scale structure formation in the Universe and cosmology, but I also performed analysis of Galactic structure and observations for exoplanet hunting. My projects were based on huge volumes of astronomical survey data and large suites of numerical simulations.

Cosmic Voids in the Large-scale Structure of the Universe

large-scale structure
An illustration showing structures observed by CfA, 2dFGRS, and SDSS, comparing with mock catalogs constructed from Millennium Simulation. (Springel et al. 2006)

Cosmic Voids are large underdense regions present in the hierarchical structure of the Universe. Surrounded by filaments, walls and clusters, voids are an essential component of the cosmic web. They were first discovered in early galaxy redshift surveys over thirty years ago. More recent redshift surveys such as the Sloan Digital Sky Survey (SDSS), have greatly expanded our view of the large-scale structure, and provide much larger data sets to study void properties systematically and in detail.

Cosmic voids have been recognized as interesting cosmological laboratories for investigating galaxy evolution, structure formation and cosmology. To unleash the power of these cosmological applications, it is important to first find voids robustly from simulations, mock galaxy catalogs and galaxy surveys. My collaborators and I created a cosmic void catalog using the large-scale structure galaxy catalog from the Baryon Oscillation Spectroscopic Survey (BOSS). This galaxy catalog is part of the SDSS Data Release 12 and is the final catalog of SDSS-III (arXiv:1501.00963).

Voids labeled on SDSS geometry
A thin slice of BOSS galaxies (red) and random buffer particles (gray) with the identified cosmic voids labeled by blue circles. The sizes of the circles indicate the effective radii of the voids.

Using the ZOBOV void finding algorithm (Neyrinck, 2008), we identified a total of 10,643 voids. After making quality cuts to ensure that the voids represent real underdense regions, we obtained 1,228 voids with effective radii spanning the range 20–100 h-1Mpc, and with central densities that are, on average, 30% of the mean sample density. In addition, we constructed mock void catalogs from 1000 mock galaxy catalogs. These BOSS and mock void catalogs are useful for a number of cosmological and galaxy environment studies. The catalogs are publicly available here: SDSS Cosmic Void Catalogs.

Reference: A Cosmic Void Catalog of SDSS DR12 BOSS Galaxies, arXiv:1602.02771

Cosmological Constraints using Cosmic Voids

A classical method for probing the expansion history of the universe was proposed by Alcock & Paczynski (1979). The so-called Alcock-Paczynski test (AP test) is a purely geometric test that examines the ratio of the observed angular size to radial size of objects that are known to be intrinsically isotropic. First proposed by Ryden (1995) and extensively discussed by Lavaux & Wandelt (2012), cosmic voids provide an attractive alternative for applying the AP test. The low-density nature of voids make them easier to model and understand due to their lack of multi-streaming, thus it may be easier to model systematics such as redshift-space distortion (RSD) effects. Although the shapes of individual voids can be very irregular, the AP test can be applied to stacked voids to significantly reduce this "shape noise".

A stacked void
A slice through 717 stacked BOSS CMASS voids. The blue circle has a radius of 0.7, highlighting the approximate region within which we measure the void shape, while the dashed red ellipse has the ellipticity of 0.929 that is measured for this stacked void.

My collaborators and I applied the AP test to our cosmic void catalogs created with BOSS data. We also used 1000 mock galaxy catalogs that match the geometry, density, and clustering properties of the BOSS sample in order to characterize the statistical uncertainties of our measurements and take into account systematic errors such as redshift space distortions. For both BOSS data and mock catalogs, we used the ZOBOV algorithm to identify voids, we stacked together all voids with effective radii of 30–100 h-1Mpc in the redshift range 0.43–0.7, and we accurately measured the shape of the stacked voids. Our tests with the mock catalogs showed that we measured the stacked void ellipticity with a statistical precision of 2.6%. We successfully obtained a constraint of Ωm = 0.38 +0.18-0.15 at the 68% confidence level from the AP test. We discussed the sources of statistical and systematic noise that affect the constraining power of this method, and how constraints will improve in future surveys with larger volumes and densities.

Reference: Cosmic Voids in the SDSS DR12 BOSS Galaxy Sample: The Alcock-Paczynski Test, arXiv:1602.06306

Primordial Non-Gaussianity with Higher Order Moments

The earliest phases of the Universe are subject to much speculation. There are many theories about the very early universe, and a very promising paradigm is that the very early universe experienced an extremely rapid epoch of exponential expansion, called inflation. Inflation was originally motivated by several problems in the Big Bang cosmology pointed out in the 1970's. In 1980, Alan Guth found an exponential expansion of space can be driven by a negative-pressure vacuum energy density, and he proposed the hypothesis of inflation (Guth, 1981). The standard inflationary paradigm predicts nearly Gaussian and scale invariant primordial density fluctuations, which are consistent with the observations of the Cosmic Microwave Background (CMB) and Large-Scale Structure (LSS) in the last few decades. However, even the simplest inflation model predicts some small deviation from Gaussianity, and different models predict different amounts and flavors of non-Gaussianity. Detecting primordial non-Gaussianity is thus a very powerful tool for constraining inflationary models.

Skewness and kurtosis measurements
Skewness (S3) and kurtosis (S4) measurements as a function of smoothing scale for the dark matter distribution in the Gaussian and non-Gaussian simulations.

There are several avenues for constraining primordial non-Gaussianity with galaxy surveys, including the galaxy power spectrum, higher order correlations of the density field, e.g., the bispectrum, and statistics of rare peaks, i.e., the abundance of massive clusters. But a much simpler set of statistics for quantifying departures from Gaussianity are the higher order moments of the density field, of which the most frequently used are the third order normalized moment skewness and fourth order normalized moment kurtosis. Though gravitational evolution contributes most of the signal in these moments in the present day density field, small departures from Gaussianity in the primordial density field may still cause slightly different skewness and kurtosis today, which may be detectable in sufficiently large galaxy redshift surveys.

We used cosmological N-body simulations to investigate whether measurements of the moments of large-scale structure can yield constraints on primordial non-Gaussianity. We measured the variance, skewness, and kurtosis of the evolved density field from LasDamas simulations with Gaussian and three different non-Gaussian initial conditions. We showed that the moments of the dark matter density field differ significantly between Gaussian and non-Gaussian models. When we restricted our measurements to galaxy mocks with volumes equivalent to the Sloan Digital Sky Survey samples, we showed that the probability of detecting a departure from the Gaussian model is high by using measurements of the variance, but very low by using only skewness and kurtosis. We estimated that in order to detect an amount of non-Gaussianity that is consistent with recent CMB constraints using skewness or kurtosis, we would need a galaxy survey that is much larger than any planned future survey. However, future surveys should be large enough to place meaningful constraints using galaxy variance measurements.

Reference: Constraining Primordial Non-Gaussianity with Moments of the Large Scale Density Field, arXiv:1404.3725

Galactic Structure with the Spatial Correlation Function

SDSS SEGUE geometry
(Left) Sky map of the 152 SEGUE fields used in this study, shown in a Mollweide projection in Galactic coordinates. (Right) A selection of SEGUE pencil-beam fields in a slice perpendicular to the Galactic plane, with red points indicating stars with distances between 1–3 kpc.

The Milky Way provides a unique laboratory for studying the structure of a galaxy in detail, by allowing us to measure and analyze the properties of large samples of individual stars. Recent surveys have placed strong constraints on the smooth components of the Milky Way, and have discovered significant spatial substructure in the Milky Way, such as stellar streams and stellar overdensities. The Sloan Extension for Galactic Understanding and Exploration (SEGUE) is a spectroscopic sub-survey of the SDSS that focused on Galactic science. SEGUE data provides the largest spectroscopic sample of Galactic stars currently available, and covers a more extensive volume of the Milky Way than previous studies, probing from the local disk all the way to the outer stellar halo.

Constraints on Galactic components
The constraints on the thin- and thick-disk obtained from the G-dwarf correlation function using MCMC model fitting.

The spatial two-point correlation function is one of the simplest and most effective statistical tools for studying clustering in general, and it is widely used in studies of the large-scale structure of the Universe. However, it has rarely been used in Galactic structure studies, mainly due to the lack of large and homogeneous spectroscopic stellar samples. There have only been a few applications of the correlation function applied to Galactic halo stars, especially giants and blue horizontal-branch (BHB) stars, but the sample sizes were limited. We measured the two-point correlation function of G-dwarf stars within 1–3 kpc of the Sun in multiple lines-of-sight using a large G-dwarf sample created by Schlesinger et al (2012) from the SDSS SEGUE survey. The shapes of the correlation functions along individual SEGUE lines-of-sight depend sensitively on both the stellar-density gradients and the survey geometry. By fitting smooth disk galaxy models to our SEGUE clustering measurements, we were able to obtain strong constraints on the thin- and thick-disk components of the Milky Way. Moreover, we exhibited an excess of clustering at small scales (< 50 pc), which suggests the presence of small-scale substructure in the disk system of the Milky Way.

Reference: Probing Galactic Structure with the Spatial Correlation Function of SEGUE G-dwarf Stars, arXiv:1507.01593

Observations for Exoplanet Hunting

KPNO 2.1-meter inside
The 2.1-meter telescope at KPNO. The photo was shot inside the dome with the Photosynth app on my iPhone 4.

The Kilodegree Extremely Little Telescope (KELT) survey is a ground-based program designed to search for transiting exoplanets orbiting relatively bright stars. The project has two telescopes — KELT-North at Winer Observatory, Arizona and KELT-South at the South African Astronomical Observatory. These telescopes have particularly wide fields of view, allowing KELT to study a large number of potential exoplanet host stars. The exoplanet candidates detected can be followed up by small, ground-based observatories distributed around the world. You can learn more about the KELT project from the article arXiv:1605.02425 written by the KELT science team.

During 2012, I contributed about 30 nights at the Kitt Peak National Observatory (KPNO) doing follow-up observations for some of the KELT exoplanet candidates using the 2.1-meter telescope at KPNO. These observations supported the discovery of many exoplanets:

Astronomy Education and Public Outreach

SDSS Social Media Efforts

During 2013 to 2015, I was part of the Sloan Digital Sky Survey (SDSS) Educational and Public Outreach team. I was in charge of the SDSS Chinese Facebook Page and SDSS Chinese Weibo. At that time, I wrote hundreds of weibo about SDSS science and general astronomy. I also translated many astronomy related articles into Chinese and made Chinese subtitles for many educational videos. Part of these articles can be viewed at, which is my personal collection of astronomy EPO related materials in Chinese. Some of the videos I subtitled can be found on my TED translator profile.

CMB Maps
Download the free CMB Maps app for iOS! The COBE map and the WMAP map are provided by the NASA / WMAP Science Team. The Planck map is provided by ESA and the Planck Collaboration.

CMB Maps App for iOS

Cosmic Microwave Background (CMB) is the afterglow radiation from the Big Bang. Its temperature is extremely uniform, but with tiny fluctuations. In 1992, COBE satellite made the first detection of the CMB fluctuations. The following WMAP and Planck mission were able to scan the sky at better and better resolution. These CMB observations have provided us with great understanding of the Universe over the last decades.

Inspired by the WMAP lenticular card, I made a free iOS app that allows one to view and compare different generations of CMB fluctuations on your iOS devices. It includes 4K resolution all-sky COBE/WMAP/Planck CMB maps with the ability to zoom in and scroll around for map details, and one can compare different maps by simply tilting your device! The app is available on the App Store.