SSG Galaxy Research

Classification of Spectra

Spectral Classification with LLE

Through a Principal Component Analysis of 400,000 galaxy and 50,000 QSO spectra (Yip et al 2004a, 2004b) we have shown that we can describe the spectral properties of galaxies and QSOs with a small number of orthogonal components. This represents a reduction in data size by a factor of nearly 400 over the complete spectra. The coefficients of the first three eigencomponents correlate with physical properties of the galaxies such as star-formation rate, and post star-burst activity. PCA does not, however, perform well when reconstructing high-frequency structure within a spectrum (i.e. the distribution of emission lines, lines-ratios, and line-widths). High frequency, or local, features do not contribute significantly to the total variance and are, therefore, not represented in the primary eigencomponents. Other dimensionality reduction techniques such as Local Linear Embedding (Roweis and Saul 2000) can over come these issues providing (vanderPlas and Connolly 2009) a simple and automated classification.

Finding Supernova in Spectra

SN Spectra
Extracting SN from SDSS spectra.

Understanding the evolution of the Type Ia supernova rate is essential to our understanding of the star formation history of the Universe. Lately, high redshift measurements of the Type Ia rate have been receiving a lot of publicity, but understanding supernovae in the local universe is also important. In particular, comparison of detection methods and the associated rate measurements at low redshift can improve methods at higher redshift. Using a novel approach for detecting Type Ia supernovae in single epoch spectroscopic observations we identify 100+ SN within the SDSS survey data. This results in a luminosity weighted rate of 0.24 SNu at z=0.1. Our measurement is in good agreement with others at low redshift but is one of the largest local samples todate.

Resolving Star Formation and Mass

Extracting star formation from SDSS images.

Given a multi-color image of the sky or even just multi-color cutouts around individual galaxies the resolved nature of these images provides a unique opportunity to study the physical processes that drive the formation and evolution of galaxies within our universe. Previous work has shown that the environment in which agalaxy resides can significantly affect the star formation activitywithin the galaxy itself (Gomez et al 2003, Lewis etal 2002, Hogg etal 2003}. For galaxies in overdense regions the star formation appears to be more centrally condensed than for field galaxies. A number of mechanisms for causing this circumnuclear activity (as opposed to the more general star formation in HII regions within the disks of galaxies) have been suggested. These include tidal stripping of the gas from the outskirts of a galaxy as it impacts the intracluster medium, merging of galaxy sub-components and galaxy harassment. We can address these questions by decomposing the internal photometric structure of galaxies into basic constituents such as the age of the stellar population, their metallicities and their dust content by fitting spectral energy distributions (SEDs) to the observed fluxes(see Figure 2 for an example of this).

Classifying and Monitoring Active Galactic Nuclei

BAL Cat L-z Plane
Absolute i-magnitudes and redshifts of SDSS DR5 BAL quasars.

Active Galactic Nuclei (AGN) are important actors in the cosmic history of black hole growth, galaxy evolution (e.g., Di Matteo et al., 2005; Hopkins et al., 2007), galaxy clusters (e.g., Begelman, 2004), and even the ionization level of the universe (e.g., Loeb & Barkana, 2001). AGN are highly luminous across much of the electromagnetic spectrum, with each emission regime (radio, infrared, optical, UV, X-rays, gamma rays) revealing important information about a particular component of AGN emission. Unfortunately, it can be very difficult to collect large, multi-wavelength AGN samples that are representative of the full AGN population. Each observing mission has its own characteristics, and deep observations are only available in narrow "pencil beams" at many wavelengths. As an example of current research, we have recently analyzed relations between X-ray and optical/UV accretion disk emission to see whether coronae are ubiquitous components of black hole accretion (Gibson et al. 2008b). We have also fit about 80,000 SDSS quasar spectra to identify and characterize over 5000 quasars exhibiting broad absorption lines (BALs) from powerful outflows (Gibson et al. 2009a,b). These outflows are likely to be a structural or evolutionary phenomenon common to AGN, and may be an important means by which AGN affect their surroundings. Long-term monitoring of the evolution of quasar spectra may also reveal the structure and origin of BAL outflows (e.g., Gibson et al. 2008a, 2009c).