David M Goldberg

We present results for utilizing weak lensing analysis in the identification and classification of the dark matter substructure in galaxy clusters. Using a previously developed flexion analysis package (FLEXTOOL) gravitationally lensed images are analyzed for their flexion signal, the anisotropic inward "bowing" of a lensed object. The measured flexion signal can be decomposed toward a nearest-neighbor lensing galaxy as a means for identifying dark matter structure within the lensing galaxy cluster. A figure-of-merit is developed for classifying the underlying substructure using source object size, measured flexion strength and nearest-neighbor radial distance. The analysis is applied to the Abell 2744 and MACS J0416.1-2403 galaxy clusters in the HST Frontier Fields program. This is the first such work that directly identifies individual substructure using weak gravitational flexion.

Strong gravitational lensing of quasars is a powerful tool to learn about the distribution of dark matter in lensing galaxies. Multiply imaged quasar systems have symmetries which allow for an understanding of the lensing galaxy without detailed mass reconstructions. Keeton et al. (2005) defined a useful expression for the flux anomaly of "fold'' lenses, which we might naively expect to have the same flux: Rfold=(fA-fB)/(fA+fB), where "A'' and "B'' represent the positive and negative parity images straddling a critical curve. We show that the geometric configuration of the images greatly constrains the possible flux anomalies allowable from a smooth galaxy potential. Using gravlens, we create a number of simple galaxies from various mass models to put our solution to the test, and find that simulated flux anomalies are reproduced to an accuracy of |δ R| < 0.04. We then apply our approach to a radio sample of 9 well-studied fold lenses and quickly identify those with significant substructure.

We measure the distribution of void sizes in the SDSS using an objective void finding algorithm and statistically characterize the spectrum of voids by measuring the Void Probability Function for volume-limited samples of this survey. These void statistics provide powerful constraints on models that describe the relationship between luminous and dark matter. The properties of galaxies in voids provide further contraints on galaxy formation models: On average, galaxies in the rarified environments of voids exhibit bluer colors, higher specific star formation rates, lower dust content, and more disk-like morphology than objects in denser regions. This trend persists in comparisons of samples in low vs. high-density regions with similar luminosity and morphology. Large-scale modulation of the halo mass function partially explains this dependence of galaxy evolution on environment.

We are developing an integrated suite of parallelized analysis and visualization tools, designed to run on large Beowulf-class clusters of computers. Using efficient and scalable algorithms, this toolkit will harness the low-cost computing power of these parallel systems to allow real time analysis and visualization of data sets in the multi-terabyte to petabyte range. Software components for data input/output, transformations, statistical analysis, and 3D visualization will communicate and operate seamlessly within a symbolic user interface.

Ever since the first off-the-shelf distributed memory computing cluster (Beowulf) was constructed at the Goddard Space Flight Center, systems composed of commodity hardware have been attractive solutions to difficult scientific computing problems. Beowulfs are relatively easy to construct, maintain, and program, and provide an unbeatable cost per performance ratio. The Drexel University Physics Department was the first academic department to utilize Beowulf clusters, and has since built a successor. We are in the process of building a new cluster, one with the express design goal of maximum flexibility at very low construction price. We discuss a possible design of such a system with dual Athlon MP processors and channel-bonded 100 Mbps Ethernet. This hardware combination provides maximum performance at a minimum price, which is further strengthened by in-house construction of the entire system. We present a review of Beowulfs, comparisons of other hardware solutions and performance projections.

Developed at Yale for use in undergraduate astronomy courses, Showfits is the premier web-based image analysis tool. Showfits provides all the functionality necessary for undergraduate level (and often research level) image manipulation, including image convolutions, rotations, aperture photometry, centroid calculation, flat-fielding, dark subtracts, and stacking. Moreover, the Graphical User Interface is platform independent and highly intuitive. Most of the calculations are transparent to the user, making the program useful pedagogically. Finally, as a web-based tool, it is a convenient way to make images taken on a local telescope easily viewable to the outside public.