Astro-Data Distributed Computing

Robert Coleman* and Tom Devine

Lane Department of Computer Science and Electrical Engineering, West Virginia University, Morgantown, WV 26505

Presentation Category: Physical Sciences & Engineering (Poster Presentation #146)

Student’s Major: Computer Science

The volume of data collected through astronomical observation has prompted data scientists to research ways to classify space objects. The need to process data meant building a network of computers with high scalability, security, etc. has become normal. Distributed computing environments (DCE) take a system of distributed computers and feed data through to yield a result. Scientists use the power of DCEs to help identify and categorize objects, like pulsars. Pulsars result from massive stars dying. They emit electromagnetic radiation from their magnetic poles and spin at incredibly fast rates, some as fast as 700 rotations per second. Measuring the attributes of pulsars is conducted by studying their emissions that reach earth. The PRESTO suite records this data for observers. All-sky surveys using extremely sensitive instruments have produced petabytes of observational data, and experts find it very difficult to locate pulsars because many recorded measurements are Radio Frequency Interference. DCEs can process pulsar data and determine which potential candidates are likely pulsars and should be observed. This is important because it eliminates manual labor of locating pulsar candidates to study. Scientists use pulsars as giant regular clocks in space that enable us to measure distances between galaxies, detect gravitational waves, and study extreme physics. To accomplish radio pulsar candidate classification, we will locally emulate a DCE and use it as a development environment for distributed machine learning algorithms. Our aim is to further test and develop distributed semi-supervised learning algorithms that have shown promise for radio pulsar candidate classification.

Funding:

Program/mechanism supporting research/creative efforts: WVU's Research Apprenticeship Program (RAP) & accompanying HONR 297-level course