REU Site: EXERCISE - Explore Emerging Computing in Science and Engineering

Salisbury University, Salisbury MD

Investigators

Abstract

The project is a renewal of the Research Experiences for Undergraduates (REU) EXERCISE (Explore Emerging Computing in Science and Engineering) site at Salisbury University (SU) for the next three years. EXERCISE is an interdisciplinary project that explores emerging paradigms in parallel computing with data and compute intensive applications in science and engineering. The goal of the project is to offer student participants, particularly from primarily undergraduate institutions (PUIs), a valuable research experience in parallel computing. The project will promote "parallel thinking", an important computational thinking skill guiding current generation students into the twenty-first century computing era. The site will prioritize recruiting under-represented students and females, and attract students from local historically black college and universities (HBCUs), PUIs, and community colleges on Maryland's Eastern Shore into computational science and engineering majors and the general Science, Technology, Engineering, and Mathematics (STEM) fields. The Principal Investigator, together with faculty mentors, will supervise a 10-week REU program that gives a diverse cohort of students a taste of computational thinking in the domain of parallel computing and also an understanding of the graduate school experience. The host institution SU will collaborate with the University of Maryland Eastern Shore, an HBCU, and the University of Maryland College Park for multi-disciplinary faculty expertise and diverse summer activities including field trips, social activities, high school outreach, and graduate school application information sessions. Processing complex information and large data in conventional von Neumann computer architectures is becoming increasingly difficult. Computers are undertaking a fundamental turn toward concurrency architectures such as hyperthreading, multi-core, and many-core architectures. Emerging parallel and distributed computing paradigms adapted to these concurrent architectures have begun to demonstrate the power of solving problems with large datasets and high computational complexity in a wide range of applications. However, there are fundamental difficulties in program semantics related to process interleaving: a parallel program can yield inconsistent answers, or even crash, due to unpredictable interactions between simultaneous tasks. Secondly, communication, memory access, and I/O overhead may result in run-time delays. Finally, it is difficult to ensure that programs consume resources in a manner that simultaneously achieves efficiency and meets performance goals. The REU Site will focus on four aspects of parallel computing, namely: algorithms, software, architecture and applications to address these parallel computing challenges. Students will work with faculty mentors in completing cutting-edge research projects to tackle data and compute intensive applications that emphasize the above four aspects. By the end of program, students will acquire valuable skills, gain a broader and deeper understanding of research, and develop greater confidence in their abilities. In particular, they will be exposed to emerging paradigms in parallel computing such as Map-Reduce and Graphical Processing Unit computing, and will have opportunities to explore concurrent software and multiprocessor architectures, and design efficient parallel algorithms, and to tackle data and compute intensive problems in computer and social networks, image and natural language processing, pattern recognition and machine learning.