Education

NCSA's new undergrad internships

by Jan Alexander, Freelance Writer

A new program at NCSA open to undergraduate students allows them to discover the role of supercomputing in science and education. The Research Experiences in Computational Science for Undergraduate Students Program, which began in fall 1992, was initiated by Bob Wilhelmson (NCSA research scientist and UIUC atmospheric sciences professor) and Nora Sabelli (NCSA senior research scientist on leave to NSF as program director, Applications of Advanced Technologies). Lisa Bievenue, specialist in Education, and Robert Panoff, senior research scientist in Education, coordinate the program.

Also known as the Undergraduate Internship in Computational Science, the program was created so that students from small colleges and universities could take advantage of the environment at a large research institution for a full semester.

Contributing to NCSA

``We encourage applicants from social science and humanities disciplines as well as basic science departments,'' says Bievenue. The program, which is open to students in all fields of study, helps students understand how computational science is instrumental in whatever field they choose.

``Because of the time and concentration of effort involved,'' says Panoff, ``it's really an immersion program rather than an internship program.'' Students generally work 30-40 hours a week on their research projects. ``They come in as undergrads, but are working at almost a graduate level.'' Panoff adds, ``They contribute significantly to the research environment. For example, many of the results [in the projects] would not have been achieved without their contributions.''

Selecting participants

Criteria for student selection includes topic for research, science and computational backgrounds, and overall credentials.

Student interns are matched with mentors who oversee their projects and give training and support. The selection committee seeks a diversity of backgrounds to give participants a range of experiences.

Funding is provided jointly through a grant from NSF and NCSA. Besides a stipend and housing expenses, the program pays for up to five class hours for students who want to take a class at UIUC. As word gets out about the program, more people apply. Last fall, ten people applied; this spring, 50. Bievenue would like to expand the program and is working on more funding. ``The goal this year is to demonstrate the utility of this approach,'' says Panoff. The hope is that positive results of NCSA's program will encourage other supercomputing centers and universities to offer similar, semester-based programs.

``We're always looking for exciting opportunities for research,'' says Panoff, who would like to expand the projects available. After this semester ends, he would like to help current students continue researching on their own campuses via Internet. This could enable former students to be electronic mentors for incoming students also.

Broadening experiences

Currently there are five interns at NCSA. Projects this semester include designing mathematics software for elementary students, implementing parallel algorithms, compiling nineteenth-century census data for study, studying light rays in black holes, and predicting the structure and function of proteins.

At weekly seminars, interns get a chance to interact with other REU students from UIUC and exchange ideas. At one of these seminars, they will also make presentations on their projects.

According to the participants, all feel they have benefitted from the program through access to new resources and ideas. They feel better prepared for graduate school, too.

Developing math software

One of the students enrolled in the program, Michael South from Valdosta State College in Georgia, is studying physics and math. He works with Panoff to develop software to introduce kids to concepts usually only seen in advanced college math, such as abstract algebra and fractal geometry. South sees a need to attract children to math and interest them in it at an early age. Like others in the program, South had no previous supercomputing experience. Access to the CRAY Y-MP supercomputer has allowed him to do many fractal experiments.

What are the benefits of the program? ``There are so many it's hard to put into words. There's so much more I can do here with visualization programs.'' He says the experience at NCSA has expanded his knowledge of teaching and given him new ways to think about computer applications. South credits Panoff with making the adjustment and training fairly easy.

After leaving NCSA, South plans to pursue a master's degree and would like to teach math at the secondary or elementary level: ``There's so many ways to do it [teach] that haven't been explored.''

Implementing algorithms for parallelization

Boyana Radenska from Southwest State University, Marshall, MN, is majoring in computer science. Working with Michael Heath, head of NCSA's Mathematics and Computer Science Group, she is implementing parallel algorithms for matrix multiplication and factorization. For this work, she uses parallel computers--the Thinking Machines' CM-5 at NCSA and an Intel iPSC/860 hypercube at Oak Ridge National Laboratory. ``He's always there when I have questions,'' she comments on Heath's helpfulness.

Heath got involved with the program because of his previous involvement in a similar program at Oak Ridge National Laboratory in Tennessee.

Radenska enjoys learning about parallel processing, using the resources available, and meeting new people. She plans to graduate in spring 1995 and go on to graduate school for a Ph.D. in computer science. Afterwards she is considering teaching.

Computing and social science

Intern Robert Shaughnessy, from Rockhurst College in Kansas City, MO, is a double major in computer science and political science. Shaughnessy works with Terry Finnegan, NCSA postdoctoral research fellow, to compile nineteenth-century census and other data to show the impact of the Civil War on an area in North Carolina between 1850 and 1890 [see access , October-December 1992]. Shaughnessy is working on a code for record linkage, a key function in the project. He likes working on his project because it involves computers and social science. Finnegan says that Shaughnessy has been a helpful addition to the project: ``We consider Robert to be our colleague.''

Shaughnessy, who takes a UIUC class in artificial intelligence, enjoys the class material. In fall 1993 he plans to graduate and go to graduate school to study computer science, particularly artificial intelligence.

Studying black holes

Joseph Libson, from Washington University, St. Louis, MO, is working with Edward Seidel, NCSA research scientist in gravitation, on a project involving the study of black hole physics. Libson is exploring where light rays go through a system in a black hole.

Seidel and Libson found out about each other, and Libson about the program, through a mutual acquaintance, Wai-Mo Suen of Washington University. Seidel is glad to have Libson's assistance.

Seidel said the program provides good research experience for undergraduates. ``Research is not like the textbook learning they're used to,'' he says. After Libson leaves, Seidel would like to continue having other interns work on his ongoing project. Libson plans to graduate in May 1993 and continue his study of relativity in graduate school.

Projects in computational biology

Intern Kevin Scott, Lynchburg College of Lynchburg, VA, is working on three research projects in various aspects of protein folding with Shankar Subramaniam, NCSA research scientist in computational biology.

Scott has been busy in his short time at NCSA--completing one project and co-authoring two papers. Collaborating on a coding scheme for storing aligned genetic sequences, he wrote programs to determine nucleic acids like those in the Human Genome Project. Another project on mutagenesis is in the conceptual stage. Using VIEW, a software program still being tested, Scott will be doing 3D molecular visualizations of proteins to study their structure and sort out their subsets. A third project, again involving visualization, will study aligned sets of structures in proteins. Soon he hopes to move code to the CONVEX C3880.

``Being here,'' says Scott,'' has solidified my goals to go to graduate school and pursue academic research'' as a career. His special interest is the knowledge-based approach to machine learning.

While at NCSA, he has ``gotten some neat ideas about what kinds of problems biologists studying protein structures need to solve.'' Scott looks forward to continuing his research via Internet after he returns to Virginia.

__________________________________________________________________

access / Spring 1993 / NCSA / pubs@ncsa.uiuc.edu