Large-scale computers - so-called cluster machines formed from thousands of computers behaving as one - have become increasingly critical to numerous fields of research in the last several years. The complexity of the machines makes them difficult to use efficiently, so for assistance in using them, scientists often work with a research computing facilitator (RCF).
The demand, however, has led to a shortage of RCFs at many universities. With that in mind, the Yale Center for Research Computing (YCRC) is working with six other institutions to help prepare college students for careers as RCFs. The YCRC, which was recently awarded a National Science Foundation (NSF) grant, will also create a regional network of RCFs to give researchers at smaller universities more access to the kind of computing expertise that they need.
“Many researchers are excellent theoreticians or laboratory scientists, but they may lack the computing expertise that would allow them to take full advantage of advanced computers to accelerate their research,” said Andrew Sherman, a senior research scientist in computer science at Yale, and the principal investigator of the project. “They may know the algorithms they need to use, but not how to configure or tune them to match the features of the machines they’ll be using. RCFs can bridge the gap between researchers and the programs, algorithms, and machines that they need to use to get their science done.”
The project, “CAREERS: Cyberteam to Advance Research and Education in Eastern Regional Schools,” is a collaboration among seven anchor institutions: Yale, Rutgers University, Rensselaer Polytechnic Institute, University of Rhode Island, Penn State University, University of Delaware, and the Massachusetts Green High Performance Computing Center.
The first goal of the project is to expand the career pipeline for RCFs. That involves promoting the field as a potential career to students with an interest in both computers and the sciences. “There are a lot of students who, if they only knew there was a career doing this kind of thing, would jump at the chance. It’s an expanding field, and it’s pretty well paid,” Sherman said. “The project will focus especially on finding potential RCFs at small-to-medium sized schools, where there hasn’t been so much exposure to advanced research computing.”
The program would pair students possessing computing expertise with science researchers who need computational assistance. A mentor (typically an experienced RCF) would also be assigned to instruct each student on how to work with their researcher. That includes teaching communication skills, as well as how to achieve the technical results needed for the particular research project. Sherman said that the program would seek out researchers across the states of Connecticut, Rhode Island, New York, New Jersey, Pennsylvania, and Delaware, especially in smaller colleges and universities that may lack the staff or resources required to support advanced research computing on campus. Students may come from the researcher’s school, or from any other institution in the region.
The second half of the project is “building a model for distributed facilitation.” That is, the anchor institutions will develop a network of RCFs of varying specialties throughout universities in the region. This will allow researchers to connect with RCFs with the right expertise, even if they’re at different schools.
“We would like to make it really easy for a biology researcher at University of New Haven (UNH), for instance, to be matched with an RCF at Wesleyan who happens to be an expert computational biologist,” Sherman said. “Or if a chemist at Southern Connecticut State University needed help, they could talk to an RCF at Penn State who knew a lot about computational chemistry.”
It’s a relatively low-cost way to provide a comprehensive RCF program for a large network of schools, many of which might not be able to justify hiring a local RCF. Sherman noted that such a program could not only help the researchers involved, but it could also help the schools by making it possible for them to attract and retain faculty doing research dependent on advanced computation.