LMU Computer Science Professor Awarded NSF Grant for Research on Task Scheduling Algorithms

Jared Coleman, assistant professor of computer science at Loyola Marymount University, wants to help other researchers - from astrophysicists to geneticists - use their large sets of data and algorithms to solve real-world scientific problems. Thanks to a $150,000 grant from the National Science Foundation (NSF), he's currently working on this research for a project titled "Novel Approaches for Task Graph Scheduling Algorithm Design, Evaluation, and Comparison."

Coleman's research addresses a decades-old problem in distributed computing called task scheduling. Task scheduling algorithms are a way computer scientists tell a cluster of computers, servers, and networks how and when to map dependent tasks to process large sets of data to solve one common problem or application.

"Over the last decade, so many algorithms have been proposed that it's become hard for domain scientists to know which algorithms will perform well or perform poorly for their specific applications," said Coleman. "Through this research, we are working on a software library that is all about making it easier for people to understand which algorithms will work well for a particular domain-researcher's task application."

Coleman's goal is to create a framework for comparing task scheduling algorithms to help other researchers make decisions. Another piece of his work is integrating the open-source task scheduling software library (called SAGA) with popular and widely used distributed computing tools and platforms like WfCommons, Pegasus, and Chameleon Cloud. This will allow other researchers easy access to the tool in their day-to-day work.

This two-year NSF grant is also significant because it focuses specifically on providing undergraduate students with research experience. "It's usually graduate students doing research, so it's really satisfying to see undergrads involved in this high-level computer science research - while also giving them the opportunity to see what computer science research in academia can look like," Coleman said.

Computer science research and working in academia can be hard for a lot of young students to imagine or see themselves doing. "When I was an undergrad, I had no idea what research in computer science looked like," Coleman said. "You typically think of lab coats and test tubes when someone says research but research in chemistry and computer science look very different."

For this research, Coleman's undergrads are programming, coding, and running experiments - and then presenting empirical results on those experiments. In fact, one of his students, computer science major Jason Chamorro '27, presented his first paper on their research at a super computing conference in November. "He did a great job!" Coleman said.

Coleman has a team of five undergraduate computer science majors assisting with the project. As they continue their work, they are already making strides and discovering compelling trends in their research.

"We've found that while a lot of the existing scheduling algorithms were designed for specific applications, they can be adapted in small ways to work for other, very different, problems and can work very well," he said. "This has been an interesting theme that has emerged."

Coleman and his students will continue to work on extending the software library, publishing papers on their findings, and presenting their results at conferences. His goal is to keep expanding the software library over time.