Bradford Larsen
Tools for Concurrent Computing
Brad Larsen, a philosophy/computer science dual major from Nashua, NH, traveled to Germany to study at Friedrich Alexander University Erlangen-Nuremberg. With funding from an International Research Opportunities Program (IROP) grant, he assisted in the implementation of a new programming language that will allow computers to handle the growing data sets produced by industry and science.
Historically computers have been sequential in operation, processing single instructions at any given instant. But the amount of data associated with modern technologies, such as medical imaging, demands increased computational power. As a result, concurrent computing, or the use of dual-core processors, is becoming the choice way to accomplish this. In moving away from the sequential computing model and toward a more complex concurrent model, we are faced with new challenges in software development. Specifically, more errors are likely and more programming is necessary. Brad sought a solution to these issues by working on high-level abstractions, program analysis, and model checking. The main projects he worked on were the Jackal Distributed Shared Memory (a system for the Java language) and Tapir, a new low-level concurrent language designed to prove the correctness of programs.
"The general idea behind this research is to work around some of the difficulties of using many computers in concert rather than a single computer to solve problems," explains Brad. "This is a relevant issue particularly in scientific computing—climate modeling, for example—but also in mainstream computing. Most consumer computers these days have two or more processors in one box."
After returning to UNH, Brad presented his research at the 2007 IROP Symposium. In 2008, he and his foreign mentor, Ronald Veldema, submitted a paper on their research to the 20th IASTED conference, where they won the Best Paper in Distributed Information and Systems award. The three-day conference in Orlando, Florida, attracted computer scientists from around the world and featured 50+ presentations on parallel and distributed computing and systems.
The computer science building at Friedrich Alexander University
From left to right: Michael Klemm, Ronald Veldema, Brad Larsen, Alexander Dreweke, Michael Philippsen (Veldema and Philippsen, Larsen's mentors, are co-authors of his award-winning paper)