James L. Carmon Scholarship Award 2011
Junqi Yin, a PhD candidate in physics, works on computer simulations of model systems in condensed matter physics. The simulations use algorithms to drive graphics processors to solve problems that can’t be addressed analytically. This enables scientists to study biophysical interactions at extremely small scales. The work has long-range implications in fields as diverse as thermodynamics and biophysics.
Yin’s algorithms and codes take advantage of the platform used in new graphic processing units (GPUs), powerful new tools for supercomputing. Yin’s code, running on a single GPU, outperforms conventional codes of the type running on a 32-CPU cluster by an order of magnitude. Yin’s “Wang-Landau sampling code” provides linear scalability, which extends to large applications. The sheer speed opens up whole new possibilities to run regimes previously considered intractable because of their size and complexity. In fact, Yin’s code has already been used to resolve a longstanding controversy about a specific type of magnetic phase transition.