Ying Wai Li presents work on bottlenecks with Monte Carlo algorithms

In addition to the opportunity to present her own findings, Li emphasizes that conferences such as PASC and CCP are a valuable opportunity to meet peers and take a big-picture look at the state of the field
Photo courtesy of the Swiss National Supercomputing Centre

Ying Wai Li’s research has provided valuable insight into parallel computing that aids in reducing computing time on systems such as Titan, America’s most powerful supercomputer.

Li, a research and development computational scientist at the Oak Ridge Leadership Computing Facility (OLCF) at the US Department of Energy’s (DOE’s) Oak Ridge National Laboratory (ORNL), has identified and resolved bottlenecks that occur within Monte Carlo algorithms—computational tools that rely on statistical probability to estimate configurations or solutions. By targeting and resolving these computational speed bumps, Li has provided a pathway to perform the computer simulation at least ten times faster.

Li was invited to share her findings at four conferences in 2017, including two this summer: Conference on Computational Physics (CCP) and Platform for Advanced Scientific Computing Conference (PASC17). Through these conferences, Li has different platforms from which to share her research and expertise with her high-performance computing colleagues. At CCP, Li gave a keynote lecture in the conference’s classical statistical mechanics session. Within this keynote address, Li shared the work she has completed in parallel Monte Carlo algorithms since coming to ORNL. In her work, Li has identified and resolved bottlenecks within the algorithm—techniques that are valuable to other researchers wishing to decrease the amount of computing hours and to improve the accuracy of Monte Carlo simulations.

In addition to her keynote presentation at CCP, Li participated in the PASC17 meeting in Switzerland. Li was invited to the conference to present her paper entitled “A Histogram-Free Multicanonical Monte Carlo Algorithm for the Basis Expansion of Density of States.”

“The goal of the paper was to illustrate a breakthrough to perform classical Monte Carlo simulations faster than existing algorithms, while getting an improved accuracy. With these techniques, researchers can simulate the properties of materials with much reduced computing resources,” Li said.

In addition to the opportunity to present her own findings, Li emphasizes that conferences such as PASC and CCP are the greatest opportunity to meet peers and take a big picture look at the state of the field. They are also essential to increase the visibility of her research.

“Attending conferences such as CCP and PASC are important for researchers who need to stay on top of the most recent developments in the field,” Li said. “My experience has been that after hearing my own presentations, people will come to me later to speak about collaborating or simply learn more about my methods and how they can be implemented in their own work.”

The OLCF is a DOE Office of Science User Facility located at ORNL. ORNL is supported by the US Department of Energy’s Office of Science. The single largest supporter of basic research in the physical sciences in the United States, the Office of Science is working to address some of the most pressing challenges of our time. For more information, please visit science.energy.gov.