Wednesday, January 25
Center for Research at Extreme Scales, Indiana U.
High Performance Computing in Physics or: How I learned to stop worrying and love Exascale
The field of High Performance Computing has delivered explosive computational performance improvements for physics research beginning from a thousand basic operations per second in the 1940s to performance in excess of 100 quadrillion floating point operations per second (100 Petaflops) today. During the span of a human lifetime, supercomputer speeds have improved by a staggering 14 orders of magnitude (100 trillion times improvement). But the true value of these supercomputing systems is their service to computational science as the “third pillar of science” complementing traditional theoretical and empirical methods. In fact, many discoveries credited to empirical tools such as particle accelerators or interferometers also required the use of supercomputers in order to produce their results.
Within as little as five years and quite possibly less, the Exaflops barrier will be reached by several leadership class supercomputers enabling computational research in physics at the Exascale level. This will bring enormous opportunities. But practical Exascale computing presents multiple challenges both for physics application developers and hardware architecture designers. The ratio of memory to flops continues to drop in the most recently deployed leadership machines, and system power requirements remain very high at the same time as increases in component heterogeneity continue to complicate machine programmability. Because of their complexity, many supercomputer systems are swamped with codes that are poorly matched to the hardware on which they execute and written in ways that are not well suited for the scientific domain from which they originate.
This colloquium will explore some of the challenges facing scientists on current and emerging HPC architectures both at the leadership and mainstream scales, and will discuss a wide range of solutions ranging from domain specific languages, technologies including field programmable gate arrays, and run-time system software.
Wednesday, February 1
Mechanical Engineering, Yonsei University
Modeling of blast waves: introducing “Hugonions”
Wednesday, February 8
Space Vehicles Directorate, Air Force Research Lab
Optical atomic clocks: ultra-precise instruments for science and technology
decimal place. Conventional atomic clocks probe a hyperfine transition in alkali atoms, such as the 9.2 GHz transition in cesium that is used to define and realized the SI second. Over the past 15 years, a superior technology has been developed and refined based on laser spectroscopy of atoms and ions; here, the higher clock frequency (hundreds of THz) leads to improved precision and accuracy. In this talk, I will describe the principles of operation of state-of-the-art research-grade optical clocks, the underlying physics that drives these instruments to such high performance, and a few of the scientific and technological applications that will benefit from these devices. thAtomic clocks play a vital role in many technologies including communications and navigation, and the best atomic clocks contribute to international atomic time with inaccuracy at the 16
Wednesday, February 15
Physics & Astronomy
3-Minute Thesis Competition
Wednesday, February 22
X: Self-Driving Car Project
Lidar for self-driving cars
Cars have evolved
slowly for over a century, becoming increasingly safe and convenient. Recently,
developments in self-driving technology may herald a rare step-function in the
transportation industry. At the simplest level, moving control from human to
machine could result in fewer accidents and less time wasted driving, and it
empowers the disabled and those too young or old to drive. More fundamentally
though, it can change how we spend our time, where we live, how we build
cities, how we work, and even how we think.
Waymo (formerly the Google
Self-driving Car Project) has been building and testing self-driving cars for
several years, and has driven more than 2 million autonomous miles. I’m on the
team that is developing Waymo’s next-generation lidar system. I will discuss
the current and future state of self-driving technology generally, then dive a
bit deeper into the lidar system as an example of physics in industry.
Wednesday, March 1
Honeywell Aerospace Advanced Technology
Wednesday, March 8
Elizabeth Jeffery Kraczek
Physics & Astronomy, Brigham Young University
Wednesday, March 15
Wednesday, March 22
Mechanical Engineering, Brigham Young University
Wednesday, March 29
University of Texas Arlington
Wednesday, April 5
Physics & Astronomy, Brigham Young University
We welcome anyone who wish to attend, and typically serve refreshments
ten minutes before the colloquium begins. Speakers generally keep
their presentation accessible to undergraduate physics students.