The UC Davis Complexity Sciences Center was recognized for its achievements in high-performance computing technologies with an HPC Innovation Excellence Award from Hyperion Research. The award was announced June 18 at the ISC19 supercomputer industry conference in Frankfurt, Germany.
Additional winners for 2019 include the California Institute of Technology, Lawrence Livermore National Laboratory, and the University of Illinois at Urbana-Champaign.
The Complexity Sciences Center is led by James Crutchfield, professor of physics in the UC Davis College of Letters and Science. Crutchfield and graduate student Adam Rupe used the Cori supercomputer to develop Project DisCo, the first distributed high-performance computing application of a physics-based, data-driven technique called “local causal states.” This technique offers an alternative to deep learning for the goal of scientific discovery from data.
The UC Davis team tested the technique on climate data from the CAM5.1 global atmospheric model, processing almost 90 terabytes of data in less than seven minutes on the Cori supercomputer.
New data-driven methods are required that discover and mathematically describe complex emergent phenomena, uncover the physical and causal mechanisms underlying these phenomena, and are better able to predict these phenomena and how they evolve over time. The local causal states have the potential to do exactly this, and to do so directly from unlabeled data. Tools like the DisCo implementation of local causal states will allow scientists to ask complicated questions of their data sets without knowing all the technical details of how to properly answer these questions.
— Adapted from a news release from Hyperion Research.