Oak Ridge National Laboratory is testing a ‘deep-learning supercomputer in a box’ as it looks ahead to machines that automatically find insights in data.
DOE researchers design algorithms that model a million atoms and probe the detailed chemical behavior of materials.
Sandia researcher taps high-performance computers to design metamaterials that mold electromagnetics.
Data from experiments and advanced codes combine with supercomputing muscle to help explain a half-century-old mystery.
Developing practical fusion energy has been impeded for decades by high heat loss from magnetically confined plasmas. Researchers from MIT, University of California at San Diego and General Atomics captured the dynamics of plasma turbulence linked to heat loss on unprecedented scales. Pictured here is the inside of the Alcator C-Mod tokamak, the inset depicting plasma turbulence simulations that show long wavelength blobs coexisting with short wavelength streamers – small, finger-like structures that comprise the turbulence in the core of the experimental plasmas.View full highlight »