Argonne National Laboratory played host to the Computational Research Leadership Council (CRLC) meeting on September 4–5, 2019. The council consists of the computational leadership at the eight U.S. Department of Energy (DOE) multi-program national laboratories: Argonne, Brookhaven, Lawrence Berkeley, Lawrence Livermore, Los Alamos, Oak Ridge, Pacific Northwest, and Sandia.
The CRLC quarterly meetings were initiated in August 2017 to provide a forum for in-depth discussions focused around collaborative opportunities among the national laboratories. Each quarter the host venue changes.
The first day of the September 2019 meeting began with a welcome by Mike Papka, deputy associate laboratory director of Computing, Environment and Life Sciences (CELS) and director of the Argonne Leadership Computing Facility (ALCF), and with an overview of computing at Argonne, presented by the meeting host Valerie Taylor, an Argonne Distinguished Fellow and director of the Mathematics and Computer Science (MCS) Division. Researchers from MCS, Data Science and Learning (DSL), Computational Science (CPS), and ALCF then presented technical talks on optimization, statistics, and machine learning and on systems advances. A recurring theme was the challenges the research teams are addressing — including multimodal data analysis, dependencies in space and time, and integration of machine learning techniques. In each case, the researchers gave specific examples of the work being done at Argonne to address these challenges: the Waggle platform for artificial intelligence at the edge, the Chameleon cloud testbed, the Swift/T language and runtime for high-performance computing (HPC) workflows, and cyber infrastructure for autonomous science.
Following a working lunch, attendees toured the Materials Engineering Research Facility and the Argonne Tandem Linac Accelerator System. The afternoon concluded with a meeting among the CRLC members, as well as individual meetings with Steven Lee and Robinson Pino from DOE’s Office of Advanced Scientific Computing Research (ASCR).
The second day of the workshop began with a meeting of ASCR and CRLC leaders. The remainder of the morning was devoted to technical talks. Three CELS researchers involved in quantum computing sciences discussed their work in projects ranging from hybrid quantum-classical architectures, networks, and algorithms to highly scalable HPC quantum simulations. Concluding the morning were presentations on computational methods for exascale computing — in particular, the EZ and SZ lossy compressors being used in hurricane and cosmology simulations and OpenMC, a collaborative effort to provide a capability for coupled Monte Carlo particle transport calculations at exascale.
After a working lunch, attendees toured the Center for Nanoscale Materials and the Advanced Photon Source.
“These in-person quarterly CRLC meetings enable DOE and the national laboratories to share their experiences, keep abreast of exciting new research, and view the various facilities available at the labs” said Taylor. “They also provide an opportunity to identify areas for future collaborations; for example, in domains ranging from quantum research to data analytics.”