Building the Roads for Self-Driving Scientific Facilities

W&M Physics Colloquium
Friday, April 30, 2021
4:00 PM

Amber Boehnlein [Host: J. Stevens]
Computational Science and Technology Division - Thomas Jefferson National Accelerator Facility
“Building the Roads for Self-Driving Scientific Facilities”

Abstract:
Addressing fundamental questions across the physical sciences motivates new experimental facilities and methods of theoretical inquiry. Major investments in experiments often require novel approaches in computing and software. An example is the development of the distributed grid computing model to meet the needs of the ATLAS and CMS detectors at the Large Hadron Collider. That spirit of innovation is being applied today to make use of technical advances in computing hardware and to modernize computing models for the current and upcoming generation of experiments. The once sharp line between online and offline computing tasks is being blurred by application of technologies such as streaming readout and machine learning.

Beyond that classic ‘operational’ approach to scientific computing, the DOE Exascale Computing Project, combined with advances in data science, has ushered in a new era of possibilities, including real time feedback that could be used to steer data collection, and setting up automated process ‘agents’ that could enable scientists to focus more on research.

At Jefferson Lab, we are organizing our computing and software efforts towards this vision within the context of the 12 GeV science program. This talk will cover the current Jefferson Lab activities and prospects for the future.