by Cindy Lee
The Genesis Mission unites DOE National Labs, industry, academia, and more to harness AI for breakthroughs in energy dominance, discovery science, and national security. As part of the mission, the American Science and Security Cloud integrates America’s most advanced high-performance computing systems, scientific experimental facilities, data resources, and production capabilities into a single, coordinated AI-driven discovery system. Researchers at the Advanced Light Source (ALS) have been using this platform to accelerate our science and are contributing to push the limits of what is possible.
The ALS leads a Genesis Mission initiative called SYnergistic Neutron and Photon Science-Intelligence (SYNAPS-I) that has brought together experts from the DOE light and neutron sources to accelerate discovery in energy technologies, chemical and materials manufacturing, agriculture, and medicine. By applying AI tools to neutron and photon science, what used to take researchers months now takes mere minutes. In May, the team demonstrated at the AI+ Expo that this discovery process will continue to improve by using the American Science and Security Cloud.
Dylan McReynolds, data platforms program lead within the ALS Photon Science Computing group, shared more details.
Why is the American Science and Security Cloud necessary?
Oak Ridge National Laboratory is coordinating the American Science and Security Cloud, and they’re building these services to help all of us do AI. “All of us” includes ALS computer scientists, of course, but also our beamline and instrumentation experts and user partners from academia and industry.
Some of the services in the platform address longstanding needs, like transferring files between an ALS beamline and a computing cluster or a user’s home institution. Other services have become necessary as we grow in using AI to accelerate science. For instance, a common model repository and a common data catalog are both central to our research that is ever increasingly interconnected, like a user from the USDA physically scanning samples at the ALS in California, deploying computing tools at the Argonne Leadership Computing Facility in Illinois, and then making the results available to the agricultural research field at large.
What has been the impact of the platform on light and neutron sources?
Some of the tools in the platform are improvements from what we’ve been using. With SYNAPS-I, we developed neural networks that accelerated discovery at the ALS, Advanced Photon Source (APS), Linac Coherent Light Source (LCLS), National Synchrotron Light Source II (NSLS-II), Stanford Synchrotron Radiation Lightsource (SSRL), and the Spallation Neutron Source (SNS). The networks, models, and automated processes we developed at all these facilities previously relied on an external company for file transfer. Now, the Genesis Mission platform has a service for file transfer. We also started using the platform’s job scheduler to kick off reconstruction.
While we were able to do this all before, the American Science cloud provides common and centralized tools that will make it easier to share models and workflows with other facilities while also making the system easier for us to maintain. We’re really excited about using the American Science and Security Cloud “Inference as a Service.” That means that the models we develop will be stored in the cloud and running on DOE HPC facilities. We won’t need to reload them every time we analyze new datasets, saving us so much time and compute power.
Across the SYNAPS-I facilities, we have 17 tomography beamlines. Using American Science and Security Cloud workflows, we’ll be able to improve data processing for hundreds of experiments each year!
Why is it necessary for the ALS to be involved in the American Science and Security Cloud?
Not only does this platform enhance the research we’re doing, we have to provide critical use cases for facilities to use the new features. This is an exciting time to dream up new workflows and services. By running our projects on this platform, we can share important insights into how the workflows are functioning for us and make sure the platform continues to have the services we need the most.
