While artificial intelligence (AI) is already part of our daily lives in countless ways — from the facial recognition on our smartphones to e-commerce to a doctor’s ability to make more accurate medical diagnoses — researchers at the U.S. Department of Energy’s (DOE) Argonne National Laboratory are leading efforts to leverage AI to accelerate new and potentially transformative discoveries in science.
“What we are interested in is how we can apply the same advances to scientific problems — to discover things faster, to discover things we could not have previously known,” said Ian Foster, director of Argonne’s Data Science and Learning division. “We believe AI methods can provide humans with very powerful, knowledgeable and imaginative assistance that can accelerate the discovery process.”
Broadly speaking, the term AI refers to a set of ideas and concepts that involve building intelligent artificial systems that can observe, learn and solve problems. An important component of AI is machine learning, which is a data-intensive application of artificial intelligence that allows systems to automatically learn and improve from experience.
“While short-term things might move more slowly than we like, long-term AI will change many aspects of society, even in ways we don’t fully understand yet.” — Ian Foster, director of Argonne’s Data Science and Learning division
“With machine learning and artificial intelligence, the computer plays an active role, which means it gives you its answers, predictions or possibilities that maybe a human didn’t think of,” said Brahim Mustapha, an accelerator physicist in Argonne’s Physics division, who oversees one of the many AI-related projects in development at the laboratory.
“This is all made possible by some significant methodological advances, and also by the fact that we now have much more data and far more powerful computers than we used to,” added Foster. “The techniques we’re using combine all three of those aspects to achieve success.”
Uniquely positioned to take a leadership role
Argonne has been at the forefront of AI research during its emergence in recent years. In 2019, Argonne partnered with DOE’s Oak Ridge and Berkeley national laboratories to host a series of town hall meetings, culminating in an expansive AI for Science Report that explored scientific opportunities and outlined a roadmap for the coming decade. As part of its own labwide AI for Science initiative, Argonne has been building an infrastructure of AI technologies and expertise to accelerate pivotal discoveries in practically all areas of research, from drug design to advanced manufacturing to cosmology.
Currently at the lab there are many AI-focused projects underway, and more still that involve partnerships with other national laboratories — bolstered by a recent $37 million research commitment from DOE. These collaborations within the lab and across the DOE landscape highlight the depth and breadth of Argonne’s capabilities, according to Subramanian Sankaranarayanan, group leader in the Theory and Modeling Group in the Center for Nanoscale Materials (CNM).
“Our staff’s range of expertise, the computing resources, data infrastructure, emerging modalities, our world-class experimental facilities, all of that is uniquely positioning us to take a leadership role in AI and machine learning science,” he said.
Sankaranarayanan is heading a project team involving five different laboratories that is using AI to build a digital environment that allows X-ray experiments or electron microscopy studies to be conducted virtually. The goal, he said, is to provide researchers with an opportunity to create a digital twin that mirrors their own research environment.
“The competition for time at facilities such as the Advanced Photon Source is so great that users would be much better served if they knew beforehand how their experimental controls will impact their expected outcomes,” he said. “They would be much better informed about what they are expecting to see, and then could make the most of their actual research time.”
Nicholas Schwarz, a principal computer scientist and group leader at the Advanced Photon Source (APS) — which is undergoing an upgrade that, when completed, will generate X-rays that are up to 500 times brighter than the current light source — believes that AI and machine learning have the potential to be particularly impactful there.
“The upgrade will generate tremendous amounts of new data that we couldn’t capture before, that we can then train AI models on to make new discoveries,” he explained. “Machine learning is all about data, so the APS is the starting point where we’ll be able to collect the data to then make advances using machine learning and other forms of artificial intelligence.”
Among the projects Schwarz currently supports is one using machine learning to better understand what happens when a material fails, so that a stronger material can be designed in its place. Another initiative, with DOE’s Lawrence Berkeley National Laboratory, is collecting extensive amounts of data to categorize and synthesize new materials for future experiments.
Mathew Cherukara leads the Computational X-ray Science group at the APS, where his team is working on several AI projects that have the potential to revolutionize every aspect of experiments at the facility. Examples include AI-accelerated data analysis so scientists can get real-time feedback during their experiment, AI-enhanced techniques that can image materials beyond hardware limitations, and AI agents that make experimental decisions so scientists can make the most efficient use of their limited time on the APS’s instruments.
“To take the world-leading scientific instruments at Argonne to their full potential, we need to leverage the transformational capabilities provided by AI,” Cherukara said.
Mustapha, from the Physics division, is overseeing an effort to use AI and machine learning to better operate Argonne’s user facilities, starting with the Argonne Tandem Linac Accelerator System (ATLAS). A leading facility for research into the structure of nuclei, ATLAS is undergoing an upgrade of its own over the next few years, which will allow it to simultaneously deliver beams to two experimental stations. Mustapha’s project will also employ AI techniques to reduce the time needed to switch between experiments at these stations.
“An improved ATLAS will be an important resource for the accelerator physics community,” Mustapha said.
His team is using the Argonne Wakefield Accelerator, a premier electron accelerator facility, as a test bed for these AI and machine learning developments for accelerators. These techniques, he said, could benefit both APS and ATLAS, as well as future accelerators worldwide.
Foster is involved in a project that is employing AI methods alongside state-of-the-art resources and expertise at the Argonne Leadership Computing Facility (ALCF) — which soon will include Aurora, the exascale supercomputer due to come online in 2022 — so that scientists can better use the vastly increasing amounts of data produced by the APS and other similar facilities. Aurora’s design supports both AI-powered research and traditional modeling and simulation workloads, giving scientists an unprecedented set of tools to pursue data-driven discoveries.
“People want to be able to steer their experiments in real time, not take the data home, analyze it and figure out what to do next, three months later. For that, you need to be able to process the data almost instantaneously,” he said. “We’re creating a workflow where you collect large amounts of data at a beamline, process it on a supercomputer, and train the machine learning system to deliver the actionable information right back to the beamline.”
Foster added that this was successfully done recently, on a limited scale, to help guide the data collection process for medical researchers using APS to identify COVID-19 protein structures.
The next generation of the way we do science
Though these represent just a few of the many projects underway, they paint a picture of a laboratory that has made AI and machine learning a priority. Mustapha said that these efforts are ultimately undertaken for one simple reason: to help solve intractable problems and benefit society.
“There’s no doubt that AI will accelerate scientific discovery,” he said. “As a result, we can expect to achieve better outcomes and discover new things faster than if humans alone were to do it.”
On a practical level, according to Sankaranarayanan, leveraging AI will allow scientists to learn more from the growing amounts of data available to them and make better predictions about what the data means — which would directly affect our everyday lives.
“For example, can we make a prediction about what the next mutation of the virus that causes COVID-19 is going to look like? The use of AI now makes it possible to search through this giant space and identify the kinds of mutations that may happen,” he said. “This could also apply to research on materials for battery applications, various forms of energy storage or the effects of pollutants on the environment. There are so many areas that will benefit.”
While Foster believes it may take some time for scientists to make the most effective use of AI and machine learning techniques — for example, he noted that many aspects of the current research environment will need to be reimagined, such as automating laboratories so they can be controlled with AI assistance — he left little doubt about the transformative potential of these methods.
“There’s an old saying that people always overestimate the short-term impact of innovations, but underestimate the long-term aspects,” he said. “While short-term things might move more slowly than we like, long-term AI will change many aspects of society, even in ways we don’t fully understand yet.”
Schwarz calls AI and machine learning “the next generation of the way we do science,” adding that Argonne is perfectly suited to lead the way and help shape this future.
“It’s incredibly exciting to be part of this, because what we’re really doing at Argonne is assembling teams of the best and the brightest in this space. We’re breaking away from the old ‘silos’ and we’re bridging all of this together. This is really an unprecedented opportunity for us.”
The APS, ALCF, ATLAS and CNM are DOE Office of Science User Facilities. Funding for the projects discussed in this article was provided by DOE’s Office of Science.
About Argonne’s Center for Nanoscale Materials
The Center for Nanoscale Materials is one of the five DOE Nanoscale Science Research Centers, premier national user facilities for interdisciplinary research at the nanoscale supported by the DOE Office of Science. Together the NSRCs comprise a suite of complementary facilities that provide researchers with state-of-the-art capabilities to fabricate, process, characterize and model nanoscale materials, and constitute the largest infrastructure investment of the National Nanotechnology Initiative. The NSRCs are located at DOE’s Argonne, Brookhaven, Lawrence Berkeley, Oak Ridge, Sandia and Los Alamos National Laboratories. For more information about the DOE NSRCs, please visit https://science.osti.gov/User-Facilities/User-Facilities-at-a-Glance.
The Argonne Leadership Computing Facility provides supercomputing capabilities to the scientific and engineering community to advance fundamental discovery and understanding in a broad range of disciplines. Supported by the U.S. Department of Energy’s (DOE’s) Office of Science, Advanced Scientific Computing Research (ASCR) program, the ALCF is one of two DOE Leadership Computing Facilities in the nation dedicated to open science.
About the Advanced Photon Source
The U. S. Department of Energy Office of Science’s Advanced Photon Source (APS) at Argonne National Laboratory is one of the world’s most productive X-ray light source facilities. The APS provides high-brightness X-ray beams to a diverse community of researchers in materials science, chemistry, condensed matter physics, the life and environmental sciences, and applied research. These X-rays are ideally suited for explorations of materials and biological structures; elemental distribution; chemical, magnetic, electronic states; and a wide range of technologically important engineering systems from batteries to fuel injector sprays, all of which are the foundations of our nation’s economic, technological, and physical well-being. Each year, more than 5,000 researchers use the APS to produce over 2,000 publications detailing impactful discoveries, and solve more vital biological protein structures than users of any other X-ray light source research facility. APS scientists and engineers innovate technology that is at the heart of advancing accelerator and light-source operations. This includes the insertion devices that produce extreme-brightness X-rays prized by researchers, lenses that focus the X-rays down to a few nanometers, instrumentation that maximizes the way the X-rays interact with samples being studied, and software that gathers and manages the massive quantity of data resulting from discovery research at the APS.
This research used resources of the Advanced Photon Source, a U.S. DOE Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under Contract No. DE-AC02-06CH11357.
Argonne National Laboratory seeks solutions to pressing national problems in science and technology. The nation’s first national laboratory, Argonne conducts leading-edge basic and applied scientific research in virtually every scientific discipline. Argonne researchers work closely with researchers from hundreds of companies, universities, and federal, state and municipal agencies to help them solve their specific problems, advance America’s scientific leadership and prepare the nation for a better future. With employees from more than 60 nations, Argonne is managed by UChicago Argonne, LLC for the U.S. Department of Energy’s Office of Science.
The U.S. Department of Energy’s Office of Science is the single largest supporter of basic research in the physical sciences in the United States and is working to address some of the most pressing challenges of our time. For more information, visit https://energy.gov/science.