Nobel Prize Celebrates Interplay of Physics and AI

On October 8, the Nobel Prize in physics was awarded to John Hopfield and Geoffrey E. Hinton for their foundational discoveries and inventions that have enabled artificial neural networks to be used for machine learning—a widely used form of AI. The award highlights how the field of physics is intertwined with neural networks and the field of AI.(Credit: © Johan Jarnestad/The Royal Swedish Academy of Sciences)(Credit: © Johan Jarnestad/The Royal Swedish Academy of Sciences)

An artificial neural network is a collection of nodes that connect in a way inspired by neurons firing in a living brain. The connections allow a network to store and manipulate information. While neural network research is closely tied to the fields of neuroscience and computer science, it is also connected with physics. Ideas and tools from physics were integral in the development of neural networks for machine learning tasks. And once machine learning was refined into a powerful tool, physicists across the world, including at the University of Maryland, have been deploying it in diverse research efforts.

Hopfield invented a neural network, called the Hopfield network, that can work as a memory that stores patterns in data—this can be used for tasks like recognizing patterns in images. Each node in its network can be described as a pixel in an image, and it can be used to find an image in its memory from its training that most closely resembles a new image that it is presented to it. But the process used to compare images can also be described in terms of the physics that govern the quantum property of spin.

The spin of a quantum particle makes it behave like a tiny bar magnet, and when many magnets are near each other, they all work to orient themselves in specific ways (north poles repelled by the other north poles and attracted to south poles). Physicists can characterize a group of spins based on their interactions and the energy associated with the orientation all the spins want to fall into. Similarly, a Hopfield network can be described as characterizing images based on an energy defined by the connections between nodes.

Hinton used the tools of statistical physics to build on Hopfield’s work. He developed an approach to using neural networks called Boltzmann machines. The learning method of these neural networks fit the description of a specific type of spin orientation found in materials, called a spin glass. A Boltzmann machine can be used to identify characteristics of the data it was trained with. These networks can help classify images or create new elements that fit the pattern it has been trained to recognize.

Following the initial work of Hopefield and Hinton a broad variety of neural networks and machine learning applications have arisen. As neural networks have expanded beyond the forms developed by Hopfield and Hinton, they still resemble common physics models, and some physicists have chosen to apply their skills to understanding the large, often messy, models that describe neural networks.

“Artificial neural networks represent a very complicated, many-body problem, and physicists, especially condensed matter physicists, that's what we do,” says JQI Fellow Maissam Barkeshli, a theoretical physicist who has applied his expertise to studying artificial neural networks. “We study complex systems, and then we try to tease out interesting, qualitatively robust behavior. So neural network research is really within the purview of physicists.”

In a paper that Barkeshli and UMD graduate student Dayal Kalra shared at the Conference on Neural Information Processing Systems last year, the pair presented the results of their investigation of the impact of the learning rate—the size of steps that are made each time the network’s parameters are changed during training—on the optimization of a neural network. Their analysis predicted distinct behaviors for the neural network depending on the learning rate used.

“Our work focused on documenting and explaining some intriguing phenomena that we observed as we tuned the parameters of the training algorithm of a neural network,” Barkeshli says. “It is important to understand these phenomena because they deeply affect the ability of neural networks to learn complicated patterns in the data.”

Other similar questions in neural network research remain, including how the information is encoded in a network, and Barkeshli says that many of the open questions are likely to benefit from the perspectives and tools of physicists.

In addition to the field of machine learning benefitting from the tools of physics, it has also provided valuable tools for physicists to use in their research. Similar to the diverse uses of AI to play board games, create quirky images and write emails, the applications of machine learning have taken many forms in physics research.

“The laureates’ work has already been of the greatest benefit,” says Ellen Moons, Chair of the Nobel Committee for Physics. “In physics we use artificial neural networks in a vast range of areas, such as developing new materials with specific properties.”

Neural networks can also be useful for physicists who need to identify relevant data so they can focus their attention effectively. For instance, the scientific background document that was shared as part of the prize announcement cites the use of neural networks in the analysis of data from the IceCube neutrino detector at the South Pole. The project was a collaboration of many researchers, including UMD physics professors Kara Hoffman and Gregory Sullivan and UMD physics assistant professor Brian Clark, that produced the first neutrino image of the Milky Way.

Neutrinos are a group of subatomic particles that are notoriously difficult to detect since they have little mass and no electrical charge, which makes interactions uncommon (only about one out of 10 billion neutrinos is expected to interact as it travels all the way through the earth). The lack of interactions means neutrinos can be used to observe parts of the universe where light and other signals have been blocked or deflected, but it also means researchers have to work hard to observe them. The IceCube detector includes a cubic kilometer of ice that neutrinos can interact with. When an interaction is observed, researchers must determine if it was actually an interaction with a neutrino interaction or another particle. They must also determine the direction the detected neutrino came from and whether it likely originated from a distant source or if it was produced by particle interactions that occurred in the earth’s atmosphere.

UMD postdoctoral researcher Stephen Sclafani worked on the project as a graduate student at Drexel University. He was an author of the paper sharing the results and was a lead in the project’s use of machine learning in their analysis. Neural networks helped Sclafani and his colleagues select the desired neutron interactions observed by the detector from data that had been collected over ten years. The approach increased their efficiency at identifying relevant events and provided them with approximately 30 times the amount of data to use in generating their neutrino map of our galaxy.

“Initially, as many people were, I was surprised by this year's prize selection,” says Sclafani. “But neural networks are currently revolutionizing the way we do physics. The breakthroughs of Hopfield and Hinton are responsible for many other results and are grounded in statistical physics.”

Some physics research applications of neural networks draw more directly on the physics foundation of neural networks. Earlier this year, JQI Fellow Charles Clark and JQI graduate student Ruizhi Pan proposed new tools to expand the use of machine learning in quantum physics research. They investigated a type of neural network called a restricted Boltzmann machine (RBM)—a variation of Boltzmann machines with additional restrictions on the networks. Their research returned to the spin description of the network and investigated how well various numbers of nodes can do at approximating the state that results from the spin interactions of many quantum particles.

"We, and many others, thought that the RBM framework, applied by 2024 Nobel Physics Laureate Geoffrey Hinton to fast learning algorithms about 20 years ago, might offer advantages for solving problems of quantum spin systems," says JQI Fellow Charles Clark. "This proved to be the case, and the research on quantum models using the RBM framework is an example of how advances in mathematics can lead to unanticipated developments in the understanding of physics, as was the case for calculus, linear algebra, and the theory of Hilbert spaces."

There are numerous additional ways that neural networks are also being developed into valuable tools for physics research, and this year’s Nobel Prize in physics celebrates that contribution as well as its roots in the field.

For more information about the prize winners and their research that the award recognizes, see the press releases from the Royal Swedish Academy of Sciences.

Original story by Bailey Bedfod:  jqi.umd.edu/news/nobel-prize-celebrates-interplay-physics-and-ai

Related news stories:

https://jqi.umd.edu/news/attacking-quantum-models-ai-when-can-truncated-neural-networks-deliver-results

https://jqi.umd.edu/news/neural-networks-take-quantum-entanglement

 

Chacko Elected APS Fellow

Professor Zackaria Chacko has been elected Fellows of the American Physical Society. APS Fellowship recognizes excellence in physics and exceptional service to the physics community.

Chacko, who is a member of the Maryland Center for Fundamental Physics (MCFP), was cited for discovering two of the major theoretical scenarios for particle physics beyond the Standard Model — neutral naturalness and gaugino mediated supersymmetry breaking — and for inspiring experimental programs to test them.

Following his B.S. and M.S. degrees at the Indian Institute of Technology, Kharagpur, Chacko earned his Ph.D. in Physics at the University of Maryland in 1999, working with Markus Luty. While a graduate student at Maryland, he received the Michael J. Pelczar Award for Excellence in Graduate Study and was named an Outstanding Teaching Assistant.

He then held postdoctoral positions at the University of Washington and the University of California, Berkeley, before accepting the role of Assistant Professor at the University of Arizona. In 2007, he returned to the University of Maryland. He was promoted to Associate Professor in 2009 and full Professor in 2016. He has served on the department’s Priorities, Curriculum Review, and Graduate Admissions committees, and is currently a member of the editorial board of JHEP, the premier journal dedicated to elementary particle physics.

The primary focus of Chacko’s research is on proposing new theories that address the known problems of the Standard Model of particle physics which can be tested in current and future experiments. The work for which he received the award is related to a theoretical problem of the Standard Model, known as the “hierarchy problem”. The Higgs boson has a mass of order the weak scale, the mass scale of the force carriers of the weak interactions. However, in the Standard Model, quantum effects tend to make the Higgs many orders of magnitude heavier than the observed value. The fact that the Higgs is light then arises from a very delicate cancellation between completely independent effects, which seems extremely contrived.

An elegant class of solutions to the hierarchy problem involve extending the Standard Model to include new particles related to the known particles by a new symmetry of nature. The quantum effects of the new particles cancel against those of the Standard Model, explaining the lightness of the Higgs boson. Chacko was recognized for proposing two paradigms that realize this framework, gaugino mediated supersymmetry breaking and neutral naturalness, which have been enormously influential in the field and inspired novel experimental searches to discover them.    

“Chacko’s APS Fellowship highlights his highly original and influential proposals to solve one of the deepest mysteries of particle physics, the Hierarchy Problem,” said Raman Sundrum, Director of the MCFP. “This distinction is richly deserved.”

 

High Altitude Water Cherenkov Observatory Sheds Light on Origin of Galactic Cosmic Rays

HAWC observes Ultra-High Energy gamma rays confirming Galactic Center as a source of Ultra-High Energy cosmic ray protons in the Milky Way

The High-Altitude Water Cherenkov (HAWC) Observatory, located on the slopes of the Sierra Negra volcano in Mexico, has achieved a groundbreakingHAWC by Jordan GoodmanHAWC by Jordan Goodman milestone with the first detection of gamma rays exceeding 100 TeV from the Galactic Center. This provides strong evidence for the existence of a PeVatron—a source capable of accelerating particles to energies of up to petaelectronvolts (PeV), which is over one hundred times the energy achieved by particle accelerators on Earth. PeVatrons have long intrigued astrophysicists due to their role in high-energy cosmic particle acceleration. While magnetic fields in space deflect charged particles, making it difficult to pinpoint their origin, gamma rays offer a direct view into these extreme acceleration processes, shedding light on their origins within our Galaxy.

The figure shows the best-fit spectrum of the source detected by HAWC, and the resulting spectrum after subtracting two known point sources that are coincident with ours. This resulting spectrum corresponds to the diffuse emission from the Galactic Center and it shows that it extends without evidence of a cutoff to over 100 TeV.The figure shows the best-fit spectrum of the source detected by HAWC, and the resulting spectrum after subtracting two known point sources that are coincident with ours. This resulting spectrum corresponds to the diffuse emission from the Galactic Center and it shows that it extends without evidence of a cutoff to over 100 TeV.The center of our Galaxy hosts a range of remarkable astrophysical objects, including Sagittarius A*, a supermassive black hole with a mass approximately four million times that of the Sun. It is surrounded by neutron stars, white dwarfs stripping material from nearby stars, and extremely hot, dense gas clouds with temperatures reaching millions of degrees. These environments provide ideal conditions for the interaction of PeV protons, freshly accelerated by the suspected PeVatron, with protons from the surrounding matter. These  interactions produce neutral pions, which quickly decay into gamma rays, contributing to the observed photon spectrum between 6 and 114 TeV. The lack of a spectral cutoff strongly suggests a hadronic origin for these gamma rays. Furthermore, the short escape time of the PeV protons suggests the need for a quasi-continuous injection of particles into the gas to maintain the observed gamma-ray production.

The dense interstellar gas between Earth and the Galactic Center obscures this intriguing region from optical observation. Thus, the findings from the HAWC Gamma-Ray observatory provide valuable insights into the high-energy processes occurring at the core of our Galaxy, shedding light on the origin of Galactic cosmic rays.

The particle astrophysics group at UMD plays an important role in the operations of the HAWC Observatory. This particular study was led by Sohyoun Yun-Cárcamo (Ph.D. candidate), Dezhi Huang (postdoc), and Jason Fan (former UMD Ph.D. student). Other UMD authors of this paper are Jordan Goodman, Andrew Smith, Kristi Engel, Elijah Willox, and Zhen Wang.

Publication: https://iopscience.iop.org/article/10.3847/2041-8213/ad772e

William Douglass Dorland, 1965-2024

Bill Dorland, an esteemed plasma and computational physicist who last week received the American Physical Society’s James Clerk Maxwell Prize, has died at age 58. Since a 2004 diagnosis of chordoma, a rare cancer affecting the spine, he optimistically pursued emerging therapies while advocating for the chordoma community, engaging in continued physics research and serving as a superb mentor and teacher.

After completing his undergraduate studies at the University of Texas (and winning the campus foosball tournament), Dorland earned both a Ph.D. in astrophysical sciences and a Master’s degree in public affairs at Princeton University. He returned to Texas, working at the Institute for Fusion Studies, before joining the University of Maryland in 1998 when his wife, Sarah C. Penniston-Dorland, accepted a fellowship at Johns Hopkins University.

Early in his career, Dorland’s calculations revealed that an international plan to build a gigantic fusion reactor was based on flawed science, thereby saving $10 billion and preventing a probable scientific debacle.

His work modeling plasma turbulence merited the prestigious E. O. Lawrence Medal of the Department of Energy. A Diamondback profile described Dorland’s reluctance to leave his class for a call from “the secretary”, who turned out to be Secretary of Energy Steven Chu relaying news of the award and its $50,000 honorarium.

During his career, Dorland held appointments at the University of Vienna, the University of Oxford and Imperial College, London.  From 2020-23, he served as associate laboratory director for Computational Science at the U.S. Department of Energy's Princeton Plasma Physics Lab, which is managed by Princeton University.

Dorland studied in Japan during high school, and found the experience invaluable and insightful. Arriving at the University of Texas, he was shocked by paucity of such opportunities, and launched a vigorous campaign to direct a fraction of student fees toward international exchanges. The number of students studying abroad from UT grew from eight his freshmen year to more than a thousand four years later. In 2000, he received a special award by the Council on International Education Exchange.

At UMD, he co-developed new curricula, including Physics for Decision Makers: The Global Energy Crisis, a Marquee course to instruct non-science majors in perhaps the world’s most pressing challenge. He was a remarkable mentor; three of Dorland’s undergraduate advisees have received the University Medal.  Twice he officiated the weddings of UMD graduate students.

When his chordoma diagnosis prompted an assessment of his life and priorities, he sought the role of director of the UMD Honors College, recalling his own transformative experience as a UT undergrad. For seven years, he advocated for new programs and encouraged study abroad experiences. In a Maryland Today  article during that time, he described continuing his work through his tortuous medical odyssey with the support of his wife, a professor in the Department of geology, and his daughter Kendall.  The family asks that those interested in commemorating Bill do so with a donation to the Chordoma Foundation.

In 2010, Dorland was named a UMD Distinguished Scholar-Teacher (DST). In a letter supporting the nomination, one student described Dorland as “the sort of genius who, while always impressive, is never intimidating….His cheerful encouragement, quirky sense of humor, and constant support were what kept me going in graduate school, even when finishing the dissertation seemed like an impossible goal.”

In his own DST essay, Dorland wrote that after his diagnosis, “I had occasion to reconsider all the decisions I had made in life, and to adjust my trajectory accordingly for the time remaining. I spent a few weeks thinking hard, and was extraordinarily happy to find that I was already doing exactly that which gives me the most satisfaction.”

He concluded: “I generally work as hard as I can to challenge the best students at the University of Maryland to perform at their very best level. This is my mission. It is nothing more than teaching, research, and love.”