Twisted Light Gives Electrons a Spinning Kick

Details

Published: Saturday, November 30 2024 01:00

It’s hard to tell when you’re catching some rays at the beach, but light packs a punch. Not only does a beam of light carry energy, it can also carry momentum. This includes linear momentum, which is what makes a speeding train hard to stop, and orbital angular momentum, which is what the earth carries as it revolves around the sun.

In a new paper, scientists seeking better methods for controlling the quantum interactions between light and matter demonstrated a novel way to use light to give electrons a spinning kick. They reported the results of their experiment, which shows that a light beam can reliably transfer orbital angular momentum to itinerant electrons in graphene, on Nov. 26, 2024, in the journal Nature Photonics.

Having tight control over the way that light and matter interact is an essential requirement for applications like quantum computing or quantum sensing. In particular, scientists have been interested in coaxing electrons to respond to some of the more exotic shapes that light beams can assume. For example, light carrying orbital angular momentum swirls around its axis as it travels. When viewed head-on, a light beam with orbital angular momentum contains a dark spot in the middle, a vortex opened up by the beam’s corkscrew character.In a new experiment, light beams carrying orbital angular momentum caused electrons in graphene to gain (blue beam) and lose (red beam) angular momentum, transporting them across the sample and generating a current that researchers measured. (Credit: Mahmoud Jalali Mehrabad/JQI)

“The interaction of light that has orbital angular momentum with matter has been thought about since the 90s or so,” says Deric Session, a postdoctoral researcher at JQI and the University of Maryland (UMD) who is the lead author of the new paper. “But there have been very few experiments actually demonstrating the transfer.”

Part of the challenge has been a size mismatch. In order for electrons to feel a tug from a light beam carrying momentum, they have to experience the way that the beam changes as it passes by. In many cases, the length over which a light beam changes dwarfs the size of the matter that researchers are interested in manipulating, making it especially challenging to pick out electrons as targets.

For instance, atoms and their orbiting electrons—mainstays of quantum physics experiments and favorite targets for precise manipulation—are roughly 1,000 times smaller than the light beams that researchers use to interact with them. Light travels as repeating waves of electric and magnetic fields, and the length that a light beam travels before it repeats is called the wavelength. In addition to being an important characteristic size, the wavelength of light also determines the amount of energy carried by individual particles of light called photons. Only photons that carry particular amounts of energy can interact with atoms, and those photons tend to have wavelengths much bigger than the atoms themselves. So while atoms as a whole will happily absorb energy and momentum from these photons, the wavelength is too big for the internal pieces of the atom—the nucleus and the electrons—to notice any relative difference. This makes it very difficult to transfer orbital angular momentum solely to an atom’s electrons.

One way to overcome this difficulty is to shrink the wavelength of light. But that increases the energy carried by each photon, ruling out atoms as reliable targets. In the new experiment, the researchers, who included JQI Fellows Mohammad Hafezi and Nathan Schine, JQI Co-Director Jay Sau and JQI Adjunct Fellow Glenn Solomon, pursued an alternate approach: Instead of shrinking the wavelength of the light, they puffed up electrons to make them occupy more space.

Electrons bound to the nucleus of an atom can only roam so far before they are liberated from the atom and useless for experiments. But in conductive materials, electrons have more latitude to travel far and wide while remaining under control. The researchers turned to graphene, a flat material that is one of the best known electrical conductors, in search of a way to make electrons take up more space.

By cooling a sample of graphene down to just 4 degrees above absolute zero and subjecting it to a strong magnetic field, electrons that are ordinarily free to move around become trapped in loops called cyclotron orbits. As the field gets stronger, the orbits become tighter and tighter until many circulating electrons are packed in so tightly that no more can fit. Although the orbits are tight, they are still much larger than the electron orbitals in atoms—the perfect recipe for getting them to notice light carrying orbital angular momentum.

The researchers used a sample of graphene wired up with electrodes for their experiments. One electrode was in the middle of the sample, and another made a ring around the outer edges. Earlier theoretical work, developed in 2021 by former JQI and UMD graduate student Bin Cao and three other authors of the new paper, suggested that electrons circulating in such a sample could gain angular momentum in chunks from incoming light, increase the size of their orbits and eventually get absorbed by the electrodes.

“The idea is that you can change the size of the cyclotron orbits by adding or subtracting orbital angular momentum from the electrons, thus effectively moving them across the sample and creating a current,” Session says.

In the new paper, the research team reported observing a robust current that survived under a wide range of experimental conditions. They hit their graphene sample with light carrying orbital angular momentum that circulated clockwise and observed the current flowing in one direction. Then they hit it with light carrying counterclockwise orbital angular momentum and found that the direction of the current flipped. They flipped the direction of the applied magnetic field and observed the current flip directions, too—an expected finding since changing the magnetic field direction also swaps the direction electrons flow in their cyclotron orbits. They changed the voltage across the inner and outer electrodes and continued to see the same difference between currents generated by clockwise and counterclockwise vortex light. They also tested sending circularly polarized light, which carries an intrinsic angular momentum, at the sample, and they found that it barely generated any current. In all cases, the signal was clear: The current only appeared in the presence of light carrying orbital angular momentum, and the direction of the current was correlated with whether the light carried momentum that spun clockwise or counterclockwise.

The result was the culmination of several years of work, which included some false starts with sample fabrication and difficulties collecting enough good data from the experiment.

“I spent over a year just trying to make graphene samples with this kind of geometry,” Session says. Ultimately, Session and the team reached out to a group they had worked with before, led by Roman Sordan, a physicist at the Polytechnic University of Milan in Italy and an expert at preparing graphene samples. “They were able to come through and make the samples that we used,” says Session.

Once they had samples that worked well, they still had trouble aligning their twisted light with the sample to observe the current.

“The signal we were looking at was not quite consistent,” says Mahmoud Jalali Mehrabad, a postdoctoral researcher at JQI and UMD and a co-author of the paper. “Then one day, with Deric, we started to do this spatial sweep. And we kind of mapped the sample with really high accuracy. Once we did that—once we nailed down the very peak, optimized position for the beam—everything started to make sense.” Within a week or so, they had collected all the data they needed and could pick out all the signals of the current’s dependence on the orbital angular momentum of the beam.

Mehrabad says that, in addition to demonstrating a new method for controlling matter with light, the technique might also enable fundamentally new measurements of electrons in quantum materials. Specially prepared light beams, combined with interference measurements, could be used as a microscope that can image the spatial extent of electrons—a direct measurement of the quantum nature of electrons in a material.

“Being able to measure these spatial degrees of freedom of free electrons is an important part of measuring the coherence properties of electrons in a controllable manner—and manipulating them,” Mehrabad says. “Not only do you detect, but you also control. That’s like the holy grail of all this.”

Original story: https://jqi.umd.edu/news/twisted-light-gives-electrons-spinning-kick

In addition to Session, Hafezi, Schine, Sau, Solomon, Cao, Sordan and Mehrabad, the paper had several other authors: Nikil Paithankar, a graduate student at the Polytechnic University of Milan in Italy; Tobias Grass, a former JQI postdoctoral researcher who is now a research fellow at the Donostia International Physics Center in Donostia, Spain; Christian Eckhardt, a graduate student at the Max Planck Institute for the Structure and Dynamics of Matter in Hamburg, Germany; Daniel Gustavo Suárez Forero, a former JQI postdoctoral researcher who will be starting as an assistant professor of physics at the University of Maryland, Baltimore County in 2025; Kevin Li, a former JQI and UMD undergraduate student; Mohammad Alam, a former JQI and UMD undergraduate student who now works at IonQ; and Kenji Watanabe and Takashi Taniguchi, both researchers at the National Institute for Materials Science in Tsukuba, Japan.

This work was supported by ONR N00014-20-1-2325, AFOSR FA95502010223, ARO W911NF1920181, MURI FA9550-19-1-0399, FA9550-22-1-0339, NSF IMOD DMR-2019444, ARL W911NF1920181, Simons and Minta Martin foundations, and EU Horizon 2020 project Graphene Flagship Core 3 (grant agreement ID 881603). Tobias Grass acknowledges funding by BBVA Foundation (Beca Leonardo a Investigadores en Fisica 2023) and Gipuzkoa Provincial Council (QUAN-000021-01).

Repurposing Qubit Tech to Explore Exotic Superconductivity

Details

Published: Wednesday, November 20 2024 01:00

Decades of quantum research are now being transformed into practical technologies, including the superconducting circuits that are being used in physics research and built into small quantum computers by companies like IBM and Google. The established knowledge and technical infrastructure are allowing researchers to harness quantum technologies in unexpected, innovative ways and creating new research opportunities.

For superconducting circuits to be used as qubits—the basic building blocks of a quantum computer—the circuits need to reliably interact with delicate quantum states while keeping them carefully isolated from other influences. Superconducting circuits and the quantum states that occur in them are both sensitive to external influences, like shifting temperatures, stray electric fields or a passing particle, so manipulating qubits without external disruption is a delicate art. Instead of fashioning qubits in their superconducting circuits, researchers incorporated a sample with a magnetic layer atop a superconductor. They were able to use the sensitivity of the circuit to explore the quantum world hidden in the sample. The above image shows the changes in the circuit’s behavior as an applied magnetic field and the circuit’s properties were varied, revealing a strong interaction between the superconducting and ferromagnetic properties of the sample. (Credit: Harvard University)

The sensitivity of superconducting circuits means that minor blemishes in fabrication and installation can ruin a qubit, so much work has gone into establishing precise fabrication techniques to produce usable tech. All that work now means that the underlying sensitivity that makes superconducting qubits a pain to work with can be exploited as part of a quantum sensor. 

In a paper published in the journal Nature Physics earlier this year, a collaboration between theorists at JQI and experimentalists at Harvard University presented a technique that repurposes the technology of superconducting circuits to study samples with exotic forms of superconductivity. The collaboration demonstrated that by building samples of interest into a superconducting circuit they could spy on exotic superconducting behaviors that have eluded existing measurement techniques.

“Essentially, we turned bugs of superconducting qubits into features,” says JQI postdoctoral researcher Andrey Grankin, who is an author of the paper.

The team’s experiments provided a new way to distinguish exotic forms of superconductivity from the conventional, well-understood variety. The technique also allowed them to study superconducting currents that occur in such a thin section of a sample that most existing techniques for researching superconductivity aren’t reliable. Their results also suggest that the approach might be beneficial in other areas of condensed matter research—the field that studies how the interactions of the multitude of particles in a material produce its properties.

“By capitalizing on this technology from the quantum computing side of things, it turns out that we were able to show that this is actually a very nice sensor for looking at fundamental condensed matter physics problems, in particular for understanding superconductivity in very small systems,” says Harvard physics graduate student Nick Poniatowski, who is a co-lead author of the paper.

Superconducting circuits commonly rely on resonators in which electromagnetic waves bounce around for extended periods and interact with quantum states. Researchers can use the bouncing waves to both manipulate a nearby qubit and determine what state it is in.

The project started when researchers at Harvard began experimenting with placing a sample near the resonator instead of building a qubit there. Incorporating a sample into the superconducting circuits provided the researchers with a convenient way to send currents through the sample as well as a built-in sensor, in the form of the resonator, to detect subtle changes in its electronic states.

In particular, the team studied samples where a magnetic layer lies on a superconductor, which research indicated might be a promising environment for hunting down exotic forms of superconductivity.

Ordinary superconductivity doesn’t mix well with magnetism, but researchers have discovered signs of interesting interactions that occur when magnetic and superconducting materials are brought together. Superconductivity and magnetism are connected because electrons must be paired up to create a superconducting current and magnetism influences how electrons can partner up.

Every electron has a quantum property called spin, which makes it behave similarly to a tiny magnet. Like a bar magnet will flip around when it is forced toward a magnet with the same orientation, electrons want the magnetic fields of their spins to point in opposite directions. As a result, the best-understood superconductors have electrons that pair with counterparts with the opposite spin. But magnetic materials—ferromagnets—have a magnetic field that works to twist all the electrons so that their fields point the same way, preventing such pairings.

However, experiments have indicated that there are likely more exotic forms of superconductivity with alternative ways that electrons form pairs. Some of the exotic pairings—called spin-triplet pairings—have spins that point in the same direction. Evidence suggests that spin-triplet pairs might be found in certain seemingly conventional superconductors, just in small numbers that are obscured by the large crowd of traditional superconducting pairs.

Prior results have indicated that at the interface of a ferromagnet and a superconductor, some spin-triplet electron pairs from the superconductor might be able to bleed over into the magnetic territory. Researchers hope that studying exotic superconductors and their pairing mechanisms will open new opportunities in research and quantum computing and possibly even identify a superconductor that works under more convenient conditions than the frigid temperatures or high pressures required for known superconductors.

In the new work, the experimentalists chose samples composed of a magnetic layer of a material called permalloy attached to a superconducting layer of niobium. As they performed experiments, they weren’t getting a clear picture of what was happening at the interface or identifying a smoking gun to indicate unconventional superconductivity, so they turned to theorists. Poniatowski previously studied physics as an undergraduate at UMD, and he knew that Grankin and Professor and JQI Fellow Victor Galitski had an interest in the theories that describe superconducting qubits. Poniatowski reached out, and Grankin and Galitski, who is also a Chesapeake Chair Professor of Theoretical Physics in the Department of Physics, joined the project.

“Eventually, Andrey came up with these models that then guided us a little bit more in the experiment to perform certain experimental tests,” says Yale postdoctoral associate Charlotte Bøttcher, a co-lead author of the paper who was a Harvard graduate student when she performed this research. “Before that, it was like shooting a little bit in the dark and just trying to play with whatever we had available. But then the interaction with theory was really what made things come together.”

The experiment needed a clear way of distinguishing exotic electron pairs from their conventional counterparts. Normally, measuring the electrical resistance—a material’s opposition to currents, which make electric circuits lose energy—is a cornerstone of superconductor research since the defining feature of a superconducting current is that it experiences no resistance. But all superconductors having zero resistance means resistance measurements can’t distinguish between exotic and traditional superconducting paired electrons.

However, a superconductor can have a distinctive inductance—a material’s resistance to changes in the electrical current. The inductance depends on many things, like the size and shape of the material. In superconductors, the inductance depends on the number of superconducting electron pairs that are present, with the dependence getting more dramatic the fewer electron pairs there are. The team focused on measuring the inductance to tease out what was occurring near the interface.

“Really, the advantage of our probe is we're measuring the badness of the superconductor not the goodness,” says Poniatowski. “So if a superconducting current is very weak, we can see a very strong response.”

Using this technique allowed the team not only to measure how much superconductivity occurred at the interfaces but also to observe how that amount changed as they varied the temperature. This was crucial to distinguishing what type of superconductivity they were observing. Since the exotic spin-triplet pairs are more delicate than their traditional counterparts, their population should fall off much more quickly as the temperature increases.

The team obtained even more information by performing the experiments with the sample placed in magnetic fields that pointed in various directions relative to the flow of the electrical current. Grankin’s contributions included determining how the experimental signatures should differ if the magnetic field points along the direction of the current or perpendicular to it.

The population changes they observed matched the signature of the exotic spin-triplet states described by the theory.

“Using this new technique borrowing from quantum information technologies we were able to actually see evidence for this very subtle, hard-to-detect exotic pairing state in this simple system,” says Poniatowski.

While the results indicated the presence of exotic superconductivity, they also revealed further mysteries to be investigated about how those states can exist in the material. The theories predict that in addition to being disrupted easily by heat, the superconducting states should also be easily disrupted by any mess in the material’s structure.

“Typically, this state is so fragile that any disorder should kill the state,” says Bøttcher. “And our systems are, for sure, not perfect, so it's a bit of a puzzle why this state is even there in the first place. And when there's something you don't understand, I think that that means we're not done yet, that more work needs to be done.”

The researchers hope to unravel the lingering mysteries and explore other research avenues their experiments have opened. They are working on additional research to further explore the behaviors of spins in the ferromagnet layer. And, based on their results, they are also optimistic that with further work similar devices may provide a window into the interactions of light and the magnetic excitations in a material—an emerging research field called magnonics.

“I actually think that the most important thing and our most important job as researchers is to look for what you didn't expect because I think that's where the new research and the new physics lies,” says Bøttcher. “So, I am always very excited when I see something I didn't expect, and I think this project was exactly that.”

Original story by Bailey Bedford: https://jqi.umd.edu/news/repurposing-qubit-tech-explore-exotic-superconductivity

In addition to Grankin, Galitski, Bøttcher and Poniatowski, co-authors of the paper include Harvard physics professor Amir Yacoby, Harvard postdoctoral fellow Uri Vool, and Harvard graduate students Zihan Yan and Marie Wesson.

This research was supported by the Quantum Science Center, a National Quantum Information Science Research Center of the US Department of Energy; the NSF NNIN award ECS-00335765; the Department of Defense through the NDSEG fellowship program; the National Science Foundation Grant No. DMR-2037158 and DMR-1708688; the US Army Research Office Contract No. W911NF1310172; the Simons Foundation; and the Gordon and Betty Moore Foundation Grant No. GBMF 9468.

New Design Packs Two Qubits into One Superconducting Junction

Details

Published: Monday, October 21 2024 00:01

Quantum computers are potentially revolutionary devices and the basis of a growing industry. However, their technology isn’t standardized yet, and researchers are still studying the physics behind the diverse ways to build these quantum devices. Even the most basic building blocks of a quantum computer—qubits—are still an active research topic.A superconducting circuit studied in Alicia Kollár’s lab. The middle of the three rectangles along the bottom are junctions that hold quantum states that may each be used as a qubit. A proposal to adjust the dimensions of the junctions would allow chips like this to host twice as many qubits.

In an article published September 23, 2024 in the journal Physical Review A, JQI researchers proposed a way to use the physics of superconducting junctions to let each function as more than one qubit. They also outlined a method to use the new qubit design in quantum simulations. While these proposed qubits might not immediately replace their more established peers, they illustrate the rich variety of quantum physics that remains to be explored and harnessed in the field.

Superconducting junctions are part of many diverse qubit designs, including those in the prototype quantum computers of IBM and Google. All the designs feature an island made of a superconductor joined to the rest of a superconducting circuit by a thin layer of insulator that forms the junction between the two sections. To cross the barrier, electrons in the circuit must quantum tunnel through the junction, influencing which quantum states the circuit can naturally hold.

JQI Fellow Mohammad Hafezi and JQI postdoctoral researcher Andrey Grankin, who is the first author of the paper, reviewed the research on junctions in superconducting circuits, and what they found left them wondering if the existing qubit designs were taking advantage of the full breadth of physics that can be realized in superconducting junctions. The design of a junction—the geography of the superconducting island—impacts which states it can host, and current designs have focused on small junctions and the simplest states.

In some qubit designs, the quantum states depend on the geography of the island because of how electrons in a superconductor are free to move around like a fluid. Like water in a small pool, the electrons can slosh back and forth and form waves that are influenced by their surroundings.

Certain electron waves isolated onto a superconducting island can be very stable and long lasting, which makes them useful for storing quantum information in a qubit. The waves that are stable are examples of a more general phenomenon, called standing waves, that occur when a wave isn’t interrupted during the slope of one of its hills or valleys; instead, its oscillations are perfectly completed at the edges (the walls of a pool, the points where a string is being held, etc.). A guitar string’s harmonics are also examples of standing waves. 

But just having a stable standing wave in the superconducting electrons isn’t enough to be useful for quantum calculations. To use a standing wave as a qubit, a quantum computer must be able to distinguish it from all other standing waves and individually target it. Current superconducting qubit designs circumvent this issue by using short junctions that host just a single standing wave; as long as a junction is sufficiently short, the physics governing the superconducting electrons effectively only allows a single standing wave on the superconducting island. Researchers have also studied junctions that meet along very long interfaces and found that they can easily host a vast array of standing waves. Unfortunately, the abundance of standing waves comes with a downside: The more standing waves there are, the more similar the waves become, which makes them difficult to tell apart and inconvenient for quantum computing. 

“Historically short and long junctions were researched quite extensively,” says Grankin, who is the first author of the paper. “But the intermediate junction lengths have not been studied.”

Hafezi—who is also a Minta Martin professor of electrical and computer engineering and physics at the University of Maryland (UMD) and a Senior Investigator at the National Science Foundation Quantum Leap Challenge Institute for Robust Quantum Simulation (RQS)—and Grankin became interested in this intermediate regime. The pair consulted with JQI Fellow Alicia Kollár, another author of the paper who works with superconducting qubits in her research.

“Andrey and Mohammad came to me with a creative new idea for how to make a junction host multiple qubit excitations,” says Kollár, who is also a Chesapeake Assistant Professor of Physics and a Co-Associate Director of Research for RQS. “Our main challenge was coming up with a design that would yield practical device parameters and a device that is actually within reach of current state-of-the-art fabrication techniques.”

Together, the group explored the behaviors of electrons in the intermediate case and if it is practical to produce multiple excitations that can be easily distinguished and separately manipulated. Since the pool of electrons is within a solid superconductor, setting them into motion isn’t as simple as plucking a string. To push and pull on electrons you need an electric field. One way that physicists like to push electrons around is using a special reflective chamber—or resonator—that is full of electric and magnetic fields in the form of light. 

Light in a resonator can form its own standing waves that act on the electrons. Similar to a guitar string vibrating due to the sound waves from another string—or more dramatically the sound of a singer’s voice shaking a glass until it shatters—the right light waves inside a resonator can excite electrons in a superconducting junction into a standing wave, which physicists call a mode of the junction. 

The team analyzed how medium-sized junctions should behave inside a resonator and found promising results. The various modes of a junction each respond more or less strongly to particular frequencies of light, so light can be selected to target a specific mode. The response of a mode to a standing wave of light in a resonator also depends on whether the symmetries of the mode and the light match. If the waves of light in the resonator are symmetrical across the center of a junction, they naturally push electrons into waves with a similar symmetry. For instance, if the light waves crossing the junction form a hill on one side and a valley on the other, they can’t push the electrons into a simple hill reflected across the center of the device, but they might be able to excite a similarly lopsided mode of the junction. 

So, light that creates one mode in the superconducting electrons may be ignored by another mode. In the paper, the team described a method of exploiting these two effects to excite or manipulate only a targeted mode. The researchers proposed a design where two distinct modes are targeted so that a single junction functions as two independent qubits. They also described a method to use a one-dimensional line of junctions to simulate interactions between two-dimensional grids of quantum particles. However, they haven’t yet tackled fabricating the junctions and demonstrating the feasibility of their proposal.

“This project started from a fundamental interest in the electrodynamics of extended junctions,” Grankin says. “Then it turned out to be also useful from the quantum information and simulation perspective.”

Original story by Bailey Bedford: https://jqi.umd.edu/news/new-design-packs-two-qubits-one-superconducting-junction

HAWC Finds High-Energy Gamma-Ray Emissions from Microquasar V4641 Sagittarii

Details

Published: Wednesday, October 16 2024 00:01

A new study in Nature, “Ultra-high-energy gamma-ray bubble around microquasar V4641 Sgr,"   has  revealed a groundbreaking discovery by researchers from the High Altitude Water Cherenkov (HAWC) observatory:  TeV gamma-ray emissions from V4641 Sagittarii (V4641 Sgr), a binary system composed of a black hole and a main sequence B-type companion star. This discovery provides fresh insights into particle acceleration in large-scale jets emitted from microquasars, which serve as natural laboratories for studying high-speed jets produced by matter falling onto spinning black holes. The findings show that V4641 Sgr's gamma-ray emissions occur at similar distances from the black hole as those observed in another well-known microquasar, SS 433. This makes V4641 Sgr stand out for its super-Eddington accretion and one of the fastest superluminal jets in the Milky Way. With a gamma-ray spectrum that ranks among the hardest of any known TeV sources, the emissions are detected at energies exceeding 200 TeV. Schematic illustration of the V4641 Sgr region.

The study's results indicate that the gamma rays are likely produced by extremely high-energy protons. While studies of SS 433 indicate that the emission from this object is likely electrons, the high energy emission at large distances from V4641 argue against electrons due to their rapid energy loss at high energies. The implications are profound: the environment around the microquasar may play a critical role in determining whether the emission from the large-scale jets comes from electrons or protons, and they could play a significant role as a source of Galactic cosmic rays. These observations open new avenues for understanding particle acceleration in extreme environments and contribute to the broader study of high-energy astrophysics. 

 "At a zenith angle of 46° from HAWC's field of view, the microquasar V4641 Sgr has accelerated particles to the knee of the cosmic-ray spectrum, pushing the boundaries of our understanding of particle acceleration and transport in such extreme environments," said Dr. Dezhi Huang, a Post-Doctoral researcher at the University of Maryland. "HAWC observatory continues to deliver exceptional performance, providing valuable insights into these high-energy processes that were previously beyond our reach. "

“The HAWC survey has discovered for the first time very high energy gamma rays from the extended 100 pc jet of the microquasar V4641 Sgr. This proves that jets launched in accreting systems can accelerate particles up to PeV energies, and therefore that microquasars are potentially significant contributors to the Galactic cosmic ray population at high energies," saidDr. Sabrina Casanova, Professor from Institute of Nuclear Physics of the Polish Academy of Sciences. "Furthermore, although very high energy protons are strongly suspected to exist in the kpc jets of active galactic nuclei (AGN), the associated hundred TeV emission has never been observed from an extended region from an AGN jet due to strong gamma-ray absorption over the long distances to Earth.”Differential spectrum weighted by E2 for the northern and southern sources in a model with two point sources and for the asymmetric extended source in a model with a single asymmetric extended source. The shaded regions indicate the best-fit spectra and 1σ statistical uncertainties when fitting a single-power-law model to the data from 10 to >200 TeV. The markers correspond to the best-fit values and their 1σ statistical uncertainties obtained when fitting a single-power-law model to data in individual energy bins. The chosen energy range for plotting the spectrum is specified in the Methods.

This study was supported by the collaborative efforts of multiple institutions, with major contributions from the University of Maryland. Distinguished University Professor Jordan Goodman, a member of the collaboration's internal editorial board, played a key role in guiding the publication. Dr. Dezhi Huang, one of the corresponding authors, led significant aspects of the analysis, while Dr. Kristi Engel helped refine the paper. Additional support came from UMD HAWC group members Research Scientist Dr. Andrew Smith, Project Engineer Michael Schneider, Dr. Zhen Wang, Dr. Jason Fan and graduate students Sohyoun Yun-Cárcamo.

More on the finding: https://www.nature.com/articles/d41586-024-03191-x

Nobel Prize Celebrates Interplay of Physics and AI

Details

Published: Friday, October 11 2024 00:01

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)

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