Tagged: Quanta Magazine Toggle Comment Threads | Keyboard Shortcuts

  • richardmitnick 8:11 pm on May 15, 2023 Permalink | Reply
    Tags: , "Physicists Create Elusive Particles That Remember Their Pasts", , , In 1982 Frank Wilczek helped open physicists’ minds to the menagerie of particles that could exist in two dimensions., Last fall researchers with Google celebrated the first clear intertwining of non-abelian objects., , Quanta Magazine, , , Quantum processors are changing the hunt for anyons., , Researchers have spent millions of dollars over the past three decades or so trying to capture and tame the particle-like objects which go by the cryptic moniker of "non-abelian anyons"., The next milestone will be real error correction which neither Google nor Quantinuum attempted., The shared memory of non-abelian anyons could serve as an ideal qubit.,   

    From “Quanta Magazine” : “Physicists Create Elusive Particles That Remember Their Pasts” 

    From “Quanta Magazine”

    5.9.23
    Charlie Wood

    1
    By “braiding” particles around each other, quantum computers could store and manipulate information in a way that protects against errors.
    Merrill Sherman/Quanta Magazine.

    Forty years ago, Frank Wilczek was mulling over a bizarre type of particle that could live only in a flat universe. Had he put pen to paper and done the calculations, Wilczek would have found that these then-theoretical particles held an otherworldly memory of their past, one woven too thoroughly into the fabric of reality for any one disturbance to erase it.

    However, seeing no reason that nature should allow such strange beasts to exist, the future Nobel prize-winning physicist chose not to follow his thought experiments to their most outlandish conclusions — despite the objections of his collaborator Anthony Zee, a renowned theoretical physicist at the University of California-Santa Barbara.

    “I said, ‘Come on, Tony, people are going to make fun of us,’” said Wilczek, now a professor at the Massachusetts Institute of Technology.

    Others weren’t so reluctant. Researchers have spent millions of dollars over the past three decades or so trying to capture and tame the particlelike objects, which go by the cryptic moniker of “non-abelian anyons”.

    Now two landmark experiments have finally succeeded, and no one is laughing. “This has been a target, and now it’s hit,” Wilczek said.

    Physicists working with the company Quantinuum announced today that they had used the company’s newly unveiled, next-generation H2 processor to synthesize and manipulate non-abelian anyons in a novel phase of quantum matter.

    2
    Researchers used Quantinuum’s new H2 processor to simulate a novel state of matter in which non-abelian anyons can be created and manipulated.
    Credit: Quantinuum.

    Their work follows a paper posted last fall in which researchers with Google celebrated the first clear intertwining of non-abelian objects, a proof of concept that information can be stored and manipulated in their shared memory. Together, the experiments flex the growing muscle of quantum devices while offering a potential glimpse into the future of computing: By maintaining nearly indestructible records of their journeys through space and time, non-abelian anyons could offer the most promising platform for building error-tolerant quantum computers.

    “As pure science, it’s just, wow,” said Ady Stern, a condensed matter theorist at the Weizmann Institute of Science in Israel who has spent his career studying the objects. “This brings you closer [to topological quantum computing]. But if there’s one thing the last few decades have shown us, it’s a long and winding road.”

    Flatland Computing

    In 1982, Wilczek helped open physicists’ minds to the menagerie of particles that could exist in two dimensions. He worked out the consequences of confining quantum laws to a hypothetical, entirely flat universe, and found that it would contain strange particles with fractional spins and charges. Moreover, swapping otherwise indistinguishable particles could change them in ways that were impossible for their three-dimensional counterparts. Wilczek cheekily named these two-dimensional particles anyons, since they seemed to be capable of nearly anything.

    Wilczek focused on the simplest “abelian” anyons, particles that, when swapped, change in subtle ways that are not directly detectable.

    He stopped short of exploring the wilder option — non-abelian anyons, particles that share a memory. Swapping the positions of two non-abelian anyons produces a directly observable effect. It switches the state of their shared wave function, a quantity that describes a system’s quantum nature. If you stumble upon two identical non-abelian anyons, by measuring which state they are in, you can tell whether they have always been in those positions or whether they’ve crossed paths — a power no other particle can claim.

    To Wilczek, that notion seemed too fantastical to develop into a formal theory. “What kinds of states of matter support those?” he recalled thinking.

    But in 1991, two physicists identified those states [Nuclear Physics B (below)]. They predicted that, when subjected to strong enough magnetic fields and cold enough temperatures, electrons stuck to a surface would swirl together in just the right way to form non-abelian anyons. The anyons would not be fundamental particles — our 3D world forbids that — but “quasiparticles.” These are collections of particles, but they are best thought of as individual units. Quasiparticles have precise locations and behaviors, just as collections of water molecules produce waves and whirlpools.

    In 1997, Alexei Kitaev, a theorist at the California Institute of Technology, pointed out that such quasiparticles could lay the perfect foundation for quantum computers. Physicists have long salivated at the possibility of harnessing the quantum world to perform calculations beyond the reach of typical computers and their binary bits. But qubits, the atom-like building blocks of quantum computers, are fragile. Their wave functions collapse at the lightest touch, erasing their memories and their ability to perform quantum calculations. This flimsiness has complicated ambitions to control qubits long enough for them to finish lengthy calculations.

    Kitaev realized that the shared memory of non-abelian anyons could serve as an ideal qubit. For starters, it was malleable. You could change the state of the qubit — flipping a zero to a one — by exchanging the positions of the anyons in a manner known as “braiding.”

    You could also read out the state of the qubit. When the simplest non-abelian anyons are brought together and “fused,” for instance, they will emit another quasiparticle only if they have been braided. This quasiparticle serves as a physical record of their crisscrossed journey through space and time.

    And crucially, the memory is also nigh incorruptible. As long as the anyons are kept far apart, poking at any individual particle won’t change the state the pair is in — whether zero or one. In this way, their collective memory is effectively cut off from the cacophony of the universe.

    “This would be the perfect place to hide information,” said Maissam Barkeshli, a condensed matter theorist at the University of Maryland.

    Unruly Electrons

    Kitaev’s proposal came to be known as “topological” quantum computing because it relied on the topology of the braids. The term refers to broad features of the braid — for example, the number of turns — that aren’t affected by any specific deformation of their path. Most researchers now believe that braids are the future of quantum computing, in one form or another. Microsoft, for instance, has researchers trying to persuade electrons to form non-abelian anyons directly. Already, the company has invested millions of dollars into building tiny wires that — at sufficiently frigid temperatures — should host the simplest species of braidable quasiparticles at their tips. The expectation is that at these low temperatures, electrons will naturally gather to form anyons, which in turn can be braided into reliable qubits.

    After a decade of effort, though, those researchers are still struggling to prove that their approach will work. A splashy 2018 claim that they had finally detected the simplest type of non-abelian quasiparticle, known as “Majorana zero modes,” was followed by a similarly high-profile retraction in 2021. The company reported new progress in a 2022 paper, but few independent researchers expect to see successful braiding soon.

    Similar efforts to turn electrons into non-abelian anyons have also stalled. Bob Willett of Nokia Bell Labs has probably come the closest [Physical Review X] in his attempts to corral electrons in gallium arsenide, where promising but subtle signs of braiding exist. The data is messy, however, and the ultracold temperature, ultrapure materials, and ultrastrong magnetic fields make the experiment tough to reproduce.

    “There has been a long history of not observing anything,” said Eun-Ah Kim of Cornell University.

    Wrangling electrons, however, is not the only way to make non-abelian quasiparticles.

    “I had given up on all of this,” said Kim, who spent years coming up with ways to detect anyons as a graduate student and now collaborates with Google. “Then came the quantum simulators.”

    Compliant Qubits

    Quantum processors are changing the hunt for anyons. Instead of trying to coax hordes of electrons to fall into line, in recent years researchers have begun using the devices to bend individual qubits to their will. Some physicists consider these efforts simulations, because the qubits inside the processor are abstractions of particles (while their physical nature varies from lab to lab, you can visualize them as particles spinning around an axis). But the quantum nature of the qubits is real, so — simulations or not — the processors have become playgrounds for topological experiments.

    “It breathes new life” into the field, said Fiona Burnell, a condensed matter theorist at the University of Minnesota, “because it’s been so hard to make solid-state systems.”

    Synthesizing anyons on quantum processors is an alternate way to leverage the power of Kitaev’s braids: Accept that your qubits are mediocre, and correct their errors. Today’s shoddy qubits don’t work for very long, so anyons built from them would also have short lifetimes. The dream is to quickly and repeatedly measure groups of qubits and correct errors as they crop up, thereby extending the life span of the anyons. Measurement erases an individual qubit’s quantum information by collapsing its wave function and turning it into a classical bit. That would happen here too, but the important information would remain untouchable — hidden in the collective state of many anyons. In this way, Google and other companies hope to shore up qubits with fast measurements and swift corrections (as opposed to low temperatures).

    “Ever since Kitaev,” said Mike Zaletel, a condensed matter physicist at the University of California-Berkeley, “this has been the way people think quantum error correction will likely work.”

    Google took a major step toward quantum error correction in the spring of 2021, when researchers assembled about two dozen qubits into the simplest grid capable of quantum error correction, a phase of matter known as the toric code.

    Creating the toric code on Google’s processor amounts to forcing each qubit to strictly cooperate with its neighbors by gently nudging them with microwave pulses. Left unmeasured, a qubit points in a superposition of many possible directions. Google’s processor effectively cut down on those options by making each qubit coordinate its spin axis with its four neighbors in specific ways. While the toric code has topological properties that can be used for quantum error correction, it doesn’t natively host non-abelian quasiparticles. For that, Google had to turn to a strange trick long known [Physical Review Letters(below)] to theorists: certain imperfections in the grid of qubits, dubbed “twist defects,” can acquire non-abelian magic.

    Last fall, Kim and Yuri Lensky, a theorist at Cornell, along with Google researchers, posted a recipe for easily making and braiding pairs of defects in the toric code. In a preprint posted shortly after, experimentalists at Google reported implementing that idea, which involved severing connections between neighboring qubits. The resulting flaws in the qubit grid acted just like the simplest species of non-abelian quasiparticle, Microsoft’s Majorana zero modes.

    “My initial reaction was ‘Wow, Google just simulated what Microsoft is trying to build. It was a real flexing moment,” said Tyler Ellison, a physicist at Yale University.

    4
    Merrill Sherman/Quanta Magazine.

    By tweaking which connections they cut, the researchers could move the deformations. They made two pairs of non-abelian defects, and by sliding them around a five-by-five-qubit chessboard, they just barely eked out a braid. The researchers declined to comment on their experiment, which is being prepared for publication, but other experts praised the achievement.

    “In a lot of my work, I’ve been doodling similar-looking pictures,” Ellison said. “It’s amazing to see that they actually demonstrated this.”

    Paint by Measurement

    All the while, a group of theorists headed up by Ashvin Vishwanath at Harvard University was quietly pursuing what many consider an even loftier goal: creating a more complicated phase of quantum matter where true non-abelian anyons — as opposed to defects — arise natively in a pristine phase of matter. “[Google’s] defect is kind of a baby non-abelian thing,” said Burnell, who was not involved in either effort.

    Anyons of both types live in phases of matter with a topological nature defined by intricate tapestries of gossamer threads, quantum connections known as entanglement. Entangled particles behave in a coordinated way, and when trillions of particles become entangled, they can ripple in complicated phases sometimes likened to dances. In phases with topological order, entanglement organizes particles into loops of aligned spins. When a loop is cut, each end is an anyon.

    Topological order comes in two flavors. Simple phases such as the toric code have “abelian order.” There, loose ends are abelian anyons. But researchers seeking true non-abelian anyons have their sights set on a completely different and much more complicated tapestry with non-abelian order.

    Vishwanath’s group helped cook up a phase with abelian order in 2021. They dreamt of going further, but stitching qubits into non-abelian entanglement patterns proved too intricate for today’s unstable processors. So the crew scoured the literature for fresh ideas.

    They found a clue in a pair of papers [ https://arxiv.org/pdf/quant-ph/0108118.pdf and https://arxiv.org/pdf/quant-ph/0407255.pdf ] decades before. Most quantum devices compute by massaging their qubits much as one might fluff a pillow, in a gentle way where no stuffing flies out through the seams. Carefully knitting entanglement through these “unitary” operations takes time. But in the early 2000s Robert Raussendorf, a physicist now at the University of British Columbia, hit on a shortcut. The secret was to hack away chunks of the wave function using measurement — the process that normally kills quantum states.from decades before. Most quantum devices compute by massaging their qubits much as one might fluff a pillow, in a gentle way where no stuffing flies out through the seams. Carefully knitting entanglement through these “unitary” operations takes time. But in the early 2000s Robert Raussendorf, a physicist now at the University of British Columbia, hit on a shortcut. The secret was to hack away chunks of the wave function using measurement — the process that normally kills quantum states.

    “It’s a really violent operation,” said Ruben Verresen, one of Vishwanath’s collaborators at Harvard.

    Raussendorf and his collaborators detailed how selective measurements on certain qubits could take an unentangled state and intentionally put it into an entangled state, a process Verresen likens to cutting away marble to sculpt a statue.

    The technique had a dark side that initially doomed researchers’ attempts to make non-abelian phases: Measurement produces random outcomes. When the theorists targeted a particular phase, measurements left non-abelian anyons speckled randomly about, as if the researchers were trying to paint the Mona Lisa by splattering paint onto a canvas. “It seemed like a complete headache,” Verresen said.

    Toward the end of 2021, Vishwanath’s group hit on a solution: sculpting the wave function of a qubit grid with multiple rounds of measurement. With the first round, they turned a boring phase of matter into a simple abelian phase. Then they fed that phase forward into a second round of measurements, further chiseling it into a more complicated phase. By playing this game of topological cat’s cradle, they realized they could address randomness while moving step by step, climbing a ladder of increasingly complicated phases to reach a phase with non-abelian order.

    “Instead of randomly trying measurements and seeing what you get, you want to hop across the landscape of phases of matter,” Verresen said. It’s a topological landscape that theorists have only recently begun to understand.

    Last summer, the group put their theory to the test on Quantinuum’s H1 trapped-ion processor, one of the only quantum devices that can perform measurements on the fly. Replicating parts of Google’s experiment, they made the abelian toric code and created a stationary non-abelian defect in it. They tried for a non-abelian phase but couldn’t get there with only 20 qubits.

    But then a researcher at Quantinuum, Henrik Dreyer, took Verresen aside. After swearing him to secrecy with a nondisclosure agreement, he told Verresen that the company had a second-generation device. Crucially, the H2 had a whopping 32 qubits. It took substantial finagling, but the team managed to set up the simplest non-abelian phase on 27 of those qubits. “If we had one or two fewer qubits, I don’t think we could have done it,” Vishwanath said.

    Their experiments marked the first unassailable detection of a non-abelian phase of matter. “To realize a non-abelian topological order is something people have wanted to do for a long time,” Burnell said. “That’s definitely an important landmark.”

    Their work culminated in the braiding of three pairs of non-abelian anyons such that their trajectories through space and time formed a pattern known as Borromean rings, the first braiding of non-abelian anyons. Three Borromean rings are inseparable when together, but if you cut one the other two will fall apart.

    “There’s a kind of gee-whiz factor,” Wilczek said. “It takes enormous control of the quantum world to produce these quantum objects.”

    The Big Chill

    As other physicists celebrate these milestones, they also emphasize that Google and Quantinuum are running a different race than the likes of Microsoft and Willett. Creating topological phases on a quantum processor is like making the world’s tiniest ice cube by stacking a few dozen water molecules — impressive, they say, but not nearly as satisfying as watching a slab of ice form naturally.

    “The underlying math is extremely beautiful, and being able to validate that is definitely worthwhile,” said Chetan Nayak, a researcher at Microsoft who has done pioneering work on non-abelian systems. But for his part, he said, he’s still hoping to see a system settle into a state with this sort of intricate entanglement pattern on its own when cooled.

    “If this was unambiguously seen in [Willett’s experiments], our minds would be blown,” Barkeshli said. Seeing it in a quantum processor “is cool, but no one’s getting blown away.”

    The most exciting aspect of these experiments, according to Barkeshli, is their significance for quantum computation: Researchers have finally shown that they can make the necessary ingredients, 26 years after Kitaev’s initial proposal. Now they just need to figure out how to really put them to work.

    One snag is that like Pokémon, anyons come in a tremendous number of different species, each with its own strengths and weaknesses. Some, for example, have richer memories of their pasts, making their braids more computationally powerful. But coaxing them into existence is harder. Any specific scheme will have to weigh such trade-offs, many of which aren’t yet understood.

    “Now that we have the ability to make different kinds of topological order, these things become real, and you can talk about these trade-offs in more concrete terms,” Vishwanath said.

    The next milestone will be real error correction, which neither Google nor Quantinuum attempted. Their braided qubits were hidden but not protected, which would have required measuring the crummy underlying qubits and quickly fixing their errors in real time. That demonstration would be a watershed moment in quantum computation, but it’s years away — if it’s even possible.

    Until then, optimists hope these recent experiments will launch a cycle where more advanced quantum computers lead to a better command over non-abelian quasiparticles, and that control in turn helps physicists develop more capable quantum devices.

    “Just bringing out the power of measurement,” Wilczek said, “that’s something that might be a game-changer.”

    Nuclear Physics B 1991
    Physical Review X
    Physical Review Letters 2010

    See the full article here.

    Comments are invited and will be appreciated, especially if the reader finds any errors which I can correct. Use “Reply”.


    five-ways-keep-your-child-safe-school-shootings

    Please help promote STEM in your local schools.

    Stem Education Coalition

    Formerly known as Simons Science News, Quanta Magazine is an editorially independent online publication launched by the Simons Foundation to enhance public understanding of science. Why Quanta? Albert Einstein called photons “quanta of light.” Our goal is to “illuminate science.” At Quanta Magazine, scientific accuracy is every bit as important as telling a good story. All of our articles are meticulously researched, reported, edited, copy-edited and fact-checked.

     

  • richardmitnick 1:23 pm on May 1, 2023 Permalink | Reply
    Tags: "How Pools of Genetic Diversity Affect a Species’ Fate", A richer understanding of genetic diversity and population dynamics can help conservation practitioners make better choices about how to save imperiled animals., A species is a breeding pool that holds all the variant forms (or alleles) of genes that one might find in a type of organism., , Biologists have developed a deeper understanding of how the relationship between genetic diversity and population structure can influence the fate of a species., , Distinct populations can sometimes act as refuges or reservoirs for uncommon genes., , , , In recent years biologists have developed a deeper understanding of how the relationship between genetic diversity and population structure can influence the fate of a species., Quanta Magazine, Researchers are discovering how much the genetic dynamics within and among populations can affect how resiliently a species can evolve and adapt to changing conditions over time., Sprinkled throughout the genomes of even healthy populations are detrimental genes called deleterious recessive alleles., The species concept is a linchpin of modern biology and evolutionary theory., The total genetic diversity of a species can be parceled out among subspecies and other populations that express different traits., When a species or population gets too small it can lose more than its adaptive potential for surviving future threats. It can lose its ability to sustain itself., When we try to save a species what exactly are we trying to save?, Working to turn stem cells into embryos for in vitro fertilization.   

    From “Quanta Magazine” : “How Pools of Genetic Diversity Affect a Species’ Fate” 

    From “Quanta Magazine”

    4.25.23
    Anna Funk

    1
    The total genetic diversity of a species can be parceled out among subspecies and other populations that express different traits. Preserving that genetic diversity when those groups dwindle can be a challenge. Credit: Samantha Mash for Quanta Magazine.

    In March 2018 at the Ol Pejeta Conservancy in Kenya, surrounded by his devoted keepers, Sudan the northern white rhino breathed his last. He wasn’t the only remaining northern white rhino because three females in protective captivity survived him. But Sudan’s death ended any hope of those females breeding naturally and rendered the northern white rhino effectively extinct. The moment made headlines, and the world lamented the high-profile extinction.

    But it wasn’t a true species extinction event. Northern white rhinos are just a distinct subset, or subspecies, of white rhino. More than 10,000 white rhinos are still left in the southern subspecies. White rhinos as a species aren’t even endangered.

    Nevertheless, ever since Sudan’s death, Cynthia Steiner, an associate director in conservation genetics for the San Diego Zoo Wildlife Alliance, and her colleagues have gone to great lengths to try to reboot the northern subspecies. They’re working to turn stem cells collected from the remaining females into embryos for in vitro fertilization [Genome Research (below)]. They want more northern white rhinos, and mixing in genes from southern white rhino males just won’t do.

    “Before, people would say, we need to save the species,” Steiner explained. “But that’s not enough if you think about it. When you talk about species, you’re not considering the whole evolutionary potential of all the different groups that make up that species. That’s why the new notion is: We really want to save the genetic structure of the species, these different populations — subspecies — that have unique characteristics at the genomic level.”

    In recent years, biologists have developed a deeper understanding of how the relationship between genetic diversity and population structure can influence the fate of a species. They’ve long understood how geography and ecological variations often partition species into subspecies or other small distinct populations of individuals who are more closely related to one another than to outsiders. Now, with the help of better tools and techniques for studying the genomes of creatures in the wild, researchers are discovering the full extent of how much the genetic dynamics within and among those populations can affect how resiliently a species can evolve and adapt to changing conditions over time. Those distinct populations can sometimes act as refuges or reservoirs for uncommon genes, and they can become the salvation of a species if new threats suddenly make those genes more valuable. On the other hand, if the smaller populations become too isolated, they can die out and make chunks of a species’ genetic diversity vanish forever.

    It’s an important insight into how species naturally adapt and evolve. Luckily for the many species of rhinos, whales, panthers, amphibians and other endangered organisms around the world, this richer understanding of genetic diversity and population dynamics can also help conservation practitioners make better choices about how to save imperiled animals.

    But the choices involved sometimes mean probing more deeply into a crucial question: When we try to save a species, what exactly are we trying to save?

    Not Just Species

    Biologists typically use “species” as a label for “reproductively isolated” populations of organisms that generally breed among themselves and not with outsiders. In effect, a species is a breeding pool that holds all the variant forms (or alleles) of genes that one might find in a type of organism. The species concept is a linchpin of modern biology and evolutionary theory, although it has been intensely criticized for failing to capture the reality of how some organisms actually behave and breed.

    Yet although species are pools of genes, those pools are not evenly mixed. Subgroups of lineages within them may tend to breed with one another more often, and they may develop sets of distinctive traits. Naturalists sometimes recognize such groups as subspecies. But even without a subspecies label, there can be populations of individuals within a species (or even within a subspecies) that have a recognizable identity over time.

    The U.S. Endangered Species Act of 1973 was ahead of its time on this: It has always allowed the listing of “any distinct population segment,” not just fully distinct species. A quarter of U.S.-listed endangered “species” are actually subspecies, including well-known examples like Florida panthers, northern spotted owls and Mexican gray wolves.

    The idea is to consider not just the current survival of a species but its potential evolutionary ability to adapt when faced with environmental changes or emerging diseases in the future. The source of this resilience is its genetic diversity. If a new pest or disease strikes a genetically diverse population, any individuals that happen to be naturally resistant can survive it and reproduce. If all the individuals are genetically alike and lack resistance, the population will die out.

    This is currently happening in North American forests, where tens of millions of ash trees are succumbing to a beetle, the emerald ash borer. Some individual ash trees, however, are genetically resistant to the beetles, and they are the last hope for the species.

    Population structure bears on genetic diversity because sometimes rare variants of genes are only able to survive by getting sequestered in subpopulations more conducive to their survival. Moreover, researchers have discovered that the importance of genetic diversity goes beyond a population’s specific inventory of genetic traits. Recent studies have helped to confirm a proposal by Michael Lynch, an evolutionary biologist at Arizona State University, that the efficiency of natural selection depends on the “effective population size,” which describes the amount of genetic diversity; this dictates how quickly or slowly species (and populations within them) may evolve, which in turn affects how well species may adapt or whether they will splinter into new species.

    That’s why conservationists didn’t simply release the last lonely northern white rhino females into the southern population and call it a day, and why Steiner’s group has been trying to create northern white rhino embryos. Keeping both the northern and southern subspecies in existence separately will better preserve the adaptive potential of the white rhino species as a whole by preserving the unique characteristics and the full genetic diversity of each.

    An Inbreeding Bottleneck

    When a species or population gets too small, it can lose more than its adaptive potential for surviving future threats. It can lose its ability to sustain itself, as was seen in a study of killer whales, or orcas, that was published in March in Nature Ecology & Evolution [below].

    2
    Most of the orcas, or killer whales, that live off the Pacific Northwest coast have rebounded slightly in recent decades, but the “southern resident” population has not. A recent study found that those whales seem to be declining because of inbreeding depression. Credit:Sara Hysong Shimazu/Shutterstock.

    Off the coast of the Pacific Northwest, the killer whales that reign over the food web have had a rough century. Those orcas have seen their food supplies depleted by humans overfishing their staple chinook salmon. Their waters have grown increasingly polluted. And a rising cacophony of underwater noise from commercial vessels now interferes with the mammals’ echolocation. But their circumstances haven’t all been dire: Although these coastal orcas are still listed as threatened, laws in the U.S. and Canada passed in the 1970s have shielded them from hunting and harassment, allowing most — but not all — populations to rebound at least slightly in recent decades.

    The one exception has been a population known as the southern residents. This group of fewer than 100 orcas range from British Columbia to the California coast, and despite having the same protections as their more northerly cousins, they have continued to decline. (They were designated as endangered in 2005.) The reason why was uncovered in the new study: The orcas suffer from a classic case of “inbreeding depression” that hampers their ability to produce fertile young.

    Sprinkled throughout the genomes of even healthy populations are detrimental genes called deleterious recessive alleles. Only organisms that have two copies of these alleles, one from each parent, suffer health issues. Individuals who have only one copy are carriers of the condition. In most populations, the odds that two carriers will get together are relatively low, but if they do, one quarter of their offspring on average will express the defect.

    When populations are so small that close relatives are breeding, individuals are more likely to have the same gene variants as their mate. Carriers are then much more likely to breed with other carriers. The deleterious recessive alleles become increasingly common in the population, as does the expression of the defects they cause.

    Surprisingly, recent research has revealed that not all small populations are at risk of inbreeding depression. Inbreeding seems to be most detrimental to formerly large populations that are now small, as seems to be the case with the southern resident orcas. Organisms that have always had a tiny population, however, sometimes develop a kind of resistance to it.

    One of the tiniest populations in the wild today is the critically endangered vaquita porpoise, which swims only in the northernmost waters of the Gulf of California in Mexico. Its numbers have shrunk to only around 10 individuals because of continued illegal gill net fishing in the area, which is intended to catch similarly sized fish but often ensnares the tiny porpoises. With so few remaining animals, future efforts to save the species might seem to be doomed by inbreeding.

    1
    Only about 10 of the highly endangered vaquita porpoises are left. Surprisingly, because their numbers have always been fairly small, inbreeding depression is not likely to hurt efforts to rebuild their population. Courtesy of Save The Whales.

    Yet when Chris Kyriazis, a conservation biologist at the University of California-Los Angeles, and an international team of scientists recently analyzed the genomes of the known remaining vaquitas and some preserved specimens, they came to a surprising conclusion. As they described last year in Science [below], the remaining vaquitas all seemed to be healthy and reproducing. Hardly any of the individual vaquitas carried copies of deleterious alleles, so inbreeding was essentially a nonissue for them.

    Using population models, the researchers reconstructed what the genetic profile of vaquitas might have been in the past. They deduced that because the vaquita population has always been relatively small, natural selection eliminated most of the harmful genes from the population and reduced the frequency of the remaining ones, long before gill nets became a problem.

    “When populations are historically small, with low numbers going back many tens of thousands of years, the consequences … are underappreciated in conservation,” Kyriazis said. “It prepares you for bottlenecks.”

    The findings underscore the importance of understanding the history of a species as well as its current genetic structure in order to fully understand its diversity. And they show that a better understanding of genetic diversity can reveal insights into a species’ history.

    Genetic Rescue

    Most endangered or threatened species can’t claim the vaquita’s resistance to the harm of inbreeding. For species or populations at the brink of extinction, the best conservation strategy can sometimes be to bring in outsiders to restore healthy dominant alleles to the breeding pool. “By introducing genetically diverse individuals, the evolutionary potential of populations in peril might increase and, in turn, the resilience of the species as a whole,” Steiner said. “This is the notion of genetic rescue.”

    Genetic rescue has had some great successes. By the 1990s, the population of Florida panthers (an endangered subspecies of puma) had dropped to fewer than 30 individuals, and the remaining cats had heart defects, kinked tails, poor sperm quality, genetic diseases and reproductive issues. Then conservationists introduced just eight new female pumas from a different subspecies in West Texas. Today, genetic variation has doubled, and the number of Florida panthers has bounced back to an estimated 120 to 230 individuals.

    4
    Florida panthers were suffering from severely diminished diversity until a prudent program of “genetic rescue” introduced genes from puma individuals in Texas. Credit: Calv-6304/Alamy Stock Photo.

    Outbreeding like this isn’t even unnatural. Studies show that even hybridization between different species is more common in nature than one might expect. Hybrids and populations in boundary zones are often important for protecting species by infusing them with new genes or serving as repositories for genes that maybe aren’t currently advantageous but could be in the future.

    Not everyone is on board with the idea of genetic rescue through interbreeding populations, however. Some of the resistance is idealistic: Within conservation, “a lot of people feel that preserving the ‘pureness’ and evolutionary history of species should be the main goal of conservation strategies,” Steiner said. “But more aggressive conservation actions may be needed when the adaptive potential of populations is at risk.”

    Some of the hesitation over genetic rescue also likely comes from memories of past mistakes, when interbreeding organisms that were too distantly related led to its own problems, called outbreeding depression. Some organisms from different populations look very similar, are physically capable of mating and can even produce living offspring, but the long-term viability or fertility of those hybrids turns out to be poor. For instance, separate populations within a species can have different numbers of chromosomes — it’s a surprisingly common phenomenon in rodents. Mismatches in traits that affect the timing of reproduction can leave a hybrid looking for mates either before or after mating season.

    “It also depends on the traits of the specific population,” said Amro Zayed, a conservation biologist at York University in Toronto. “Are they locally adapted? If they are, they’d be at risk of extinction from admixture.” Because populations are often highly adapted to their own locales, mixed offspring can lack characteristics critical to survival in either of their parents’ habitats.

    All of these problems mean that trying to introduce more genetic diversity to a population too recklessly can sometimes have the counterintuitive effect of reducing that diversity. Conservationists need to weigh carefully whether mixing specific populations will accomplish what they hope.

    5
    Mountain yellow-legged frogs live in two distinct populations in California. Conservationists are now studying whether interbreeding the two might help to preserve them. Credit: Natural History Collection/Alamy Stock Photo.

    For example, the endangered mountain yellow-legged frog (Rana muscosa) lives in two distinct populations in California that biologists refer to as management units. A captive breeding program is now underway to boost their numbers. For the moment the breeders are keeping the frogs from the two units separate, but one of the wild populations is declining faster than the other and is showing signs of inbreeding and low genetic diversity. “There’s a discussion now about whether genetic rescue is an appropriate conservation strategy for this species,” Steiner said.

    Today’s conservationists don’t have to guess at what-ifs, though. With ever more information at their disposal about species’ genetics and traits, they’re well equipped to compare the risk of outbreeding depression with the risk of keeping populations separate, especially when those populations are low on genetic diversity.

    Moreover, researchers are finding increasingly often that species may not be the most important biological unit for conservation. The 2022 World Wildlife Fund Living Planet Report looked at reports on tens of thousands of populations from around the world in recent decades. It found that between 1970 and 2018, the sizes of those populations declined on average by 69%. Declines were even steeper in some parts of the world, with Latin America and the Caribbean seeing losses reaching 94%, mainly due to tropical deforestation. Losses like these are a huge blow to diversity, and they highlight how much adaptive potential is likely being lost in ecosystems worldwide.

    “It’s important not to be overwhelmed, and keep in mind that we do what we can do,” Steiner said. “Monitoring declines of populations is, in my opinion, much more effective than just tracking the number of species that are disappearing, period, because by tracking the decline of populations, you can actually take actions that are more effective.”

    Nature Ecology & Evolution
    Science 2022
    Genome Research

    See the full article here .

    Comments are invited and will be appreciated, especially if the reader finds any errors which I can correct. Use “Reply”.


    five-ways-keep-your-child-safe-school-shootings

    Please help promote STEM in your local schools.

    Stem Education Coalition

    Formerly known as Simons Science News, Quanta Magazine is an editorially independent online publication launched by the Simons Foundation to enhance public understanding of science. Why Quanta? Albert Einstein called photons “quanta of light.” Our goal is to “illuminate science.” At Quanta Magazine, scientific accuracy is every bit as important as telling a good story. All of our articles are meticulously researched, reported, edited, copy-edited and fact-checked.

     

  • richardmitnick 12:38 pm on May 1, 2023 Permalink | Reply
    Tags: "The Computer Scientist Peering Inside AI’s Black Boxes", , , At least 581 AI models involved in medical decisions have received authorization from the Food and Drug Administration., , Cynthia Rudin, Cynthia Rudin wants machine learning models-responsible for increasingly important decisions-to show their work., Cynthia Rudin works to make machine learning more transparent through “interpretable” models., , Machine learning models are incredibly powerful tools. They extract deeply hidden patterns in large data sets that our limited human brains can’t parse., Machine learning was designed to be black box — predictive models that are either too complicated for any human to understand or proprietary., Many algorithms are black boxes — either because they’re proprietary or because they’re too complicated for a human to understand., Quanta Magazine, Rudin and her team set out to prove that even the most complex machine learning models can be transformed into interpretable glass boxes that show their work.   

    From “Quanta Magazine” : “The Computer Scientist Peering Inside AI’s Black Boxes” Cynthia Rudin 

    From “Quanta Magazine”

    4.27.23
    Allison Parshall

    1
    Cynthia Rudin works to make machine learning more transparent through “interpretable” models. Rudin at Duke University, where she studies machine learning. Credit: Alex M. Sanchez for Quanta Magazine.

    Cynthia Rudin wants machine learning models-responsible for increasingly important decisions-to show their work.

    Machine learning models are incredibly powerful tools. They extract deeply hidden patterns in large data sets that our limited human brains can’t parse. These complex algorithms, then, need to be incomprehensible “black boxes,” because a model that we could crack open and understand would be useless. Right?

    That’s all wrong, at least according to Cynthia Rudin, who studies interpretable machine learning at Duke University. She’s spent much of her career pushing for transparent but still accurate models to replace the black boxes favored by her field.

    The stakes are high. These opaque models are becoming more common in situations where their decisions have real consequences, like the decision to biopsy a potential tumor, grant bail or approve a loan application. Today, at least 581 AI models involved in medical decisions have received authorization from the Food and Drug Administration. Nearly 400 of them are aimed at helping radiologists detect abnormalities in medical imaging, like malignant tumors or signs of a stroke.

    Many of these algorithms are black boxes — either because they’re proprietary or because they’re too complicated for a human to understand. “It makes me very nervous,” Rudin said. “The whole framework of machine learning just needs be changed when you’re working with something higher-stakes.”

    But changed to what? Recently, Rudin and her team set out to prove that even the most complex machine learning models, neural networks doing computer vision tasks, can be transformed into interpretable glass boxes that show their work to doctors.

    Rudin, who grew up outside Buffalo, New York, grew to share her father’s love of physics and math — he’s a medical physicist who helped calibrate X-ray machines — but she realized she preferred to solve problems with computers. Now she leads Duke’s Interpretable Machine Learning lab, where she and her colleagues scrutinize the most complex puzzle boxes in machine learning — neural networks — to create accurate models that show their work.

    Quanta spoke with Rudin about these efforts, ethical obligations in machine learning and weird computer poetry. The interviews have been condensed and edited for clarity.

    Did you always dream of being a computer scientist?

    No, definitely not. As a kid, I wanted to be an orchestra conductor, or something like it. And I wanted to be a composer and write music.

    What kind of music?

    That’s the problem. I write French music from the turn of the previous century, like Ravel and Debussy. And then I realized that few people cared about that kind of music, so I decided not to pursue it as a career. As an undergraduate, I wanted to be an applied mathematician — but I went in the opposite direction, which was machine learning.

    When did you begin thinking about interpretability?

    After I graduated, I ended up working at Columbia with the New York City power company, Con Edison. And they were doing real-world work. We were supposed to predict which manholes were going to have a fire or an explosion — at the time, it was about 1% of the manholes in Manhattan every year. I joked that I was always trying to take a picture of myself on the “most likely to explode” manhole — though I never actually did.

    I found out very quickly that this was not a problem that machine learning was helping with, because the data was so messy. They had accounting records dating back to the 1890s. So we processed all the data and turned it into these tiny models that the company could understand and work with. It was interpretable machine learning, though I didn’t know that at the time.

    What did you know about interpretability back then?

    I didn’t really know anything about interpretability because they didn’t teach it to anyone. Machine learning was designed to be black box — predictive models that are either too complicated for any human to understand or proprietary, somebody’s secret sauce. The whole idea was that you didn’t need to deal with the data; the algorithm would handle all that under the hood. It was so elegant, but that just made it very difficult to figure out what was going on.

    2
    Many researchers accept that models of machine learning are “black boxes” and impossible for humans to understand, but Rudin has shown that interpretable variations can work just as well. Credit: Alex M. Sanchez for Quanta Magazine.

    But why does knowing what’s going on under the hood matter?

    If you want to trust a prediction, you need to understand how all the computations work. For example, in health care, you need to know if the model even applies to your patient. And it’s really hard to troubleshoot models if you don’t know what’s in them. Sometimes models depend on variables in ways that you might not like if you knew what they were doing. For example, with the power company in New York, we gave them a model that depended on the number of neutral cables. They looked at it and said, “Neutral cables? That should not be in your model. There’s something wrong.” And of course there was a flaw in the database, and if we hadn’t been able to pinpoint it, we would have had a serious problem. So it’s really useful to be able to see into the model so you can troubleshoot it.

    When did you first get concerned about non-transparent AI models in medicine?

    My dad is a medical physicist. Several years ago, he was going to medical physics and radiology conferences. I remember calling him on my way to work, and he was saying, “You’re not going to believe this, but all the AI sessions are full. AI is taking over radiology.” Then my student Alina [Barnett] roped us into studying [AI models that examine] mammograms. Then I realized, OK, hold on. They’re not using interpretable models. They’re using just these black boxes; then they’re trying to explain their results. Maybe we should do something about this.

    So we decided we would try to prove that you could construct interpretable models for mammography that did not lose accuracy over their black box counterparts. We just wanted to prove that it could be done.

    How do you make a radiology AI that shows its work?

    We decided to use case-based reasoning. That’s where you say, “Well, I think this thing looks like this other thing that I’ve seen before.” It’s like what Dr. House does with his patients in the TV show. Like: “This patient has a heart condition, and I’ve seen her condition before in a patient 20 years ago. This patient is a young woman, and that patient was an old man, but the heart condition is similar.” And so I can reason about this case in terms of that other case.

    We decided to do that with computer vision: “Well, this part of the image looks like that part of that image that I’ve seen before.” This would explain the reasoning process in a way that is similar to how a human might explain their reasoning about an image to another human.

    These are high-complexity models. They’re neural networks. But as long as they’re reasoning about a current case in terms of its relationship to past cases, that’s a constraint that forces the model to be interpretable. And we haven’t lost any accuracy compared to the benchmarks in computer vision.

    Would this ‘Dr. House’ technique work for other areas of health care?

    You could use case-based reasoning for anything. Once we had the mammography project established, my students Alina Barnett and Stark Guo, and a physician collaborator named Brandon Westover, transferred their knowledge directly to EEG scans for critically ill patients. It’s a similar neural architecture, and they trained it within a couple of months, very quick.

    If this approach is just as accurate as black boxes, why not use it for everything?

    Well, first of all, it’s much harder to train an interpretable model, because you have to think about the reasoning process and make sure that’s correct. For low-stakes decisions, it’s not really worth it. Like for advertising, if the ad gets to the right people and makes money, then people tend to be happy. But for high-stakes decisions, I think it’s worth that extra effort.

    Are there other ways to figure out what a neural network is doing?

    Around 2017, people started working on “explainability,” which was explaining the predictions of a black box. So you have some complicated function — like a neural network. You can think about these explanation methods as trying to approximate these functions. Or they might try to pick out which variables are important for a specific prediction.

    And that work has some serious problems with it. The explanations have to be wrong, because if their explanations were always right, you could just replace the black box with the explanations. And so the fact that the explainability people casually claim the same kinds of guarantees that the interpretability people are actually providing made me very uncomfortable, especially when it came to high-stakes decisions. Even with an explanation, you could have your freedom denied if you were a prisoner and truly not understand why. Or you could be denied a loan that would give you a house, and again, you wouldn’t be able to know why. They could give you some crappy explanation, and there’s nothing you could do about it, really.

    Are people taking interpretability more seriously now?

    I think so. It used to be that I would give a talk and some people would come up and yell at me after. And they’d be like, “We don’t need interpretable models; we just test it really carefully and it’s fine.” Now people are coming up afterward and saying, “Yeah, I agree with you, and I’m working on this too.” I think you still have the explainability people ruling the land at the moment — again, it’s easier to poke at a black box than it is to replace it. Those guys I haven’t managed to convince, and I view that as somewhat of a personal failure, but I’m working on it. [Laughs.] I’m hoping that this next generation will help me out.

    Would any low-stakes applications of machine learning benefit from more interpretability?

    People are working on interpretable models for natural language processing. These large language-generation models like ChatGPT are very difficult to understand. We’ve realized now when they say something offensive, it would be useful to know why they did that. It’s really hard to troubleshoot these black box models. Before ChatGPT, I used to run our computer-generated poetry team at Duke. We were working with GPT-2, a predecessor to ChatGPT, and I often felt like we were trying to convince it to do something it really didn’t want to do. It just wasn’t good at figuring out which words generally make sense together.

    Why did you make computer-generated poetry?

    Well, I was hoping to do something meta-creative. The team started with sonnets, then went on to limericks. They wrote this paper called There Once Was a Really Bad Poet, It Was Automated but You Didn’t Know It We forced the model to follow a certain template — like Mad Libs on steroids. There were a whole bunch of poems that were just a riot. It’s so fun when you get some weird piece of poetry that the computer wrote and you’re like, “Wow, that’s pretty funky.”

    But all of this was before ChatGPT, which has no trouble with text generation, even with very difficult constraints like rhyming and iambic pentameter. But ChatGPT taught me something important. If we don’t have interpretability on large scale language and image generation models, they are harder to control, which means they are likely to assist in propagating dangerous misinformation more quickly. So they changed my mind on the value of interpretability — even for low-stakes decisions it seems we need it.

    Do you ever use machine learning to compose music?

    We published a beautiful computer generation algorithm for four-part harmony that is fully interpretable, written by one of my students, Stephen Hahn. All of the co-authors were musicians, and we incorporated music theory into the algorithm. It isn’t a neural network, and it produces beautiful music.

    I mean, when we find a tiny little model for predicting whether someone will have a seizure, I think that’s beautiful, because it’s a very small pattern that someone can appreciate and use. And music is all about patterns. Poetry is all about patterns. They’re all beautiful patterns.

    See the full article here .

    Comments are invited and will be appreciated, especially if the reader finds any errors which I can correct. Use “Reply”.


    five-ways-keep-your-child-safe-school-shootings

    Please help promote STEM in your local schools.

    Stem Education Coalition

    Formerly known as Simons Science News, Quanta Magazine is an editorially independent online publication launched by the Simons Foundation to enhance public understanding of science. Why Quanta? Albert Einstein called photons “quanta of light.” Our goal is to “illuminate science.” At Quanta Magazine, scientific accuracy is every bit as important as telling a good story. All of our articles are meticulously researched, reported, edited, copy-edited and fact-checked.

     

  • richardmitnick 12:22 pm on May 1, 2023 Permalink | Reply
    Tags: "Tiny Jets on the Sun Power the Colossal Solar Wind", A new analysis argues that the solar wind is powered by a collective set of intermittent small-scale jetlike eruptions in the sun’s corona., , , Coronal regions that previously appeared devoid of magnetic flux were in fact filled with complex magnetic fields., Jetlets could be involved with the large-scale events at the sun such as flares and coronal mass ejections., Quanta Magazine, Scientists already knew that the corona was home to small jetlets that typically last for several minutes., , , The team found that the jetlets-each between 1000 and 3000 kilometers wide-are present even during the solar minimum., Torrents of charged particles continuously lift off the sun’s atmosphere and radiate outward at millions of kilometers per hour.   

    From “Quanta Magazine” : “Tiny Jets on the Sun Power the Colossal Solar Wind” 

    From “Quanta Magazine”

    4.24.23
    Theo Nicitopoulos


    NASA’s Solar Dynamics Observatory captured an ultraviolet image of an erupting flare. Credit: NASA.

    Torrents of charged particles continuously lift off the sun’s atmosphere and radiate outward at millions of kilometers per hour, yielding a solar wind so immense that its limit defines the outer edge of our solar system.

    Despite the vast reach of this wind, its formation has long been a puzzle. Now a new analysis argues that the solar wind is powered by a collective set of intermittent, small-scale jetlike eruptions in the sun’s corona, or outer layer. “The idea is similar to how individual clapping sounds in an auditorium become a steady roar as an audience applauds,” said Craig DeForest, a solar physicist at the Southwest Research Institute in Boulder, Colorado, and a co-author of the study.

    While scientists already knew that the corona was home to small jetlets that typically last for several minutes, they had previously discovered only a small number of them, mainly at the base of plumes emerging from cooler, less dense regions of the corona known as coronal holes.

    The new study reveals that they’re ubiquitous. “Once you know how to find them, you see that they are everywhere in basically every structure in the corona all the time,” said co-author Dan Seaton, a solar physicist who is also at the Southwest Research Institute.

    The team found that the jetlets, each between 1,000 and 3,000 kilometers wide, are present even during the solar minimum, the least active phase of the sun’s 11-year cycle — a result that’s consistent with the solar wind’s pervasive nature. “You can randomly pick any day and the jetlets are there, just like the solar wind,” said Nour Raouafi, a solar physicist at the Johns Hopkins University Applied Physics Laboratory and lead author of the study.

    In the paper laying out the new findings, published last month in The Astrophysical Journal [below], the team provides evidence that the jetlets are sparked by a process called magnetic reconnection, which heats and accelerates a plasma of charged particles.

    The researchers suggest that the jetlets then produce waves that heat the corona and enable the plasma to escape the sun’s gravity and coalesce to form the solar wind.

    Figure 1.
    2
    (a) Composite of SDO/AIA and GOES-R/SUVI 171 Å images showing the small-scale activity at the base of the solar corona and its extension to higher altitudes. The maximum extent of the jetlets in the AIA field of view is limited by the instrument sensitivity. Estimates of their occurrence rate and size are also limited by the temporal and spatial resolution of the instrument. The SUVI image maps the structures observed at the coronal base into the solar wind. The accompanying movies illustrate the highly dynamic and continuous nature of this phenomenon. (b) AIA image (171 Å) showing the jetlet structures as elongated features above the solar polar limb. Examples of jetlet events are indicated by the arrows.

    3
    GOES-R. https://www.researchgate.net

    5
    Solar Ultraviolet Imager (SUVI). https://www.goes-r.gov

    “The numbers come out looking promising and show it is really quite possible that jetlets could supply the mass lost by the sun to the solar wind,” said Charles Kankelborg, a solar physicist at Montana State University who was not involved with the study.

    The Engine

    The idea that small-scale, intermittent events could collectively drive the solar wind stems from the work of Eugene Parker, a pioneering solar physicist who died last year.

    In 1988, he suggested that a “swarm of nanoflares” driven by tiny bursts of magnetic reconnection could heat the corona enough to power the wind.

    3
    The astrophotographers Andrew McCarthy and Jason Guenzel created this image out of over 90,000 individual photographs taken through a custom-modified solar telescope. Credit: Andrew McCarthy and Jason Guenzel.

    However, finding evidence of this small-scale reconnection has proved to be elusive due to the low resolution of magnetic measurements.

    For the new study, the researchers examined high-resolution images from a variety of sources, including NASA’s Solar Dynamics Observatory, the GOES-R satellites — best known as weather satellites — and the Goode Solar Telescope at the Big Bear Solar Observatory.

    They found that coronal regions that previously appeared devoid of magnetic flux were in fact filled with complex magnetic fields. The team was also able to link several jetlets to specific reconnection events. The researchers expect that even finer-scale magnetic field data could reveal higher reconnection and jetting rates.

    The team went on to suggest that the jetlets create a specific kind of wave, called Alfvén waves, that heat the corona. Alfvén waves had been thought of as a competing mechanism that might explain the solar wind. But there is a growing view that these processes can work together. “The global presence of these reconnection-driven jetlets provides a natural explanation for both reconnection and Alfvén waves powering the solar wind,” said Judith Karpen, a solar physicist at NASA’s Goddard Space Flight Center.

    Researchers anticipate that upcoming efforts will reveal coronal processes in unprecedented detail. Their hope lies in newer telescopes, such as the Daniel K. Inouye Solar Telescope at the National Solar Observatory, as well as the Solar Orbiter, a joint project of NASA and the European Space Agency that launched in 2020.


    “It could turn out that the spectrum of jetting goes from relatively large events ending with Parker’s nanoflares on the smallest scales,” said Raouafi.

    And jetlets could be involved with the large-scale events at the sun, such as flares and coronal mass ejections, said Jie Zhang, a solar physicist at George Mason University. “Small-scale eruptions may play a role in transforming magnetic configurations into more coherent large-scale structures that can store large amounts of energy prior to erupting,” he said.

    For now, the new jetlet findings have validated the legacy of Parker and his contemporaries. “Here are some observations 30 years later saying they were probably right,” said Kankelborg.

    The Astrophysical Journal

    See the full article here .

    Comments are invited and will be appreciated, especially if the reader finds any errors which I can correct. Use “Reply”.


    five-ways-keep-your-child-safe-school-shootings

    Please help promote STEM in your local schools.

    Stem Education Coalition

    Formerly known as Simons Science News, Quanta Magazine is an editorially independent online publication launched by the Simons Foundation to enhance public understanding of science. Why Quanta? Albert Einstein called photons “quanta of light.” Our goal is to “illuminate science.” At Quanta Magazine, scientific accuracy is every bit as important as telling a good story. All of our articles are meticulously researched, reported, edited, copy-edited and fact-checked.

     

  • richardmitnick 1:21 pm on April 21, 2023 Permalink | Reply
    Tags: "A New Kind of Symmetry Shakes Up Physics", , , Disparate observations physicists had made in the past 40 years were really manifestations of the same lurking symmetry., , Physicists and mathematicians are collaborating to work out the mathematics of these new symmetries., , Quanta Magazine, The “higher symmetries” work like a one-way street-a notable contrast to all other symmetries in physics., The most important symmetries of 20th-century physics could be extended more broadly to apply in quantum field theory-the basic theoretical framework in which physicists work today.   

    From “Quanta Magazine” : “A New Kind of Symmetry Shakes Up Physics” 

    From “Quanta Magazine”

    4.18.23
    Kevin Hartnett

    1
    The symmetries of 20th century physics were built on points. Higher symmetries are based on one-dimensional lines. Credit: Samuel Velasco/Quanta Magazine

    It is not an exaggeration to say that every major advance in physics for more than a century has turned on revelations about symmetry. It’s there at the dawn of General Relativity, in the birth of the Standard Model, in the hunt for the Higgs.

    For that reason, research across physics is now building to a crescendo. It was touched off by a 2014 paper which demonstrated that the most important symmetries of 20th-century physics could be extended more broadly to apply in quantum field theory, the basic theoretical framework in which physicists work today.

    This reformulation, which crystallized earlier work in the area, revealed that disparate observations physicists had made in the past 40 years were really manifestations of the same lurking symmetry. In doing so, it created an organizing principle that physicists could use to categorize and understand phenomena. “That’s really a stroke of genius,” said Nathaniel Craig, a physicist at the University of California-Santa Barbara.

    The principle identified in the paper came to be known as “higher symmetries.” The name reflects the way the symmetries apply to higher-dimensional objects such as lines, rather than lower-dimensional objects such as particles at single points in space. By giving the symmetry a name and language and by identifying places it had been observed before, the paper prompted physicists to search for other places it might appear.

    Physicists and mathematicians are collaborating to work out the mathematics of these new symmetries — and in some cases they’re discovering that the symmetries work like a one-way street, a notable contrast to all other symmetries in physics. At the same time, physicists are applying the symmetries to explain a wide range of questions, from the decay rate of certain particles to novel phase transitions like the fractional quantum Hall effect.

    “By putting a different perspective on a known sort of physical problem, it just opened up a huge new area,” said Sakura Schafer-Nameki, a physicist at the University of Oxford (UK).

    Symmetry Matters

    To understand why a paper that merely points out the breadth of lurking symmetries can make such a big impact, it helps to first understand how symmetry makes life easier for physicists. Symmetry means fewer details to keep track of. That’s true whether you’re doing high-energy physics or laying bathroom tile.

    The symmetries of a bathroom tile are spatial symmetries — each can be rotated, flipped upside down or moved to a new spot. Spatial symmetries play an important simplifying role in physics too. They’re prominent in Einstein’s theory of space-time — and the fact that they pertain to our universe means physicists have one less thing to worry about.

    “If you’re doing an experiment in a lab and you rotate it, that shouldn’t change your answer,” said Nathan Seiberg, a theoretical physicist at the Institute for Advanced Study in Princeton, New Jersey.

    2
    Nathan Seiberg was a co-author on the 2014 paper that developed the notion of higher symmetries. Credit: Andrea Kane/Institute for Advanced Study.

    The symmetries that are most important in physics today are subtler than spatial symmetries, but they carry the same meaning: They’re constraints on the ways that you can transform something to ensure that it’s still the same.

    In an epochal insight in 1915, the mathematician Emmy Noether formalized the relationship between symmetries and conservation laws.

    4
    Mathematician Emmy Noether. Symmetry.

    For example, symmetries in time — it doesn’t matter if you run your experiment today or tomorrow — mathematically imply the law of conservation of energy. Rotational symmetries lead to the law of conservation of angular momentum.

    “Every conservation law is associated with a symmetry, and every symmetry is associated with a conservation law,” Seiberg said. “It’s well understood and it’s very deep.”

    This is just one of the ways that symmetries help physicists understand the universe.

    Physicists would like to create a taxonomy of physical systems, classifying like with like, in order to know when insights from one can be applied to another. Symmetries are a good organizing principle: All systems exhibiting the same symmetry go in the same bucket.

    Furthermore, if physicists know a system possesses a given symmetry, they can avoid a lot of the mathematical work of describing how it behaves. The symmetries constrain the possible states of the system, which means they limit the potential answers to the complicated equations that characterize the system.

    “Typically, some random physical equations are unsolvable, but if you have enough symmetry, then the symmetry constrains the possible answers. You can say the solution must be this because it’s the only symmetric thing,” said Theo Johnson-Freyd of the Perimeter Institute for Theoretical Physics in Waterloo, Canada.

    Symmetries convey elegance, and their presence can be obvious in hindsight. But until physicists identify their influence, related phenomena can remain distinct. Which is what happened with a host of observations physicists made starting in the early 1970s.

    Fields and Strings

    The conservation laws and symmetries of 20th-century physics take pointlike particles as their primary objects. But in modern quantum field theories, quantum fields are the most basic objects, and particles are just fluctuations in these fields. And within these theories it’s often necessary to go beyond points and particles to think about one-dimensional lines, or strings (which are conceptually distinct from the strings in string theory).

    In 1973, physicists described [Nuclear Physics B (below)] an experiment that involved placing a superconducting material between poles of a magnet. They observed that as they increased the strength of the magnetic field, particles arranged themselves along one-dimensional superconducting threads running between the magnetic poles.

    The next year Kenneth Wilson identified strings — Wilson lines [Physical Review D (below)]— in the setting of classical electromagnetism. Strings also appear in the way the strong force acts among quarks, which are the elementary particles that make up a proton. Separate a quark from its antiquark, and a string forms between them that pulls them back together.

    The point is that strings play an important role in many areas of physics. At the same time, they’re mismatched to traditional conservation laws and symmetries, which are expressed in terms of particles.

    “The modern thing is to say we’re not only interested in the properties of points; we’re interested in the properties of lines or strings, and there can also be conservation laws for them,” said Seiberg, who co-wrote the 2014 paper along with Davide Gaiotto of the Perimeter Institute, Anton Kapustin of the California Institute of Technology, and Brian Willett, who was at the time a postdoc at the Institute for Advanced Study.

    The paper presented a way of measuring charge along a string and establishing that charge remains conserved as the system evolves, just as total charge is always conserved for particles. And the team did it by shifting their attention from the string itself.

    Seiberg and his colleagues imagined the one-dimensional string as being surrounded by a surface, a two-dimensional plane, so that it looked like a line drawn on a sheet of paper. Instead of measuring charge along the string, they described a method for measuring the total charge across the surface surrounding the string.

    “The really new thing is you emphasize the charged object, and you think about [surfaces] that surround it,” Schafer-Nameki said.

    The four authors then considered what happens to the surrounding surface as the system evolves. Maybe it warps or twists or otherwise changes from the completely flat surface they measured originally. Then they demonstrated that even as the surface deforms, the total charge along it remains the same.

    That is, if you measure charge at every point on a piece of paper, then distort the paper and measure again, you’ll get the same number. You can say that charge is conserved along the surface, and since the surface is indexed to the string, you can say it’s conserved along the string, too — regardless of what kind of string you started with.

    “The mechanics of a superconducting string and a strong-force string are completely different, yet the mathematics of these strings and the conservation [laws] are exactly the same,” Seiberg said. “That’s the beauty of this whole idea.”

    Equivalent Surfaces

    The suggestion that a surface remains the same — has the same charge — even after it’s deformed echoes concepts from the mathematical field of topology. In topology, mathematicians classify surfaces according to whether one can be deformed into the other without any ripping. According to this viewpoint, a perfect sphere and a lopsided ball are equivalent, since you can inflate the ball to get the sphere. But a sphere and an inner tube are not, as you’d have to gash the sphere to get the inner tube.

    Similar thinking about equivalence applies to surfaces around strings — and by extension, the quantum field theories inside of which those surfaces are drawn, Seiberg and his co-authors wrote. They referred to their method of measuring charge on surfaces as a topological operator. The word “topological” conveys that sense of overlooking insignificant variations between a flat surface and a warped one. If you measure the charge on each, and it comes out the same, you know that the two systems can be smoothly deformed into each other.

    Topology allows mathematicians to look past minor variations to focus on fundamental ways in which different shapes are the same. Similarly, higher symmetries provide physicists with a new way of indexing quantum systems, the authors concluded. Those systems may look completely different from each other, but in a deep way they might really obey the same rules. Higher symmetries can detect that — and by detecting it, they allow physicists to take knowledge about better-understood quantum systems and apply it to others.

    “The development of all these symmetries is like developing a series of ID numbers for a quantum system,” said Shu-Heng Shao, a theoretical physicist at Stony Brook University. “Sometimes two seemingly unrelated quantum systems turn out to have the same set of symmetries, which suggests they might be the same quantum system.”

    Despite these elegant insights about strings and symmetries in quantum field theories, the 2014 paper didn’t spell out any dramatic ways of applying them. Equipped with new symmetries, physicists might hope to be able to answer new questions — but at the time, higher symmetries were only immediately useful for re-characterizing things physicists already knew. Seiberg recalls being disappointed that they couldn’t do more than that.

    “I remember going around thinking, ‘We need a killer app,’” he said.

    From New Symmetries to New Mathematics

    To write a killer app, you need a good programming language. In physics, mathematics is that language, explaining in a formal, rigorous way how symmetries work together. Following the landmark paper, mathematicians and physicists started by investigating how higher symmetries could be expressed in terms of objects called groups, which are the main mathematical structure used to describe symmetries.

    A group encodes all the ways the symmetries of a shape or a system can be combined. It establishes the rules for how the symmetries operate and tells you what positions the system can end up in following symmetry transformations (and which positions, or states, can never occur).

    Group encoding work is expressed in the language of algebra. In the same way that order matters when you’re solving an algebraic equation (dividing 4 by 2 is not the same as dividing 2 by 4), the algebraic structure of a group reveals how order matters when you’re applying symmetry transformations, including rotations.

    “Understanding algebraic relationships between transformations is a precursor to any application,” said Clay Córdova of the University of Chicago. “You can’t understand how the world is constrained by rotations until you understand ‘What are rotations?’”

    4
    Merrill Sherman/Quanta Magazine.

    By investigating those relationships, two separate teams — one involving Córdova and Shao and one that includes researchers at Stony Brook and the University of Tokyo — discovered that even in realistic quantum systems, there are non-invertible symmetries that fail to conform to the group structure, a feature that every other important type of symmetry in physics fits into. Instead, these symmetries are described by related objects called categories which have more relaxed rules for how symmetries can be combined.

    For example, in a group, every symmetry is required to have an inverse symmetry — an operation that undoes it and sends the object it acts on back to where it started. But in separate papers published last year, the two groups showed that some higher symmetries are non-invertible, meaning once you apply them to a system, you can’t get back to where you started.

    This non-invertibility reflects the way that a higher symmetry can transform a quantum system into a superposition of states, in which it is probabilistically two things at once. From there, there’s no road back to the original system. To capture this more complicated way higher symmetries and non-invertible symmetries interact, researchers including Johnson-Freyd have developed a new mathematical object called a higher fusion category.

    “It’s the mathematical edifice that describes the fusions and interactions of all these symmetries,” Córdova said. “It tells you all the algebraic possibilities for how they can interact.”

    Higher fusion categories help to define the non-invertible symmetries that are mathematically possible, but they don’t tell you which symmetries are useful in specific physical situations. They establish the parameters of a hunt on which physicists then embark.

    “As a physicist the exciting thing is the physics we get out of it. It shouldn’t just be math for the sake of math,” Schafer-Nameki said.

    Early Applications

    Equipped with higher symmetries, physicists are also reevaluating old cases in light of new evidence.

    For example, in the 1960s physicists noticed a discrepancy in the decay rate of a particle called the pion. Theoretical calculations said it should be one thing, experimental observations said another. In 1969, two papers seemed to resolve the tension by showing that the quantum field theory which governs pion decay does not actually possess a symmetry that physicists thought it did. Without that symmetry, the discrepancy disappeared.

    But last May, three physicists proved that the 1969 verdict was only half the story. It wasn’t just that the presupposed symmetry wasn’t there — it was that higher symmetries were. And when those symmetries were incorporated into the theoretical picture, the predicted and observed decay rates matched exactly.

    “We can reinterpret this mystery of the pion decay not in terms of the absence of symmetry but in terms of the presence of a new kind of symmetry,” said Shao, a co-author of the paper.

    Similar reexamination has taken place in condensed matter physics. Phase transitions occur when a physical system switches from one state of matter to another. At a formal level, physicists describe those changes in terms of symmetries being broken: Symmetries that pertained in one phase no longer apply in the next.

    But not all phases have been neatly described by symmetry-breaking. One, called the fractional quantum Hall effect, involves the spontaneous reorganization of electrons, but without any apparent symmetry being broken. This made it an uncomfortable outlier within the theory of phase transitions. That is, until a paper in 2018 by Xiao-Gang Wen of the Massachusetts Institute of Technology helped establish that the quantum Hall effect does in fact break a symmetry — just not a traditional one.

    “You can think of [it] as symmetry-breaking if you generalize your notion of symmetry,” said Ashvin Vishwinath of Harvard University.

    These early applications of higher and non-invertible symmetries — to the pion decay rate, and to the understanding of the fractional quantum Hall effect — are modest compared to what physicists anticipate.

    In condensed matter physics, researchers hope that higher and non-invertible symmetries will help them with the fundamental task of identifying and classifying all possible phases of matter. And in particle physics, researchers are looking to higher symmetries to assist with one of the biggest open questions of all: what principles organize physics beyond the Standard Model.

    “I want to get the Standard Model out of a consistent theory of quantum gravity, and these symmetries play a critical role,” said Mirjam Cvetic of the University of Pennsylvania.

    It will take a while to fully reorient physics around an expanded understanding of symmetry and a broader notion of what makes systems the same. That so many physicists and mathematicians are joining in the effort suggests they think it will be worth it.

    “I have not yet seen shocking results that we didn’t know before, but I have no doubt it’s quite likely this will happen, because this is clearly a much better way of thinking about the problem,” Seiberg said.

    Nuclear Physics B 1973
    Physical Review D 1974

    See the full article here .

    Comments are invited and will be appreciated, especially if the reader finds any errors which I can correct. Use “Reply”.


    five-ways-keep-your-child-safe-school-shootings

    Please help promote STEM in your local schools.

    Stem Education Coalition

    Formerly known as Simons Science News, Quanta Magazine is an editorially independent online publication launched by the Simons Foundation to enhance public understanding of science. Why Quanta? Albert Einstein called photons “quanta of light.” Our goal is to “illuminate science.” At Quanta Magazine, scientific accuracy is every bit as important as telling a good story. All of our articles are meticulously researched, reported, edited, copy-edited and fact-checked.

     

  • richardmitnick 4:31 pm on April 14, 2023 Permalink | Reply
    Tags: "A New Approach to Computation Reimagines Artificial Intelligence", "Hyperdimensional computing" - each piece of information is represented as a single entity: a hyperdimensional vector., , , Information in the brain is represented by the activity of numerous neurons-thousands of neurons., Modern ANNs are elaborate networks of computational units-trained to do specific tasks., Quanta Magazine, Such systems are so complicated that no one truly understands what they’re doing or why they work so well., The artificial neural networks ("ANNs") that underpin ChatGPT and other large language model systems might be on the wrong track., The limitations of ANNs have long been obvious., The strengths of hyperdimensional computing lie in the ability to compose and decompose hypervectors for reasoning.   

    From “Quanta Magazine” : “A New Approach to Computation Reimagines Artificial Intelligence” 

    From “Quanta Magazine”

    4.13.23
    Anil Ananthaswamy

    1
    Myriam Wares for Quanta Magazine.

    Despite the wild success of ChatGPT and other large language models, the artificial neural networks (“ANNs”) that underpin these systems might be on the wrong track.

    For one, “ANNs” are “super power-hungry,” said Cornelia Fermüller, a computer scientist at the University of Maryland. “And the other issue is [their] lack of transparency.” Such systems are so complicated that no one truly understands what they’re doing, or why they work so well. This, in turn, makes it almost impossible to get them to reason by analogy, which is what humans do — using symbols for objects, ideas and the relationships between them.

    Such shortcomings likely stem from the current structure of ANNs and their building blocks: individual artificial neurons. Each neuron receives inputs, performs computations and produces outputs. Modern ANNs are elaborate networks of these computational units, trained to do specific tasks.

    Yet the limitations of ANNs have long been obvious. Consider, for example, an ANN that tells circles and squares apart. One way to do it is to have two neurons in its output layer, one that indicates a circle and one that indicates a square. If you want your ANN to also discern the shape’s color — blue or red — you’ll need four output neurons: one each for blue circle, blue square, red circle and red square. More features mean even more neurons.

    This can’t be how our brains perceive the natural world, with all its variations. “You have to propose that, well, you have a neuron for all combinations,” said Bruno Olshausen, a neuroscientist at the University of California-Berkeley. “So, you’d have in your brain, [say,] a purple Volkswagen detector.”

    Instead, Olshausen and others argue that information in the brain is represented by the activity of numerous neurons. So the perception of a purple Volkswagen is not encoded as a single neuron’s actions, but as those of thousands of neurons. The same set of neurons, firing differently, could represent an entirely different concept (a pink Cadillac, perhaps).

    This is the starting point for a radically different approach to computation known as “hyperdimensional computing”. The key is that each piece of information, such as the notion of a car, or its make, model or color, or all of it together, is represented as a single entity: a hyperdimensional vector.

    A vector is simply an ordered array of numbers. A 3D vector, for example, comprises three numbers: the x, y and z coordinates of a point in 3D space. A hyperdimensional vector, or hypervector, could be an array of 10,000 numbers, say, representing a point in 10,000-dimensional space. These mathematical objects and the algebra to manipulate them are flexible and powerful enough to take modern computing beyond some of its current limitations and foster a new approach to artificial intelligence.

    “This is the thing that I’ve been most excited about, practically in my entire career,” Olshausen said. To him and many others, hyperdimensional computing promises a new world in which computing is efficient and robust, and machine-made decisions are entirely transparent.

    Enter High-Dimensional Spaces

    To understand how hypervectors make computing possible, let’s return to images with red circles and blue squares. First we need vectors to represent the variables SHAPE and COLOR. Then we also need vectors for the values that can be assigned to the variables: CIRCLE, SQUARE, BLUE and RED.

    The vectors must be distinct. This distinctness can be quantified by a property called orthogonality, which means to be at right angles. In 3D space, there are three vectors that are orthogonal to each other: One in the x direction, another in the y and a third in the z. In 10,000-dimensional space, there are 10,000 such mutually orthogonal vectors.

    But if we allow vectors to be nearly orthogonal, the number of such distinct vectors in a high-dimensional space explodes. In a 10,000-dimensional space there are millions of nearly orthogonal vectors.

    Now let’s create distinct vectors to represent SHAPE, COLOR, CIRCLE, SQUARE, BLUE and RED. Because there are so many possible nearly orthogonal vectors in a high-dimensional space, you can just assign six random vectors to represent the six items; they’re almost guaranteed to be nearly orthogonal. “The ease of making nearly orthogonal vectors is a major reason for using hyperdimensional representation,” wrote Pentti Kanerva, a researcher at the Redwood Center for Theoretical Neuroscience at the University of California-Berkeley, in an influential 2009 paper [Cognitive Computation].

    2
    Pentti Kanerva (left) and Bruno Olshausen, researchers at the University of California-Berkeley.

    The paper built upon work done in the mid-1990s by Kanerva and Tony Plate, at the time a doctoral student with Geoff Hinton at the University of Toronto. The two independently developed the algebra for manipulating hypervectors and hinted at its usefulness for high-dimensional computing.

    Given our hypervectors for shapes and colors, the system developed by Kanerva and Plate shows us how to manipulate them using certain mathematical operations. Those actions correspond to ways of symbolically manipulating concepts.

    The first operation is multiplication. This is a way of combining ideas. For example, multiplying the vector SHAPE with the vector CIRCLE binds the two into a representation of the idea “SHAPE is CIRCLE.” This new “bound” vector is nearly orthogonal to both SHAPE and CIRCLE. And the individual components are recoverable — an important feature if you want to extract information from bound vectors. Given a bound vector that represents your Volkswagen, you can unbind and retrieve the vector for its color: PURPLE.

    The second operation, addition, creates a new vector that represents what’s called a superposition of concepts. For example, you can take two bound vectors, “SHAPE is CIRCLE” and “COLOR is RED,” and add them together to create a vector that represents a circular shape that is red in color. Again, the superposed vector can be decomposed into its constituents.

    The third operation is permutation; it involves rearranging the individual elements of the vectors. For example, if you have a three-dimensional vector with values labeled x, y and z, permutation might move the value of x to y, y to z, and z to x. “Permutation allows you to build structure,” Kanerva said. “It allows you to deal with sequences, things that happen one after another.” Consider two events, represented by the hypervectors A and B. We can superpose them into one vector, but that would destroy information about the order of events. Combining addition with permutation preserves the order; the events can be retrieved in order by reversing the operations.

    Together, these three operations proved enough to create a formal algebra of hypervectors that allowed for symbolic reasoning. But many researchers were slow to grasp the potential of hyperdimensional computing, including Olshausen. “It just didn’t sink in,” he said.

    Harnessing the Power

    In 2018, a student of Olshausen’s named Eric Weiss demonstrated one aspect of hyperdimensional computing’s unique abilities. Weiss figured out how to represent a complex image as a single hyperdimensional vector that contains information about all the objects in the image, including their properties, such as colors, positions and sizes.

    “I practically fell out of my chair,” Olshausen said. “All of a sudden the lightbulb went on.”

    Soon more teams began developing hyperdimensional algorithms to replicate simple tasks that deep neural networks had begun tackling about two decades before, such as classifying images.

    Consider an annotated data set that consists of images of handwritten digits. An algorithm analyzes the features of each image using some predetermined scheme. It then creates a hypervector for each image. Next, the algorithm adds the hypervectors for all images of zero to create a hypervector for the idea of zero. It then does the same for all digits, creating 10 “class” hypervectors, one for each digit.

    Now the algorithm is given an unlabeled image. It creates a hypervector for this new image, then compares the hypervector against the stored class hypervectors. This comparison determines the digit that the new image is most similar to.

    Yet this is just the beginning. The strengths of hyperdimensional computing lie in the ability to compose and decompose hypervectors for reasoning. The latest demonstration of this came in March, when Abbas Rahimi and colleagues at IBM Research in Zurich used hyperdimensional computing with neural networks to solve a classic problem [Nature Machine Intelligence (below)] in abstract visual reasoning — a significant challenge for typical ANNs, and even some humans.

    4
    Abbas Rahimi, a computer scientist at IBM Research in Zurich. Courtesy of Abbas Rahimi.

    Known as Raven’s progressive matrices, the problem presents images of geometric objects in, say, a 3-by-3 grid. One position in the grid is blank. The subject must choose, from a set of candidate images, the image that best fits the blank.

    “We said, ‘This is really … the killer example for visual abstract reasoning, let’s jump in,’” Rahimi said.

    To solve the problem using hyperdimensional computing, the team first created a dictionary of hypervectors to represent the objects in each image; each hypervector in the dictionary represents an object and some combination of its attributes. The team then trained a neural network to examine an image and generate a bipolar hypervector — an element can be +1 or −1 — that’s as close as possible to some superposition of hypervectors in the dictionary; the generated hypervector thus contains information about all the objects and their attributes in the image. “You guide the neural network to a meaningful conceptual space,” Rahimi said.

    Once the network has generated hypervectors for each of the context images and for each candidate for the blank slot, another algorithm analyzes the hypervectors to create probability distributions for the number of objects in each image, their size, and other characteristics. These probability distributions, which speak to the likely characteristics of both the context and candidate images, can be transformed into hypervectors, allowing the use of algebra to predict the most likely candidate image to fill the vacant slot.

    Their approach was nearly 88% accurate on one set of problems, whereas neural network–only solutions were less than 61% accurate. The team also showed that, for 3-by-3 grids, their system was almost 250 times faster than a traditional method that uses rules of symbolic logic to reason, since that method must search through an enormous rulebook to determine the correct next step.

    A Promising Start

    Not only does hyperdimensional computing give us the power to solve problems symbolically, it also addresses some niggling issues of traditional computing. The performance of today’s computers degrades rapidly if errors caused by, say, a random bit flip (a 0 becomes 1 or vice versa) cannot be corrected by built-in error-correcting mechanisms. Moreover, these error-correcting mechanisms can impose a penalty on performance of up to 25%, said Xun Jiao, a computer scientist at Villanova University.

    Hyperdimensional computing tolerates errors better, because even if a hypervector suffers significant numbers of random bit flips, it is still close to the original vector. This implies that any reasoning using these vectors is not meaningfully impacted in the face of errors. Jiao’s team has shown [IEEEXplore (below)] that these systems are at least 10 times more tolerant of hardware faults than traditional ANNs, which themselves are orders of magnitude more resilient than traditional computing architectures. “We can leverage all [that] resilience to design some efficient hardware,” Jiao said.

    Another advantage of hyperdimensional computing is transparency: The algebra clearly tells you why the system chose the answer it did. The same is not true for traditional neural networks. Olshausen, Rahimi and others are developing hybrid systems in which neural networks map things in the physical world to hypervectors, and then hyperdimensional algebra takes over. “Things like analogical reasoning just fall in your lap,” Olshausen said. “This is what we should expect of any AI system. We should be able to understand it just like we understand an airplane or a television set.”

    All of these benefits over traditional computing suggest that hyperdimensional computing is well suited for a new generation of extremely sturdy, low-power hardware. It’s also compatible with “in-memory computing systems,” which perform the computing on the same hardware that stores data (unlike existing von Neumann computers that inefficiently shuttle data between memory and the central processing unit). Some of these new devices can be analog, operating at very low voltages, making them energy-efficient but also prone to random noise. For von Neumann computing, this randomness is “the wall that you can’t go beyond,” Olshausen said. But with hyperdimensional computing, “you can just punch through it.”

    Despite such advantages, hyperdimensional computing is still in its infancy. “There’s real potential here,” Fermüller said. But she points out that it still needs to be tested against real-world problems and at bigger scales, closer to the size of modern neural networks.

    “For problems at scale, this needs very efficient hardware,” Rahimi said. “For example, how [do you] efficiently search over 1 billion items?”

    All of this should come with time, Kanerva said. “There are other secrets [that] high-dimensional spaces hold,” he said. “I see this as the very beginning of time for computing with vectors.”

    Cognitive Computation 2009
    Nature Machine Intelligence
    IEEEXplore 2021

    See the full article here .

    Comments are invited and will be appreciated, especially if the reader finds any errors which I can correct. Use “Reply”.


    five-ways-keep-your-child-safe-school-shootings

    Please help promote STEM in your local schools.

    Stem Education Coalition

    Formerly known as Simons Science News, Quanta Magazine is an editorially independent online publication launched by the Simons Foundation to enhance public understanding of science. Why Quanta? Albert Einstein called photons “quanta of light.” Our goal is to “illuminate science.” At Quanta Magazine, scientific accuracy is every bit as important as telling a good story. All of our articles are meticulously researched, reported, edited, copy-edited and fact-checked.

     

  • richardmitnick 11:00 am on March 18, 2023 Permalink | Reply
    Tags: "LLM's": Large language models, "The Unpredictable Abilities Emerging From Large AI Models", , Quanta Magazine   

    From “Quanta Magazine” : “The Unpredictable Abilities Emerging From Large AI Models” 

    From “Quanta Magazine”

    3.16.23
    Stephen Ornes

    Large language models [LLM’s] like ChatGPT are now big enough that they’ve started to display startling, unpredictable behaviors.

    What movie do these emojis describe?

    2

    That prompt was one of 204 tasks chosen last year to test the ability of various large language models (LLMs) — the computational engines behind AI chatbots such as ChatGPT. The simplest LLMs produced surreal responses. “The movie is a movie about a man who is a man who is a man,” one began. Medium-complexity models came closer, guessing The Emoji Movie. But the most complex model nailed it in one guess: Finding Nemo.

    “Despite trying to expect surprises, I’m surprised at the things these models can do,” said Ethan Dyer, a computer scientist at Google Research who helped organize the test. It’s surprising because these models supposedly have one directive: to accept a string of text as input and predict what comes next, over and over, based purely on statistics. Computer scientists anticipated that scaling up would boost performance on known tasks, but they didn’t expect the models to suddenly handle so many new, unpredictable ones.

    Recent investigations like the one Dyer worked on have revealed that LLMs can produce hundreds of “emergent” abilities — tasks that big models can complete that smaller models can’t, many of which seem to have little to do with analyzing text. They range from multiplication to generating executable computer code to, apparently, decoding movies based on emojis. New analyses suggest that for some tasks and some models, there’s a threshold of complexity beyond which the functionality of the model skyrockets. (They also suggest a dark flip side: As they increase in complexity, some models reveal new biases and inaccuracies in their responses.)

    “That language models can do these sort of things was never discussed in any literature that I’m aware of,” said Rishi Bommasani, a computer scientist at Stanford University. Last year, he helped compile a list of dozens of emergent behaviors, including several identified in Dyer’s project. That list continues to grow.

    Now, researchers are racing not only to identify additional emergent abilities but also to figure out why and how they occur at all — in essence, to try to predict unpredictability. Understanding emergence could reveal answers to deep questions around AI and machine learning in general, like whether complex models are truly doing something new or just getting really good at statistics. It could also help researchers harness potential benefits and curtail emergent risks.

    “We don’t know how to tell in which sort of application is the capability of harm going to arise, either smoothly or unpredictably,” said Deep Ganguli, a computer scientist at the AI startup Anthropic.

    The Emergence of Emergence

    Biologists, physicists, ecologists and other scientists use the term “emergent” to describe self-organizing, collective behaviors that appear when a large collection of things acts as one. Combinations of lifeless atoms give rise to living cells; water molecules create waves; murmurations of starlings swoop through the sky in changing but identifiable patterns; cells make muscles move and hearts beat. Critically, emergent abilities show up in systems that involve lots of individual parts. But researchers have only recently been able to document these abilities in LLMs as those models have grown to enormous sizes.

    Language models have been around for decades. Until about five years ago, the most powerful were based on what’s called a recurrent neural network. These essentially take a string of text and predict what the next word will be. What makes a model “recurrent” is that it learns from its own output: Its predictions feed back into the network to improve future performance.

    In 2017, researchers at Google Brain introduced a new kind of architecture called a transformer. While a recurrent network analyzes a sentence word by word, the transformer processes all the words at the same time. This means transformers can process big bodies of text in parallel.

    2
    An illustration showing an orange and blue network of lines focus into a clear pyramid, emerging as a white light traveling into a clear eye. Avalon Nuovo for Quanta Magazine.

    Transformers enabled a rapid scaling up of the complexity of language models by increasing the number of parameters in the model, as well as other factors. The parameters can be thought of as connections between words, and models improve by adjusting these connections as they churn through text during training. The more parameters in a model, the more accurately it can make connections, and the closer it comes to passably mimicking human language. As expected, a 2020 analysis by OpenAI researchers found that models improve in accuracy and ability as they scale up.

    But the debut of LLMs also brought something truly unexpected. Lots of somethings. With the advent of models like GPT-3, which has 175 billion parameters — or Google’s PaLM, which can be scaled up to 540 billion — users began describing more and more emergent behaviors. One DeepMind engineer even reported being able to convince ChatGPT that it was a Linux terminal and getting it to run some simple mathematical code to compute the first 10 prime numbers. Remarkably, it could finish the task faster than the same code running on a real Linux machine.

    As with the movie emoji task, researchers had no reason to think that a language model built to predict text would convincingly imitate a computer terminal. Many of these emergent behaviors illustrate “zero-shot” or “few-shot” learning, which describes an LLM’s ability to solve problems it has never — or rarely — seen before. This has been a long-time goal in artificial intelligence research, Ganguli said. Showing that GPT-3 could solve problems without any explicit training data in a zero-shot setting, he said, “led me to drop what I was doing and get more involved.”

    He wasn’t alone. A raft of researchers, detecting the first hints that LLMs could reach beyond the constraints of their training data, are striving for a better grasp of what emergence looks like and how it happens. The first step was to thoroughly document it.

    3
    Ethan Dyer helped probe what kinds of unexpected abilities large language models were capable of, and what could bring them about. Credit: Gabrielle Lurie.

    Beyond Imitation

    In 2020, Dyer and others at Google Research predicted that LLMs would have transformative effects — but what those effects would be remained an open question. So they asked the research community to provide examples of difficult and diverse tasks to chart the outer limits of what an LLM could do. This effort was called the Beyond the Imitation Game Benchmark (BIG-bench) project, riffing on the name of Alan Turing’s “imitation game,” a test for whether a computer could respond to questions in a convincingly human way. (This would later become known as the Turing test.) The group was especially interested in examples where LLMs suddenly attained new abilities that had been completely absent before.

    “How we understand these sharp transitions is a great research question,” Dyer said.

    As one would expect, on some tasks a model’s performance improved smoothly and predictably as complexity increased. And on other tasks, scaling up the number of parameters did not yield any improvement. But for about 5% of the tasks, the researchers found what they called “breakthroughs” — rapid, dramatic jumps in performance at some threshold scale. That threshold varied based on the task and model.

    For example, models with relatively few parameters — only a few million — could not successfully complete three-digit addition or two-digit multiplication problems, but for tens of billions of parameters, accuracy spiked in some models. Similar jumps occurred for other tasks including decoding the International Phonetic Alphabet, unscrambling a word’s letters, identifying offensive content in paragraphs of Hinglish (a combination of Hindi and English), and generating a similar English equivalent of Kiswahili proverbs.
    Introduction

    But the researchers quickly realized that a model’s complexity wasn’t the only driving factor. Some unexpected abilities could be coaxed out of smaller models with fewer parameters — or trained on smaller data sets — if the data was of sufficiently high quality. In addition, how a query was worded influenced the accuracy of the model’s response. When Dyer and his colleagues posed the movie emoji task using a multiple-choice format, for example, the accuracy improvement was less of a sudden jump and more of a gradual increase with more complexity. And last year, in a paper presented at NeurIPS, the field’s flagship meeting, researchers at Google Brain showed how a model prompted to explain itself (a capacity called chain-of-thought reasoning) could correctly solve a math word problem, while the same model without that prompt could not.

    Yi Tay, a scientist at Google Brain who worked on the systematic investigation of breakthroughs, points to recent work suggesting that chain-of-thought prompting changes the scaling curves and therefore the point where emergence occurs. In their NeurIPS paper, the Google researchers showed that using chain-of-thought prompts could elicit emergent behaviors not identified in the BIG-bench study. Such prompts, which ask the model to explain its reasoning, may help researchers begin to investigate why emergence occurs at all.

    Recent findings like these suggest at least two possibilities for why emergence occurs, said Ellie Pavlick, a computer scientist at Brown University who studies computational models of language. One is that, as suggested by comparisons to biological systems, larger models truly do gain new abilities spontaneously. “It may very well be that the model has learned something fundamentally new and different that it didn’t have at a smaller size,” she said. “That’s what we’re all hoping is the case, that there’s some fundamental shift that happens when models are scaled up.”

    The other, less sensational possibility, she said, is that what appears to be emergent may instead be the culmination of an internal, statistics-driven process that works through chain-of-thought-type reasoning. Large LLMs may simply be learning heuristics that are out of reach for those with fewer parameters or lower-quality data.

    But, she said, finding out which of those explanations is more likely hinges on a better understanding of how LLMs work at all. “Since we don’t know how they work under the hood, we can’t say which of those things is happening.”

    Unpredictable Powers and Pitfalls

    There is an obvious problem with asking these models to explain themselves: They are notorious liars. “We’re increasingly relying on these models to do basic work,” Ganguli said, “but I do not just trust these. I check their work.” As one of many amusing examples, in February Google introduced its AI chatbot, Bard. The blog post announcing the new tool shows Bard making a factual error.

    Emergence leads to unpredictability, and unpredictability — which seems to increase with scaling — makes it difficult for researchers to anticipate the consequences of widespread use.

    “It’s hard to know in advance how these models will be used or deployed,” Ganguli said. “And to study emergent phenomena, you have to have a case in mind, and you won’t know until you study the influence of scale what capabilities or limitations might arise.”

    In an analysis of LLMs released last June, researchers at Anthropic looked at whether the models would show certain types of racial or social biases, not unlike those previously reported in non-LLM-based algorithms used to predict which former criminals are likely to commit another crime. That study was inspired by an apparent paradox tied directly to emergence: As models improve their performance when scaling up, they may also increase the likelihood of unpredictable phenomena, including those that could potentially lead to bias or harm.

    “Certain harmful behaviors kind of come up abruptly in some models,” Ganguli said. He points to a recent analysis of LLMs, known as the BBQ benchmark, which showed that social bias emerges with enormous numbers of parameters. “Larger models abruptly become more biased.” Failure to address that risk, he said, could jeopardize the subjects of these models.

    But he offers a counterpoint: When the researchers simply told the model not to rely on stereotypes or social biases — literally by typing in those instructions — the model was less biased in its predictions and responses. This suggests that some emergent properties might also be used to reduce bias. In a paper released in February, the Anthropic team reported on a new “moral self-correction” mode, in which the user prompts the program to be helpful, honest and harmless.

    Emergence, Ganguli said, reveals both surprising potential and unpredictable risk. Applications of these large LLMs are already proliferating, so a better understanding of that interplay will help harness the diversity of abilities of language models.

    “We’re studying how people are actually using these systems,” Ganguli said. But those users are also tinkering, constantly. “We spend a lot of time just chatting with our models,” he said, “and that is actually where you start to get a good intuition about trust — or the lack thereof.”

    See the full article here .

    Comments are invited and will be appreciated, especially if the reader finds any errors which I can correct. Use “Reply”.


    five-ways-keep-your-child-safe-school-shootings

    Please help promote STEM in your local schools.

    Stem Education Coalition

    Formerly known as Simons Science News, Quanta Magazine is an editorially independent online publication launched by the Simons Foundation to enhance public understanding of science. Why Quanta? Albert Einstein called photons “quanta of light.” Our goal is to “illuminate science.” At Quanta Magazine, scientific accuracy is every bit as important as telling a good story. All of our articles are meticulously researched, reported, edited, copy-edited and fact-checked.

     

  • richardmitnick 7:58 pm on March 13, 2023 Permalink | Reply
    Tags: "Shadows in the Big Bang Afterglow Reveal Invisible Cosmic Structures", , , , Blasts from supernovas and accreting supermassive black holes forced the gas away from its dark matter nodes spreading it out so that it was too thin and cold for conventional telescopes to detect., Cosmologists are beginning to look beyond the primary fluctuations in the CMB light to the secondary imprints left by interactions with galaxies and other cosmic structures., , , In the coming months scientists the Atacama Cosmology Telescope plan to unveil an updated map of CMB shadows with a notable jump in both sky coverage and sensitivity., Over the course of its nearly 14-billion-year journey the light from the CMB has been stretched and squeezed and warped by all the matter in its way., Quanta Magazine, Scientists are gaining a crisper view of the distribution of both ordinary matter and the mysterious dark matter., Scientists are repurposing CMB data to catalog the large-scale structures that developed over billions of years as the universe matured., , The CMB was a key piece of evidence that helped to establish the standard model of cosmology but the CMB backlight studies are now threatening to poke holes in that story., The vast majority of the universe’s total matter content is invisible to telescopes but both dark matter and gas leave detectable imprints on the magnification and color of the incoming CMB light.   

    From “Quanta Magazine” : “Shadows in the Big Bang Afterglow Reveal Invisible Cosmic Structures” 

    From “Quanta Magazine”

    3.13.23
    Zack Savitsky

    1
    “The universe is really a shadow theater in which the galaxies are the protagonists, and the CMB is the backlight,” said the cosmologist Emmanuel Schaan. Kristina Armitage/Quanta Magazine .

    Nearly 400,000 years after the Big Bang, the primordial plasma of the infant universe cooled enough for the first atoms to coalesce, making space for the embedded radiation to soar free. That light — the cosmic microwave background (CMB) — continues to stream through the sky in all directions, broadcasting a snapshot of the early universe that’s picked up by dedicated telescopes and even revealed in the static on old cathode-ray televisions.

    After scientists discovered the CMB radiation in 1965, they meticulously mapped its tiny temperature variations, which displayed the exact state of the cosmos when it was a mere frothing plasma. Now they’re repurposing CMB data to catalog the large-scale structures that developed over billions of years as the universe matured.

    “That light experienced a bulk of the history of the universe, and by seeing how it’s changed, we can learn about different epochs,” said Kimmy Wu, a cosmologist at SLAC National Accelerator Laboratory.

    Over the course of its nearly 14-billion-year journey the light from the CMB has been stretched and squeezed and warped by all the matter in its way. Cosmologists are beginning to look beyond the primary fluctuations in the CMB light to the secondary imprints left by interactions with galaxies and other cosmic structures. From these signals, they’re gaining a crisper view of the distribution of both ordinary matter — everything that’s composed of atomic parts — and the mysterious dark matter. In turn, those insights are helping to settle some long-standing cosmological mysteries and pose some new ones.

    “We’re realizing that the CMB does not only tell us about the initial conditions of the universe. It also tells us about the galaxies themselves,” said Emmanuel Schaan, also a cosmologist at SLAC. “And that turns out to be really powerful.”

    A Universe of Shadows

    Standard optical surveys which track the light emitted by stars overlook most of the galaxies’ underlying mass. That’s because the vast majority of the universe’s total matter content is invisible to telescopes — tucked out of sight either as clumps of dark matter or as the diffuse ionized gas that bridges galaxies. But both the dark matter and the strewn gas leave detectable imprints on the magnification and color of the incoming CMB light.

    “The universe is really a shadow theater in which the galaxies are the protagonists, and the CMB is the backlight,” Schaan said.

    Many of the shadow players are now coming into relief.

    When light particles, or photons, from the CMB scatter off electrons in the gas between galaxies, they get bumped to higher energies. In addition, if those galaxies are in motion with respect to the expanding universe, the CMB photons get a second energy shift, either up or down depending on the relative motion of the cluster.

    This pair of effects, known respectively as the thermal and kinematic Sunyaev-Zel’dovich (SZ) effects, were first theorized in the late 1960s [Nature (below)] and have been detected with increasing precision in the past decade. Together, the SZ effects leave a characteristic signature that can be teased out of CMB images, allowing scientists to map the location and temperature of all the ordinary matter in the universe.

    Finally, a third effect known as weak gravitational lensing warps the path of CMB light as it travels near massive objects, distorting the CMB as though it were viewed through the base of a wineglass. Unlike the SZ effects, lensing is sensitive to all matter — dark or otherwise.

    Taken together, these effects allow cosmologists to separate the ordinary matter from the dark matter. Then scientists can overlay these maps with images from galaxy surveys to gauge cosmic distances and even trace star formation.

    2
    Merrill Sherman/Quanta Magazine

    In companion papers in 2021 [Physical Review D (below)] and [Physical Review D (below)], a team led by Schaan and Stefania Amodeo, who is now at the Strasbourg Astronomical Observatory in France, put this approach to work. They examined CMB data taken by the European Space Agency’s Planck satellite and the ground-based Atacama Cosmology Telescope, then stacked on top of those maps an additional optical survey of nearly 500,000 galaxies. The technique allowed them to measure the alignment of ordinary matter and dark matter.

    The analysis showed that the region’s gas didn’t hug its supporting dark matter network as tightly as many models predicted. Instead, it suggests that blasts from supernovas and accreting supermassive black holes forced the gas away from its dark matter nodes, spreading it out so that it was too thin and cold for conventional telescopes to detect.

    Spotting that diffuse gas in CMB shadows has helped scientists further address the so-called missing baryons problem. It has also provided estimates for the strength and temperature of the dispersing blasts — data that scientists are now using to refine their models of galaxy evolution and the large-scale structure of the universe.

    In recent years, cosmologists have been puzzled by the fact that the observed distribution of matter in the modern universe is smoother than theory predicts. If the explosions recycling intergalactic gas are more energetic than scientists assumed, as the recent work by Schaan, Amodeo and others [MNRAS (below)] seems to suggest, these blasts could be partially responsible for having spread matter more evenly across the universe, said Colin Hill, a cosmologist at Columbia University who also works on CMB signatures. In the coming months, Hill and colleagues at the Atacama Cosmology Telescope plan to unveil an updated map of CMB shadows with a notable jump in both sky coverage and sensitivity.

    “We’ve only started to scratch the surface of what you can do with this map,” Hill said. “It’s a sensational improvement over anything that’s come before. It’s hard to believe it’s real.”

    Shades of the Unknown

    The CMB was a key piece of evidence that helped to establish the standard model of cosmology — the central framework that researchers use to understand the origin, composition and shape of the universe. But the CMB backlight studies are now threatening to poke holes in that story.

    “This paradigm really survived the test of precision measurements — until recently,” said Eiichiro Komatsu, a cosmologist at the MPG Institute for Astrophysics who worked to establish the theory as a member of the Wilkinson Microwave Anisotropy Probe, which mapped the CMB between 2001 and 2010. “We may be at the crossroads … of a new model of the universe.”

    For the past two years, Komatsu and colleagues have been investigating hints of a new character on the shadow-theater stage. The signal appears in the polarization, or orientation, of CMB light waves, which the standard model of cosmology says should remain constant on the waves’ journey across the universe. But, as theorized three decades ago by Sean Carroll and colleagues [Physical Review D (below)], that polarization could be rotated by a field of dark matter, dark energy, or some totally new particle. Such a field would cause photons of different polarizations to travel at different speeds and rotate the net polarization of the light, a property known as “birefringence” that’s shared by certain crystals, such as the ones that enable LCD screens. In 2020, Komatsu’s team reported [Physical Review Letters (below)] finding a tiny rotation in the CMB’s polarization — about 0.35 degrees. A follow-up study published last year [Physical Review D (below)] strengthened that earlier result.

    If the polarization study or another result related to the distribution of galaxies is confirmed, it would imply that the universe does not look the same in all directions to all observers. For Hill and many others, both results are tantalizing but not yet definitive. Follow-up studies are underway to investigate these hints and rule out potential confounding effects. Some have even proposed a dedicated “backlight astronomy” spacecraft [Experimental Astronomy (below)]that would further inspect the various shadows.

    “Five to 10 years ago, people thought cosmology was done,” Komatsu said. “That’s changing now. We are entering a new era.”

    Nature
    Physical Review D 2021
    Physical Review D 2021
    MNRAS 2022
    Physical Review D 1990
    Physical Review Letters 2020
    Physical Review D 2022
    Experimental Astronomy 2021

    See the full article here .

    Comments are invited and will be appreciated, especially if the reader finds any errors which I can correct. Use “Reply”.


    five-ways-keep-your-child-safe-school-shootings

    Please help promote STEM in your local schools.

    Stem Education Coalition

    Formerly known as Simons Science News, Quanta Magazine is an editorially independent online publication launched by the Simons Foundation to enhance public understanding of science. Why Quanta? Albert Einstein called photons “quanta of light.” Our goal is to “illuminate science.” At Quanta Magazine, scientific accuracy is every bit as important as telling a good story. All of our articles are meticulously researched, reported, edited, copy-edited and fact-checked.

     

  • richardmitnick 9:15 pm on March 8, 2023 Permalink | Reply
    Tags: "Researchers Argue Black Holes Will Eventually Destroy All Quantum States", , , New calculations suggest that the event horizons around black holes will ‘decohere’ quantum possibilities — even those that are far away., Quanta Magazine,   

    From “Quanta Magazine” : “Researchers Argue Black Holes Will Eventually Destroy All Quantum States” 

    From “Quanta Magazine”

    3.7.23
    Thomas Lewton

    New calculations suggest that the event horizons around black holes will ‘decohere’ quantum possibilities — even those that are far away.

    1
    Black holes effectively observe elementary particles, an effect that echoes John Wheeler’s ideas about the “participatory universe.” Kristina Armitage/Quanta Magazine

    At Princeton University in the early 1970s, the celebrated theoretical physicist John Wheeler could be spotted in seminars or impromptu hallway discussions drawing a big “U.”

    2
    John Wheeler’s “participatory universe” suggests that observers make the universe real. Samuel Velasco/Quanta Magazine; adapted from John Wheeler.

    The letter’s left tip represented the beginning of the universe, where everything was uncertain and all quantum possibilities were happening at the same time. The letter’s right tip, sometimes adorned with an eye, depicted an observer looking back in time, thus bringing the left side of the U into existence.

    In this “participatory universe,” as Wheeler called it, the cosmos expanded and cooled around the U, forming structures and eventually creating observers, like humans and measuring apparatus. By looking back to the early universe, these observers somehow made it real.

    “He would say things like ‘No phenomenon is a true phenomenon until it’s an observed phenomenon,’” said Robert M. Wald, a theoretical physicist at the University of Chicago who was Wheeler’s doctoral student at the time.

    Now, by studying how quantum theory behaves on the horizon of a black hole, Wald and his collaborators have calculated a new effect that is suggestive of Wheeler’s participatory universe. The mere presence of a black hole, they’ve found, is enough to turn a particle’s hazy “superposition” — the state of being in multiple potential states — into a well-defined reality. “It evokes the idea that these black hole horizons are watching,” said co-author Gautam Satishchandran, a theoretical physicist at Princeton.

    “What we have found might be a quantum mechanical realization of [the participatory universe], but where space-time itself plays the role of the observer,” said Daine Danielson, the third author, also at Chicago.

    Theorists are now debating what to read into these watchful black holes. “This seems to be telling us something deep about the way gravity influences measurement in quantum mechanics,” said Sam Gralla, a theoretical astrophysicist at the University of Arizona. But whether this will prove useful for researchers inching toward a complete theory of quantum gravity is still anyone’s guess.

    The effect is one of many uncovered in the past decade by physicists studying what happens when quantum theory is combined with gravity at low energies. For example, theorists have had great success thinking about Hawking radiation, which causes black holes to slowly evaporate. “Subtle effects that we hadn’t really noticed before give us constraints from which we can glean clues about how to go up toward quantum gravity,” said Alex Lupsasca, a theoretical physicist at Vanderbilt University who was not involved in the new research.

    These observant black holes seem to produce an effect that’s “very arresting,” Lupsasca said, “because it feels like somehow it’s deep.”

    Black Holes and Superpositions

    To understand how a black hole could observe the universe, start small. Consider the classic double-slit experiment, in which quantum particles are fired toward two slits in a barrier. Those that pass through are then detected by a screen on the other side.

    At first, each traveling particle seems to appear at random on the screen. But as more particles pass through the slits, a pattern of light and dark stripes emerges. This pattern suggests that each particle behaves like waves that pass through both slits at once. The bands result from the peaks and troughs of the waves either adding together or canceling one another out — a phenomenon called interference.

    Now add a detector to measure which of the two slits the particle passes through. The pattern of light and dark stripes will disappear. The act of observation changes the state of the particle — its wavelike nature disappears entirely. Physicists say that the information gained by the detection apparatus “decoheres” the quantum possibilities into a definite reality.

    Importantly, your detector doesn’t have to be close to the slits to figure out which path the particle took. A charged particle, for example, emits a long-range electric field that might have slightly different strengths depending on whether it went through the right-hand or left-hand slit. Measuring this field from far away will still allow you to gather information about which path the particle took and will thus cause decoherence.

    3
    From left: Robert Wald, Gautam Satishchandran and Daine Danielson.
    Daine Danielson (left); Sheri Lynn / Sara Kauss Photography (center); courtesy of Daine Daneilson (right)

    In 2021, Wald, Satishchandran and Danielson were exploring a paradox brought about when hypothetical observers gather information in this way. They imagined an experimenter called Alice who creates a particle in a superposition. At a later time, she looks for an interference pattern. The particle will only exhibit interference if it hasn’t become too entangled with any outside system while Alice observes it.

    Then along comes Bob, who is attempting to measure the particle’s position from far away by measuring the particle’s long-range fields. According to the rules of causality, Bob shouldn’t be able to influence the outcome of Alice’s experiment, since the experiment should be over by the time the signals from Bob get to Alice. However, by the rules of quantum mechanics, if Bob does successfully measure the particle, it will become entangled with him, and Alice won’t see an interference pattern.

    The trio rigorously calculated that the amount of decoherence due to Bob’s actions is always less than the decoherence that Alice would naturally cause by the radiation she emits (which also becomes entangled with the particle). So Bob could never decohere Alice’s experiment because she would already have decohered it herself. Although an earlier version of this paradox was resolved in 2018 [Physical Review D] with a back-of-the-envelope calculation by Wald and a different team of researchers, Danielson took it one step further.

    He posed a thought experiment to his collaborators: “Why can’t I put [Bob’s] detector behind a black hole?” In such a setup, a particle in a superposition outside the event horizon will emanate fields that cross over the horizon and get detected by Bob on the other side, within the black hole. The detector gains information about the particle, but as the event horizon is a “one-way ticket,” no information can cross back over, Danielson said. “Bob cannot influence Alice from inside of the black hole, so the same decoherence must occur without Bob,” the team wrote in an email to Quanta. The black hole itself must decohere the superposition.

    “In the more poetic language of the participatory universe, it is as if the horizon watches superpositions,” Danielson said.

    Using this insight, they set about working on an exact calculation of how quantum superpositions are affected by the black hole’s space-time. In a paper published in January, they landed on a simple formula that describes the rate at which radiation crosses over the event horizon and so causes decoherence to occur. “That there was an effect at all was, to me, very surprising,” Wald said.

    Hair on the Horizon

    The idea that event horizons gather information and cause decoherence isn’t new. In 2016, Stephen Hawking, Malcolm Perry and Andrew Strominger described how particles crossing over the event horizon could be accompanied by very low-energy radiation that records information about these particles. This insight was suggested as a solution to the black hole information paradox, a profound consequence of Hawking’s earlier discovery that black holes emit radiation.

    The problem was that Hawking radiation drains energy from black holes, causing them to completely evaporate over time. This process would appear to destroy any information that has fallen into the black hole. But in doing so, it would contradict a fundamental feature of quantum mechanics: that information in the universe can’t be created or destroyed.

    The low-energy radiation proposed by the trio would get around this by allowing some information to be distributed in a halo around the black hole and escape. The researchers called the information-rich halo “soft hair.”

    Wald, Satishchandran and Danielson were not investigating the black hole information paradox. But their work makes use of soft hair. Specifically, they showed that soft hair is created not only when particles fall across a horizon, but when particles outside a black hole merely move to a different location. Any quantum superposition outside will become entangled with soft hair on the horizon, giving rise to the decoherence effect they identified. In this way the superposition is recorded as a kind of “memory” on the horizon.

    The calculation is a “concrete realization of soft hair,” said Daniel Carney, a theoretical physicist at The DOE’s Lawrence Berkeley National Laboratory. “It’s a cool paper. It could be a very useful construction for trying to make that idea work in detail.”

    But to Carney and several other theorists working at the forefront of quantum gravity research, this decoherence effect isn’t all that surprising. The long-range nature of the electromagnetic force and gravity means that “it’s hard to keep anything isolated from the rest of the universe,” said Daniel Harlow, a theoretical physicist at the Massachusetts Institute of Technology.

    Total Decoherence

    The authors argue that there is something uniquely “insidious” about this kind of decoherence. Usually, physicists can control decoherence by shielding their experiment from the outside environment. A vacuum, for example, removes the influence of nearby gas molecules. But nothing can shield gravity, so there’s no way to insulate an experiment from gravity’s long-range influence. “Eventually, every superposition will be completely decohered,” Satishchandran said. “There’s no way of getting around it.”

    The authors therefore regard black hole horizons as taking a more active role in decoherence than was previously known. “The geometry of the universe itself, as opposed to the matter within it, is responsible for the decoherence,” they wrote in an email to Quanta.

    Carney disputes this interpretation, saying that the new decoherence effect can also be understood as a consequence of electromagnetic or gravitational fields, in combination with rules set by causality. And unlike Hawking radiation, where the black hole horizon changes over time, in this case the horizon “has no dynamics whatsoever,” Carney said. “The horizon doesn’t do anything, per se; I would not use that language.”

    To not violate causality, superpositions outside the black hole must be decohered at the maximum possible rate that a hypothetical observer inside the black hole could be collecting information about them. “It seems to be pointing toward some new principle about gravity, measurement and quantum mechanics,” Gralla said. “You don’t expect that to happen more than 100 years after gravity and quantum mechanics were formulated.”

    4
    Merrill Sherman/Quanta Magazine

    Intriguingly, this kind of decoherence will occur anywhere there is a horizon that only allows information to travel in one direction, creating the potential for causality paradoxes. The edge of the known universe, called the cosmological horizon, is another example. Or consider the “Rindler horizon,” which forms behind an observer who continuously accelerates and approaches the speed of light, so that light rays can no longer catch up with them. All of these “Killing horizons” (named after the late 19th- early 20th-century German mathematician Wilhelm Killing) cause quantum superpositions to decohere. “These horizons are really watching you in exactly the same way,” Satishchandran said.

    Exactly what it means for the edge of the known universe to watch everything inside the universe isn’t entirely clear. “We don’t understand the cosmological horizon,” Lupsasca said. “It’s super fascinating, but way harder than black holes.”

    In any case, by posing thought experiments like this, where gravity and quantum theory collide, physicists hope to learn about the behavior of a unified theory. “This is likely giving us some more clues about quantum gravity,” Wald said. For example, the new effect may help theorists understand how entanglement is related to space-time.

    “These effects have to be part of the final story of quantum gravity,” Lupsasca said. “Now, are they going to be a crucial clue along the way to gleaning insight into that theory? It’s worth investigating.”

    The Participatory Universe

    As scientists continue to learn about decoherence in all its forms, Wheeler’s concept of the participatory universe is becoming clearer, Danielson said. All particles in the universe, it seems, are in a subtle superposition until they are observed. Definiteness emerges through interactions. “That’s kind of what, I think, Wheeler had in mind,” Danielson said.

    And the finding that black holes and other Killing horizons observe everything, all the time, “whether you like it or not,” is “more evocative” of the participatory universe than the other types of decoherence are, the authors said.

    Not everyone is ready to buy Wheeler’s philosophy on a grand scale. “The idea that the universe observes itself? That sounds a little Jedi for me,” said Lupsasca, who nevertheless agrees that “everything is observing itself all the time through interactions.”

    “Poetically, you could think of it that way,” Carney said. “Personally, I’d just say that the presence of the horizon means that the fields living around it are going to get stuck on the horizon in a really interesting way.”

    When Wheeler first drew the “big U” when Wald was a student in the 1970s, Wald didn’t think much of it. “Wheeler’s idea struck me as not that solidly grounded,” he said.

    And now? “A lot of the stuff he did was enthusiasm and some vague ideas which later turned out to be really on the mark,” Wald said, noting that Wheeler anticipated Hawking radiation long before the effect was calculated.

    “He saw himself as holding out a lamp light to illuminate possible paths for other people to follow.”

    See the full article here .

    Comments are invited and will be appreciated, especially if the reader finds any errors which I can correct. Use “Reply”.


    five-ways-keep-your-child-safe-school-shootings

    Please help promote STEM in your local schools.

    Stem Education Coalition

    Formerly known as Simons Science News, Quanta Magazine is an editorially independent online publication launched by the Simons Foundation to enhance public understanding of science. Why Quanta? Albert Einstein called photons “quanta of light.” Our goal is to “illuminate science.” At Quanta Magazine, scientific accuracy is every bit as important as telling a good story. All of our articles are meticulously researched, reported, edited, copy-edited and fact-checked.

     

← Older posts

c
Compose new post
j
Next post/Next comment
k
Previous post/Previous comment
r
Reply
e
Edit
o
Show/Hide comments
t
Go to top
l
Go to login
h
Show/Hide help
shift + esc
Cancel
%d bloggers like this: