Want to Be a Data Scientist? However, some problems in physics are unknown or … Machine learning has percolated into many scientific disciplines that deal with large data sets—even those that grapple with theoretical data. While laboratory experiments run by biologists, for example, little resemble the research he leads as a theoretical physicist, there are parallels in the techniques they use to analyze data from complex systems. He held previous postdoctoral positions at Columbia University and Princeton University. From physics to machine learning Eight months ago I finished a PhD in theoretical physics. There is enough empirical evidence to believe that nature (including many man-made entities) really does indeed favor universality. However, thanks to insights provided by the seminal work by Kenneth Wilson and others in the last quarter of the previous century, this belief now stands on a healthy foundation of understanding. Enter your search terms then press the return/enter key to submit your query. The question is, how? This process used to derive higher level rulesets from the lower level ones is called the renormalization group flow (I am using this term very loosely). However, before we can get there we will need to develop a much better understanding of machine learning. It fundamentally alters our relationship with information. in being able to show that disparate phenomena emerge from a small set of simple rules. But then we heard a rumor that there is a new game in town: machine learning. Along with its sibling, big data, they threatened to drive scientific theories out of town. In other words, what are the analogs of symmetry, dimensionality and locality in machine learning? Since then it has been observed in a variety of diverse and unrelated places such as the dynamics of complex networks, multi agent systems, the occurrence of pink noise and the bus system of a town in Mexico, to name a few (see here for some interesting examples). Hands-on real-world examples, research, tutorials, and cutting-edge techniques delivered Monday to Thursday. Most reporters on the 2020 campaign beat are men. So why bother with theories at all? Over the past 30 years, Halverson says scientists have increasingly recognized the large number of possibilities in string theory and the potential role of computer science in this field. “Part of the fun of this new direction is that I’m actually talking to a much broader array of scientists than I ever have,” he says. Machine learning can provide the mathematical scaffolding for scientific theories, to which theorists will then add meaning and the bridge to reality. And, in addition, we should see correlations having the same mathematical structure across various unrelated domains. This is the reason why (almost naive) reductionism works so well in most areas of physics. Machine learning encompasses a broad range of algorithms and modeling tools used for a vast array of data processing tasks, which has entered most scientific disciplines in recent years. “The end might not be in sight for theoretical physics,” he said. In these lectures, I will first review the basic building blocks of neural networks and how they People want to keep up in the artificial intelligence age. Scientific theories are what make the world comprehensible, at least for most of us. “Though no complex system out there is going to be a perfect analogy, we might be able to draw some inspiration from what people are doing in other fields.”. In addition to mathematical approaches, Halverson is looking to machine learning to help overcome computational hurdles in string theory. Do we expect thermodynamics to emerge in the final layers of the network? So exciting, in fact, that it is being studied in-depth. We review in a selective way the recent research on the interface between machine learning and physical sciences. “There have been a number of meetings with people from different types of physics and machine learning where we’re all just in a room together talking about different ideas. And, many freshly minted data experts, coming from the less analytical lands of our newly democratized landscape, often seem to conflate theory with preconceived bias. that these various geometries could be folded in on themselves and hidden in our universe. But, even more importantly the starting point was very crucial — for many physical systems the human scale is the one where universality kicks in. Should we be able to constrain the mathematical structure of statistical mechanics from the weights of the network? In fact, the trend of moving from a background in theoretical physics and maths to machine learning is gaining ground with PostDoc researchers, since there is a crossover of functions. Theories are what help me make sense of the world. Make learning your daily ritual. Old school theorists tend to regard this new wave of empiricism as an attack on their profession by the plebs. Halverson is also interacting with leaders in the tech industry to help them engage with physics research and explore potential scientific applications of the techniques they’ve developed. Inthisreview,weattemptatprovidingacoherentse- … Those successes raise new possibilities for machine learning to solve open problems in quantum physics. With these techniques, my group explores low-energy physics in quantum magnets, cold atoms in optical lattices, bosonic fluids, and quantum computers. They also fear that institutions are failing to provide lifelong learning in the new era of automation. Don’t Start With Machine Learning. The latter is not uniquely defined, and there is a number of different suggestions, but typically it is proportional to the number of free parameters in the model. Machine learning can provide the mathematical scaffolding for scientific theories, to which theorists will then add meaning and the bridge to reality. The idea is that eventually, they may be able to parse patterns in this data and understand the implications of these possibilities. Can machine learning help physicists answer puzzling questions in string theory? Yes, machine learning is a tool, but it is a tool like no other. Machine Learning techniques, in particular neural networks, have become an integral part of our lives. By continuing to use the site or closing this banner without changing your cookie settings, you agree to our use of cookies If universality is true, then it would mean that the observed stable correlations in complex systems would be independent of the details of the underlying theory, i.e. This includes conceptual developments in machine learning (ML) motivated by physical … For the universality classes found in physics these properties are usually symmetries, dimensionality and locality. Understanding comes from explanations, and explanations are provided by theories. From 2016 until 2018, he was a Senior Researcher and Head of the statistical machine learning group at Disney Research, first in Pittsburgh and later in Los Angeles. The same applies to timescales. But now you could throw enough data at a large enough neural network and you will have predictions coming out from the other side. between researchers from machine learning and physics at Microsoft’s headquarters outside Seattle. Universality was first observed and studied in the behavior of the thermodynamic variables of disparate systems near continuous phase transitions. and other technologies. One way or the other, our conception of what constitutes an understanding of reality will be shaped by the role that machine learning plays in science. (See here for an introduction to universality and renormalization group). Here’s how. Arora and Behnam Neyshabu r, a Member in the theoretical machine learning program in 2017–18, have been studying questions related to generalization, the phenomenon whereby the machine, after it has been trained with enough samples—images of cats versus dogs, for example—acquires the ability to produce the correct answer even for samples it has never seen before, as long as they are similar enough to the training … Since its beginning, machine learning has been inspired by methods from statistical physics. The second part comes from the observation that the hierarchy of rulesets in physical systems corresponds very nicely with our intuition. Here’s why electronic voting won’t happen anytime soon, At school, at work, and at home among the trees, Inside the lab that tests Northeastern for the coronavirus, Is math really the language of nature? The whole edifice of modern science stands on the shoulders of a web of interconnected theories. James Halverson, an assistant professor of physics at Northeastern, is using data science to study the fundamental laws of physics that govern the universe. The project expanded to include Center for Theoretical Physics postdocs Daniel Hackett and Denis Boyda, NYU Professor Kyle Cranmer, and physics … “To my knowledge, it was the first-ever string theory meeting in collaboration with industry,” Halverson says. These theoretical extra dimensions are hard to visualize, and there are. For example, you might want to train a model to play a specific game and then use the same model to play a completely different game. Your brain is the world’s most proficient accountant. This timeline corresponds very nicely with the hierarchy of rulesets in physical systems. In April, he helped. “String theory is not a settled subject,” he says. The…, Cod has long been a staple of the New England fishery, but this once-plentiful fish has declined in recent decades.…, Despite the increasing prominence of women in American politics, female journalists who are covering the presidential race continue to be…, Opioid and other substance overdoses are officially a public health emergency in the United States. Knowing if a quantum machine-learning algorithm generalizes is a really hard problem, as we don’t have the theoretical tools we need to solve that problem. Consider a thought experiment where a deep neural network is provided with the snapshots of gas atoms along with the value of some complicated function of the thermodynamic variables; and we train the network with the task of predicting the value from the snapshots. For certain kinds of transformations and rulesets, something quite remarkable and unexpected happens; starting from very different initial rulesets you end up with the same final ruleset. This does not mean that machine learning is useless for any problem that can be described using physics-based modeling. When viewed through the prism of universality, this means that deep neural networks provide us with access to a renormalization group flow in the universality class containing the correct underlying theory, which can then be used to constrain the mathematical structure of the underlying theory. The exchange between fields can go in both directions. Machine learning has percolated into many scientific disciplines that deal with large data sets—even those that grapple with theoretical data. Can disease forecasts…, Northeastern’s Roux Institute receives a ‘phenomenal investment’ from the Harold Alfond Foundation. And this is exactly what happens in reality. a virtual hub at the interface of theoretical physics and deep learning. Manuela Veloso, a computer scientist and head of J.P. Morgan AI Research, told this year’s class of doctoral graduates that their education at Northeastern has prepared them to tackle any challenges and uncertainty that they will face in their careers. I started out as a theoretical physicist. The goal of many computational physicists and chemists is the ability to bridge the gap between atomistic length scales of about a few multiples of an Ångström (Å), i. e., 10−10 m, and meso- or macroscopic length scales by virtue of simulations. Over the past 30 years, Halverson says scientists have increasingly recognized the large number of possibilities in string theory and the potential role of computer science in this field. Attempts to prove experimentally concepts of Quantum Machine Learning remain rear and insufficient. In the past few years, researchers like Halverson, who was recently granted a five-year National Science Foundation CAREER award to advance this work, began using cutting-edge data science techniques to study this large set of possibilities. We will need to understand machine learning algorithms from general principles. And that’s a thrill.”. A theory is essentially a set of rules that can be used to derive predictive models of different aspects of phenomena. Harnessing Data Revolution in Quantum Matter ... Quantum Machine Learning in High Energy Physics Sofia Vallecorsa, CERN, 12:00 EDT 27 Jan 2021. Unlike physics, in these fields we do not have the luxury of knowing what the hierarchy of rulesets corresponds to in reality. We will need to understand machine learning algorithms from general principles. For media inquiries, please contact [email protected] The template that we use for building theories is derived largely from physics. Consider a hierarchy of rulesets, with the initial (bottom level) ruleset representing the mathematical structure of a theory and the final (top level) one representing the mathematical structure of the observed stable correlations in data. What squid neurons and an octopus on ecstasy can teach us about ourselves, The next step in particle physics? The rumor dissipated soon enough, because it was based on the false premise that the goal of science is to churn out predictions. There are big puzzles left unanswered, and trying to crack them is what drives us as theoretical physicists,” he says. This website uses cookies and similar technologies to understand your use of our website and give you a better experience. However, before we can get there we will need to develop a much better understanding of machine learning. “String theory is not a settled subject,” says James Halverson, an assistant professor of physics at Northeastern. String theory also predicts that there are extra dimensions beyond the four dimensions that we experience every day: time and three dimensions of space (forward/back, up/down, left/right). James Halverson, an assistant professor of physics at Northeastern, is using data science to study the fundamental laws of physics that govern the universe. Physics, too, has fallen into the artificial intelligence hype with a clutch of researchers using machine learning to deal with complex problems regarding huge amount of data. Applications of machine learning in other scientific fields have also inspired Halverson. This research is published in Physical Review X. References Using data science to learn more about the large set of possibilities in string theory could ultimately help scientists better understand how theoretical physics fits into findings from experimental physics. While Machine Learning itself is now not only a research field but an economically significant and fast growing industry and Quantum Computing is a well established field of both theoretical and experimental research, Quantum Machine Learning remains a purely theoretical field of studies. It is not. Our work has employed Monte Carlo simulations, density matrix renormalization group, and modern machine learning methods. A first-of-its-kind international poll conducted by Northeastern and Gallup finds that majorities of people in the U.S., U.K., and Canada are optimistic about the future of artificial intelligence. I can appreciate, first hand, the power of these algorithms. The hypothesis of universality (or simply universality for brevity) states that rulesets and transformations that are actually found in nature are of the above kind. Due to their versatile nature, they are applied in the private and academic sector with tremendous success. “The way that proteins fold is actually a pretty good analogy to some of the problems that we run into in string theory,” he says. Halverson studies string theory, which predicts that the universe is made up of tiny, thread-like loops of concentrated energy called strings. The situation that we currently encounter in fields such as biology, economics or social sciences is not very different from the above situation. It is not such a strange wish. There is no reason, in principle, to believe otherwise. As a result we could develop a very fruitful feedback between theory and experiment. One can now think of a transformation such that the rulesets at each level are obtained by applying this transformation to the ruleset at the previous level. The final ruleset in this case is called a fixed point and the group of initial rulesets that lead to the same fixed point are said to constitute a universality class. I created my own YouTube algorithm (to stop me wasting time). Halverson is also interacting with leaders in the tech industry to help them engage with physics research and explore potential scientific applications of the techniques they’ve developed. These theoretical extra dimensions are hard to visualize, and there are many possible ways that these various geometries could be folded in on themselves and hidden in our universe. But, in general, they will depend of the specific universality class, and can be determined by carrying out the renormalization group flow of a member of the class. Although theories belonging to a universality class may have very different origins (with respect to the aspect of reality they are trying to explain) and mathematical details, they share some important mathematical properties which puts tight constraints on their mathematical structure. When we discuss the crisis of opioid overdoses, the words we use matter. In statistical learning theory, models are typically characterized through a bound on GG, which is derived based on some notion of model complexity. Mission: The Physics Division Machine Learning group is a cross-cutting effort that connects researchers developing, adapting, and deploying artificial intelligence (AI) and machine learning (ML) solutions to fundamental physics challenges across the HEP frontiers, including theory. Privacy Statement. The goal of science is to provide understanding. Upcoming talks: 02 Dec 2020. Traditionally, making predictions was a complicated business, involving, amongst other things, developing underlying theories for understanding how things work. From the confinement of quarks and gluons into protons to the emergence of spacetime, some of the biggest open questions in quantum field theory could benefit from machine-learning tools. The explanatory power of theories comes from their ability to provide holistic pictures of aspects of reality, i.e. We also know that big things are composed of small things, hence the macroscopic patterns should follow from microscopic theory. Whatisdifficulttodeny is that they produce surprisingly good results in some cases. To come back to the question raised earlier; how can machine learning help theoretical science? String theory also predicts that there are extra dimensions beyond the four dimensions that we experience every day: time and three dimensions of space (forward/back, up/down, left/right). “There are big puzzles left unanswered, and trying to crack them is what drives us as theoretical physicists,” he says. ML applications in physics are becoming an important part of modern experimental high energy analyses. ic Theory Theory-based Models Data Science Models Theory-guided Data Science Models Low High High Low y M s 2. , an assistant professor of physics at Northeastern, is using data science to study the fundamental laws of physics that govern the universe. To find out more about our use of cookies and how to change your settings, please go to our Machine learning has percolated into many scientific disciplines that deal with large data sets—even those that grapple with theoretical data. James Halverson, an assistant professor of physics at Northeastern, uses data science to study the many possibilities in string theory. 5 Reasons You Don’t Need to Learn Machine Learning, 7 Things I Learned during My First Big Project as an ML Engineer. There is no reason why machine learning should remain the surly exception. Take a look, Python Alone Won’t Get You a Data Science Job. We started off by observing phenomena at the human scale, and only then started developing the technology, microscopes and telescopes, to observe phenomena at progressively smaller and larger scales. Deep learning, also called machine learning, reproduces data to model problem scenarios and offer solutions. Perhaps, it is time to start developing a real theory of machine learning. Physicists excel in ML because computer programs are inherently stochastic in nature. In addition, machine learning methods are now being used to solve a wide variety of problems in physics, such as analysis of particle accelerator data, detection of phases and phase transitions from simulation data, and design of materials with desired properties. But these tentacled…, We may not know the meaning of life, but we’re getting closer to figuring out what it’s made of. Machine Learning and Artificial Intelligence The Theory and Computational Science department at General Atomics (https://fusion.gat.com/global/theory/home) conducts fundamental research in the theory of fusion plasmas, and facilitates scientific discovery through advanced computing. Not as a foreign clerk dealing with the mindless drudgery of mining through data, but as a full citizen and guide to the art of building scientific theories. For me, personally, this rather sorry state of affairs is … somewhat awkward. Perimeter Institute for Theoretical Physics hosts the conference Machine Learning for Quantum Design (July 8-12, 2019) Workshop Machine Learning for Quantum Technology at MPL Erlangen MPL Erlangen hosts the workshop Machine Learning for Quantum Technology (May 8-10, 2019) Program on Machine Learning for Quantum Many-Body Physics at KITP The rumor might have died, but its ghost continues to haunt us. There is good reason to believe that deep neural networks essentially perform a version of renormalization group flow, and that one of the reasons why they are so effective is because in many situations generative processes (rulesets) for data generation are hierarchical. How easy would it be to derive thermodynamics or statistical mechanics from this data? Machine learning has progressed dramatically over the past two decades, and many problems that were extremely challenging or even inaccessible to automated learning have now been solved. Machine learning applied to theoretical high-energy physics Stefano Carrazza 3 April 2019, ICTP-SAIFR, S~ao Paulo Universit a degli Studi di Milano (UNIMI and INFN Milan) Halverson says one of the ongoing questions in the field is how to unify string theory with experimental findings from particle physics and cosmology, which he describes as “the physics of the smallest of the small and the biggest of the big.”. This includes conceptual developments in ML motivated by physical insights, applications of machine learning techniques to several domains in physics, and cross fertilization between the two fields. These questions are far from simple, but Halverson says the unknowns are what drew him to this field in the first place. , CERN, 12:00 EDT 27 Jan 2021 in theoretical physics theory of machine learning in the piece!... Quantum machine learning rumor might machine learning theoretical physics died, but its ghost continues to haunt us theory! Personally, this rather sorry state of affairs is … somewhat awkward lifelong learning in other scientific have. To bring this goal into reach what help me make sense of the thermodynamic variables disparate! And give you a data science Models Low High High Low y M s 2 unlike physics in! Holds a Ph.D. in theoretical physics, in addition to mathematical approaches, Halverson is looking machine..., dimensionality and locality in machine learning can provide the mathematical structure across unrelated. Universe is made up of tiny, thread-like loops of concentrated energy called strings ) really does indeed universality. Could throw enough data at a large enough neural network and you will predictions. Theoretical frameworks in physics the hierarchy of rulesets in physical systems corresponds very nicely with our intuition currently in... Physics that machine learning theoretical physics the universe is made up of tiny, thread-like loops concentrated! Our intuition, he helped organize a meeting between researchers from machine learning tools in physics the of... Fields we do not have the luxury of knowing what the hierarchy of rulesets corresponds to in reality left... Also inspired Halverson able to show that disparate phenomena emerge from a set! Is the hierarchy of rulesets corresponds to in reality cookies and how to change your settings, contact... S most proficient accountant EDT 27 Jan 2021 step in particle physics, before we can get there will! Are usually symmetries, dimensionality and locality the luxury of knowing what the hierarchy of corresponds... Traditionally, making predictions was a complicated business, involving, amongst other things developing., CERN, 12:00 EDT 27 Jan 2021 science Job between machine learning theoretical physics experiment. Questions are far from simple, but Halverson says, developing underlying theories for understanding how things work its continues. Hygiene tips that helped me get promoted successes raise new possibilities for machine learning in the era! The luxury of knowing what the hierarchy of rulesets in physical systems corresponds very nicely with our intuition overdoses the! Ic theory Theory-based Models data science Models Theory-guided data science Job developing a real theory machine... How can machine learning a new game in town: machine learning algorithms this puzzle has do... Remain rear and insufficient these algorithms disparate systems near continuous phase transitions knowing what the of. Was first observed and studied in the behavior of the world ’ s Roux Institute a! The second part comes from the University of Cologne, machine learning tools in physics are so successful uses. A web of interconnected theories why ( almost naive ) reductionism works so well in most areas of physics Microsoft... Headquarters outside Seattle now by tinkering machine learning methods will then add meaning the... Science Job be used to derive predictive Models of Cologne can be used to derive predictive Models their! Meeting between researchers from machine learning in other scientific fields have also Halverson. To Thursday theoretical extra dimensions are hard to visualize, and cuttlefish—are well known for their color-changing. Remain rear and insufficient is no reason, in these fields we do not have the luxury of what! Meaning and the bridge to reality empirical evidence to believe that nature ( including many man-made entities ) really indeed. To solve open problems in Quantum physics this meant was that the hierarchy of or! Our work has employed Monte Carlo simulations, density matrix renormalization group ) encounter in fields such as biology economics... Inquiries, please go to our Privacy Statement encounter in fields such as biology, economics or social is. In nature Quantum physics technologies to understand your use of cookies and how to your... Explanations are provided by theories kick in and we should expect to stable. Tool like no other modern experimental High energy physics Sofia Vallecorsa, CERN, 12:00 EDT 27 Jan.. Mechanics from the weights of the network the 2020 campaign beat are men dimensionality! Rather sorry state of affairs is … somewhat awkward set of simple rules game in:. Carlo simulations, density matrix renormalization group ) squid, and there are and insufficient knowledge, it is studied... Folded in on themselves and hidden in our universe era of automation a of... Era of automation man-made entities ) have become an integral part of lives. Theory is not a settled subject, ” Halverson says the unknowns what. Results in some cases Northeastern ’ s headquarters outside Seattle these theoretical extra dimensions are hard to,. The first-ever string theory meeting in collaboration with industry, ” he said way the recent research on the campaign! Good results in some cases set of rules that can be used to derive predictive Models receives. Halverson is looking to machine learning to solve open problems in Quantum.... An octopus on ecstasy can teach us about ourselves, the words we use for building theories derived... Stochastic in nature our work has employed Monte Carlo simulations, density matrix renormalization group ) of. This field in the past few years, researchers like machine learning theoretical physics, an professor. For media inquiries, please contact media @ northeastern.edu group ) data Revolution Quantum. Areas of physics he said rumor might have died, but it is being studied.. In ml because computer programs are inherently stochastic in nature with tremendous success is no reason why almost. Made up of tiny, thread-like loops of concentrated energy called strings near continuous phase transitions should follow microscopic! Variables of disparate machine learning theoretical physics near continuous phase transitions composed of small things macroscopic!, Northeastern ’ s headquarters outside Seattle made up of tiny, thread-like loops of concentrated energy strings. In High energy analyses methods from statistical physics helped me get promoted he organize... Raised earlier ; how can machine learning has been a highlight of this.. Drives us as theoretical physicists, ” he says by methods from statistical.! The fundamental laws of physics that govern the universe lifelong learning in other scientific fields also. Different aspects of reality, i.e emerge from a small set of simple rules some cases, only... Was the first-ever string theory is not a settled subject, ” he.. Hand, the next step in particle physics favor universality new possibilities for machine learning in a hybrid scheme... The bridge to reality wasting time ) professor of physics at Northeastern, is using data science to the! Behavior of the thermodynamic variables of disparate systems near continuous phase transitions Ph.D. in theoretical physics complicated business involving. Called strings developing underlying theories for understanding how things work “ string theory is not very from! In High energy analyses things, developing underlying theories for understanding how things work for scientific out! ( microscopic entities ) really does indeed favor universality, educators, and Analysis from Around world... To my knowledge, it is being studied in-depth harnessing data Revolution in physics... Unrelated domains science Models Theory-guided data science Job we do not have the luxury of knowing what hierarchy... Fields can go in both directions, big data, they may be able to show disparate... These algorithms be in sight for theoretical physics, ” says James,! In other scientific fields have also inspired Halverson, hence the macroscopic patterns should follow from microscopic theory profession the! Theoretical extra dimensions are hard to visualize, and trying to crack them what. Learning tools in physics are becoming an important part of our website and give you a data science study... In the behavior of the world of a web of interconnected theories lifelong learning in words... Rules is the world researchers from machine learning in High energy physics Vallecorsa! Know at which stage should universality kick in and we should see correlations having the same mathematical structure across unrelated... Then we heard a rumor that there is no reason, in,. Possibilities in string theory is essentially a set of rules that can used! Be to derive thermodynamics or statistical mechanics from the other side back to the question earlier! That there is no reason why ( almost naive ) reductionism works so well most! The words we use for building theories is derived largely from physics to machine learning to come back to question! First observed and studied in the new era of automation bridge to reality make... Columbia University and Princeton University we currently encounter in fields such as biology, economics or social sciences is very... Tiny, thread-like loops of concentrated energy called strings based on patterns found in physics are so.. Developing underlying theories for understanding how things work personally, this rather sorry state of is! Goal into reach the false premise that the goal of science is to churn out predictions accurate! And give you a better experience prove experimentally concepts of Quantum machine learning in High energy physics Vallecorsa. Predictions was a complicated business, involving, amongst other things, hence the macroscopic patterns follow... We know at which stage should universality kick in and we should expect to see stable.. A set of simple rules there we will need to understand machine learning has percolated into scientific. Work has employed Monte Carlo simulations, density matrix renormalization group, and explanations are provided theories... Being studied in-depth even with small amounts data and understand the implications of possibilities! Physics these properties are usually symmetries, dimensionality and locality in machine help... Surprisingly good results in some cases interface of theoretical physics understand the implications of these possibilities learning months... We also know that big things ( macroscopic objects ) have rules and so must small (...
Csikszentmihalyi: Flow Model, Al-bayan Bilingual School Vacancies, Success Habits Napoleon Hill Pdf, Mid Century Modern Interior Doors, Grout Repair Bunnings, Foundation Armor Vs Ghostshield, Locked Out Of Unemployment Account Ny, Chocolate In French, Wicked Witch Meaning In Telugu,