 Hello everyone and welcome to our monthly event. Great to know that there are so many of you joining us today to talk about quantum. I know that there are people joining from all over the world. Again, from the US, I know people are watching from Moscow and some people are even joining from Andalusia in Spain. So welcome. I am Katya Moskvich, the editorial lead here at IVM Research and I am here today at IVM Research in Zurich. Excellent. Our topic today could be more exciting actually. It's all about using quantum computers, as you saw in the title of the webinar, using quantum computers for unlocking secrets of the universe with particle accelerators such as the one at CERN near Geneva. And this is incredibly exciting. It's all about high energy physics, particles accelerating and colliding, producing other particles in the process and some hopefully new ones that can shed some light on some fundamental questions that we don't yet have an answer to. And this is exactly what the LHC, the Large Haddon Collider, is busy doing at CERN. I'm sure you all remember the LHC of course, especially after discovered the Higgs boson a few years ago, so that was pretty amazing. And very recently CERN, the European, which actually stands for European Organization of Nuclear Research, joined the IBM Quantum Network as a hub. So what do I mean by that? It means that scientists now working at CERN, which is of course, you know, it's a lot of scientists from a lot of universities all over the world now have access to IBM's fleet of quantum computers, which is super, super cool. They can access them through the cloud. And at the same time, CERN can provide quantum computing services to, you know, partners all over the world as well, which is really amazing. So our experts here today will, I'm sure, you know, give you a lot more details, tell you about whatever it is they're going to be doing, much better than I am telling you right now. But before I actually introduce you to them, let me remind you all joining us here today to please send us your questions in the chat, okay, and we'll try to get to them on air or through the chat because we've got actually a stellar team of scientists here behind the scenes. So you can see them here. They're all here today to answer your questions in real time and we'll try to get to as many questions as we possibly can. So please do, please do ask away. So now the experts joining me here today are Ivano Tarvenelli, Global Leader for Advanced Quantum Simulations at IBM Research Europe and Alberto DiMeglio, Coordinator of the CERN Quantum Technology Initiative and Head of CERN Open Lab. Now, we're not showing them to you just yet, why there's a reason, because we'd like to show you a really cool video of CERN. There we go. Welcome to CERN, the European Laboratory for Particle Physics. In the large Hadron Collider, we smash particles together to investigate the universe, how it works, what it is made of. This type of research results in huge IT and computing challenges. Quantum computing might not provide all the answers, but it may help us see the universe more clearly and ask better questions. All right, so joining me here today is the guy you just saw in the video, Alberto. Hi Katja. Hi everybody. Could you please give us maybe some basics to begin with? What exactly is the research at CERN and what are you guys actually doing? What are the challenges that CERN will face with, you know, new generation of experiments and particle accelerators? Just give us a brief overview. Yes, gladly. So as you know, CERN is a European but actually international research facility. It was funded quite a few years ago in 1954 as a place where collaborative science, fundamental science could be done by researchers from all over the world. The specific goal of CERN is to actually build and operate the particle accelerators. And today, the big accelerator, the main one, is the Large Hadron Collider. But this is a facility that is offered to experiments from many, many countries in the world to come and to research together. So until now, this has been extremely successful. We have made fundamental discoveries. You mentioned the exposure. This is one of the most famous, but there have been almost 60 different discoveries of new particles and new configurations of matter in the past several years. And today we use a collaboration across the world of computing and data resources that is reaching the exabyte level. It's reaching really astonishing levels. And the new experiments, the new generation of the accelerators are going to go beyond that. So we estimate that in the coming years, we might need 10 to 100 times the resources we are using today. And that doesn't work. We cannot scale. So we need to rely a lot on technologies to introduce new ways of working. Quantum technology is one of the things we are looking at. So I guess behind you, right? So this is the data center that you're referring to. So at the moment, this is what is actually dealing with all that data coming out of the collider, right? Yeah. I mean, one of the steps is not yet quantum. This is the production facilities, the data center we have here, which is, we call it the tier zero, is where the data comes into from the experiments before being distributed across the world. And I'm sure you can appreciate my t-shirt today, right? Absolutely. So I hope our viewers can also recognize that this is quite an amazing image. I actually, when I bought the t-shirt, I thought maybe I should also get an actual picture and like hang it in my living room, because that would be amazing. Maybe I'm sure some physicists that certainly probably have that actually, I wouldn't be surprised. So speaking of quantum, before we go to Ivano here, actually, I have another question for you quickly, Alberto. Why is quantum so exciting? I mean, yes, you mentioned using it for the data, but is there anything else that it can do for CERN in the future? Yeah. I mean, there are many, many ways, many places where the effects given by using quantum computers may play a fundamental role. So a lot of the problems we have to solve, a lot of the different classes of computation and data processing, actually are exponential or have a combinatorial scaling effect. This is in many, you know, many different areas. For example, data track reconstructions or quantum field simulations and radiation simulator simulations, lattice, gauge theories. All these problems would not scale well with increasing precision and accuracy by doing this in the coming years only on classic resources. And one interesting class of problems that we may talk a little bit more about it is data classification. How do you extract rare events, the difficult ones from the background of everything that happens when protons collides in the accelerator? And there are indications that by using some quantum techniques, actually in the future we might go beyond what is classically possible today. This would be very, very interesting, amazing for the community. But essentially it's the start of the journey. We are exploring the possibilities and we have good insights. The vision, if you want, this is maybe quite a few years in the future, but imagine being able to actually take the quantum state of data coming out of the detectors and feed this data, quantum data, into a quantum computer directly and get out of the answer of the universe. I'm literally looking forward for that to happen. Yeah, answers of the universe. I mean, this just sounds mind-blowing to me anyway. Super cool. Okay, well, I'd love to go to our second expert now, to Ivana Tarvinelli, my colleague actually here at IBM Research. Hi, Ivano. Hello, Katja. Hello, everybody. Hi. Cool. So, question for you. Knowing that you're obviously a physicist and a quantum computing expert, could you tell us more about the role of quantum computing for high energy physics kind of more generally? I know Alberto just kind of went into some detail here in his answer, of course, but for you specifically, where do you see the relevance of quantum computing as related to high energy physics? Yes, of course. So, in addition to the applications that Alberto just mentioned, we can use quantum algorithms to detect events compatible with new physics. So that is everything that is beyond what can be understood and interpreted today. So, in fact, what we can do is to train quantum machine learning algorithms to detect anomalies in the data and analyze them to discover new phenomena that go beyond the standard model of particle physics. That would be really very cool. So, another application of interest could be jet reconstruction. So, let me explain this concept in more details. So, when two particles collide, for instance, to proton at the large atom collider, they break into their constituents, meaning the quarks and the gluons, and eventually they also generate new particles. However, we cannot unfortunately directly observe the products of these collisions. In fact, the detectors are placed meter away from the point of the collision. And what we observe in reality are reconstructed jets of particles that travel from the point of the collision to the surface of the detector. So, with the jet reconstruction process, we recompute the particle trajectories that reach the detector in order to understand the physics back at the point of the collision. However, the number of possible parts that define the jets is growing exponentially with the number of particles that reach the detector. And therefore, it's hard to solve this problem classically. And we hope that with the use of quantum computer, we can speed up and improve this process. Finally, another application that we are exploring is the solution of the quantum field theory equations. So, that is the standard model at the low energy regime. This is a very interesting regime because it's the regime where perturbation theory is not working. So, we cannot solve it mathematically. And classical computers are limited by the exponential scaling of the complexity of the problem. And therefore, it's hard to solve these equations with a classical computer. So, we said we are exploring scalable quantum algorithms for lattice gauge theory to model these kinds of theories. So, quantum electrodynamics, for instance. Quantum chromodynamics, so the interaction between gluons and quarks, which are the fundamental theories of the subatomic scale. Cool. Yeah, that sounds really, really amazing. And just to remind our viewers here behind you, of course, and well, you've got a proper actually physical model behind you of a quantum computer, right? So, something that we have here in the lab and something that people can actually access through the cloud. So, this is what it looks like. It doesn't look like a regular computer at all. It looks like a really cool, you know, dystopian chandelier or something. I've got the same thing here on a poster. So, you know, a little bit less exciting that what you have, Ivana, so I'm a bit jealous. But, you know, what can we do, right? Anyway, actually, excitingly for you, we've got our first questions coming in from the audience. And somebody here's, I don't have names, but somebody here is asking, will quantum computers be able to simulate the processes we examine in a collider in reasonable computational time? What do you think, Ivana? Yeah, one of the applications would be indeed to put a quantum computer in between, so the experiment and the collection of data. So, as Alberto was explaining before, we can use it to filter what comes from the detector and using a support vector machine only take the events that contain interesting physics and discard all the rest so that we can handle the amount of data that Alberto was discussing before. So, that would be definitely an application. It's hard because there would be many data involved, so it's definitely probably not for near term, but is one of the applications we are looking for. Yeah, no, that's super cool. And of course now, you know, as we said before, CERN is a hub in the IBM Quantum Network, which is quite amazing, but I'm sure that many viewers here are probably wondering what that actually means. So, could you maybe explain to us what these quantum hubs are in general and where else they are and why do we need them, so what they do, just a little bit of background on quantum hubs? Yes, of course. So, the mission of the hub is to partner with academia, so the university, but also research institutes and industry to explore promising use cases. So, this is really the key point. So, we are looking for use cases where we can apply our technology. And this is in different domains that goes from natural science, obviously including high-energy physics, but also other fields like optimization, finance, and machine learning in general. As a hub, CERN has its own dedicating access, and this is really probably the most important aspect of being a hub, has access to the IBM Quantum fleet of more than 20 quantum computers. In addition, the IBM support to the hubs includes onboard training and technical assistance in terms of research. This is why we have this collaboration also with CERN, education, also very important, you mentioned it several times, and the development of new applications. We have other IBM Quantum hubs that include universities in many different countries. In addition to the US, we also have hubs in, for instance, Japan, Australia, Canada, UK, Germany, Portugal, and most probably many other places. Okay, cool. So, in the future, I think the hubs will likely be sharing tools, experiences, and best practices to boost the rapidly emerging field of quantum computing. This is the goal. So, we have many hubs, and these hubs need to talk together and help developing the field. Let's conclude saying that in addition to the hub, IBM has other models of collaboration. So, for instance, we have a dense network of academic collaborations with many other universities across the globe, and also here is where we do most of our development. Yeah, for sure. I mean, collaborations are super important. If we were just working by ourselves, it doesn't matter if you're a university or if you're a company, if you're just doing your stuff by yourself, then you're not going to go very far. So, I think collaborations be they private, public collaborations even, like what's happening now, IBM, and lots of universities that are working with CERN. Science knows no borders, I don't know, in my view. If you're a scientist, you're clever, then it doesn't matter where you work if you're working on the same goal. So, anyhow, speaking of the hub, we've got a question here for Alberto from somebody in the audience. So, here we go. Alberto, how do you envision the progress of quantum computing at CERN? What will be the next steps after becoming a quantum hub? That's a good one. Yes, we have just started already looking at the next steps. So, I mean, it's an important question. So, as I mentioned in the beginning, you mentioned already a few times. I mean, the keyword here is collaboration. CERN is a place for collaboration. And I mean, this is true in general for fundamental research. Quantum technologies, quantum computing introducing introduces, sorry, even more the need for collaboration is a truly a multidisciplinary kind of field. You need computer scientists, you need experts of the domain, you need engineers. So, what we are doing at CERN, and this is what happened last year, we have started a dedicated quantum technology initiative, which I'm the coordinator of, exactly with the goal of understanding what is happening today at CERN, how to coordinate all the possible different activities in this field to measure the impact of quantum computing and quantum technologies for fundamental research. Not only computing, for example, also sensing and theory and networks and communications are in scope, but especially how to work better with the rest of the community. So, CERN is by far not the first place where quantum technology is looked at and not the biggest. There are many other activities you mentioned, universities, you already have a lot of universities and companies in your network. So, the next steps are to build this community within, you know, as part of the high-energy physics community, understand the requirements, understand what scientists want to do, understand what are the most interesting projects to run together, and we have already quite a few running at the moment, but also provide ways of actually improving and supporting the building new knowledge. These are definitely emerging fields, and we are targeting, not necessarily, by definition, the research being done today, but the research that we have to do in the coming three, five, 10, 20 years looking at high luminosity LHC, the next generation of LHC, and even farther, the future circle collider. We need to help young researchers, engineers, scientists, physicists, to be able to understand what quantum computing is, what it means, what knowledge, the skills you need to acquire, and that will apply those skills in the field, in the best way. So, these are the next steps. It's a lot of research, a lot of collaboration, and especially building this community together. Yeah, absolutely. And, you know, you mentioned the future generations of particle colliders. I remember I visited CERN a couple times, and the last time when I visited, I think it was two years ago, everybody was super excited talking about, like, how you guys can upgrade it, what you're going to do, and all the scientists are already kind of looking into the future, and how these machines can, indeed, just keep getting better. And quantum, it's just, I think, really cool technology that could, you know, help during this process. Super cool. We've got an interesting question, actually, for you, Alberto, again, here from an intern at CERN. Hmm, I wonder who that is. So, the question is, I am an intern at CERN working on anomaly detection at CMS. Will quantum machine learning replace regular deep learning in the near future for such tasks? Interesting. This is actually throwing us a little bit ahead of the, in our discussion, because we haven't really talked about the paper you guys are working on. We'll get to it more in detail later, but maybe you can just address the quantum machine learning a little bit now, and we'll get back to it more in detail. Yeah, I think the keyword here is in the near future. So, this is a, of course, is a very interesting question. We need to understand exactly what is the role of technology. So, what we are investigating, and we can talk with them a bit more of some specific cases. What we are looking for at the moment is exactly where it makes sense to use quantum computing devices and also simulations and getting there. There are areas that we can see today are promising places for actually introducing significant changes. This is not near term for many reasons. The technology is evolving, the problems have to be understood, but it is necessary to start doing this today. I actually heard, I think this morning from another person, an interesting analogy, saying that at the beginning of the century, the last century, when people were using trains, there was some brave engineer building almost paper airplanes. Nobody at the time thought this could become a viable means of transport, but there were people already looking at how to build a business. I liked this analogy very much. It took some time, but when all the different pieces got together, the technology, the infrastructure, the knowledge, aerospace, aeronautical engineers, and material, especially, et cetera, the airplanes became one today, the COVID permitting. This is a different situation, but in principle, it's the way of traveling long distance today. There is no better. We are in the same journey. It is not near term, but it is a place where we want to be. Yeah, absolutely. It's a great analogy. Actually, I love it. And this third intern, please come and see me. We can talk directly over a coffee. That's a really great suggestion. I'm sure the intern right now is thrilled. Thank you, Alberto. So, Ivano, actually, I've got a question for you. And if we can go back to this whole quantum hub announcement, which is super exciting, of course. So, if I understand correctly, if an organization is using IBM's quantum computers as a hub, it's slightly different to just using a quantum computer from time to time over the cloud. So, what's the actual difference between being a hub or not being a hub and just connecting to IBM quantum computers through the cloud? Yes, indeed, there is some differences being a hub or just a user of the IBM facilities. So, in addition to the privileged access to the IBM quantum computers that I mentioned before, so as a hub, CERN, in this case, can invite external partners to join its network. And this you can do by signing, for instance, new contracts with universities or other research centers. In this process, they don't have to involve IBM. This is a very important point. This means that CERN can handle relations with the partners in the way it is more convenient for them. Right, yeah, okay, that makes sense. Cool. And, Alberto, what's your view? What do you think will change with CERN now becoming a quantum hub? I mean, we kind of touched upon this a little bit, but if you could just, I guess, summarize it a bit more. Yeah, yeah, indeed. So, I mean, we have been working with IBM. I mean, CERN has been part of the IBM Q network for some time, I think, 2018, 2019, as an individual member. Now, I keep mentioning collaboration. This is an important aspect. So, being a hub means that we can partner with the institutes working in the community and with experts in physics and computer science and set up joint projects that together can benefit from accessing the 20-plus fleet of computers that Ivano mentioned, IBM expertise, and we work very well with Ivano and the Zurich Research Center. So, this possibility is part of, for us, it is what it means being a hub. It means that other people within the community as part of the experiment, as part of projects, can enjoy the same access and the same possibilities that we have today and produce better results. This is really multidisciplinary. So, this is the way we work. So, it fits very well in the CERN spirit of working in this way. Yeah, now, for sure. And what about kind of the expertise of researchers in specifically in high-energy physics? Do you think they all need to be experts in quantum already now, or what's the deal here? So, it depends. I mean, I think, so, certainly, I mean, it's not enough to be an expert in quantum mechanics, to know everything about quantum computing. Probably helps, but these are different profiles, different professions. So, yes, I mean, I touched upon this, we really need to understand how we introduce this knowledge in the community. And to actually, it's actually a matter of bridging different communities sometimes. It's not about energy physics only, but very often specialists work in their field and try to solve problems in their field. Now, with the possibility, quantum computing is a very interesting paradigm of how doing that would not work. It's really necessary to bridge these different communities and work together in joint projects. This is what we call, many people call, in a way, co-development. It's necessary to have different skills together to advance the state of the art across the board. And we need to provide the means of doing that. This training is an academic type of lectures and opportunities, also working with universities to maybe change some of their parts of their CV, of their academic curriculum to enable this kind of change. So, it's fascinating. It's the start of a journey. Very exciting to be there. For sure. I remember, I never had quantum physics at high school, obviously, and not at all. And I only studied quantum physics when I did my post-grad degree much, much later. But, when my son was born, I actually bought him a tiny book, Quantum Physics for Babies. So, that was a good start there. Absolutely. Start early. So, who knows? Maybe a few years from now, there will be quantum physics even on a high school level. That would be terrific. And, Ivano, what are your thoughts on this, actually? Yeah. I mean, I also believe that it's really very important that we form a strong community and we start training people from high school to learn how to program a quantum computer and how to solve the problems of our society with this new type of technology. So, this is why IBM has developed already, sometimes, a software platform called Quantum Lab that allows everybody, even at school, to test algorithms and also execute them on a quantum computer. So, you can do really real experiments nowadays on a quantum computer. For instance, you can test the fundamentals of quantum mechanics by proving bells inequalities, so using the Quantum Lab that is put on the cloud by IBM. So, in addition, we have Qiskit, that is the software platform of IBM, also open source available on GitHub, which has a very powerful and unique library of modules for the solution of many problems. Already mentioned before, modules for the solution of problems in natural science, for instance, maths, physics, biology, medicine, material science, but also problems in optimization, machine learning, and finance. So, that's also a very promising field where we can apply quantum computer. Obviously, in order to do that, indeed, as we were mentioning before, we need a dense training program, and this is why IBM is organizing many events. You know that we are organizing schools, tutorials, hackathons, where the young people can join and learn this new technology. And obviously, CERN has a hub, we certainly partner in these activities, in these educational activities, especially when it's about applications in theoretical physics and particle physics, but not only obviously. Yeah, yeah, no, that's really cool. I mean, you mentioned hackathons and, you know, they're also Qiskit camps. This is, I think, it's really cool. It's kind of introducing young people to quantum in a really fun way. So, you know, this is quite amazing, and there are quantum gaming as well, so there's lots, lots going on. Okay, back to our chat here. So, somebody's actually asking you, Ivano, what is the prospect of quantum computing in quantum chemistry? How one can start thinking about a use case right now? Okay, so thanks for the questions. Quantum chemistry indeed was one of the first applications of quantum computing. And we searched already a long time ago, even Richard Feynman was this idea to use quantum technology for this purpose. Yes, I mean, we have already a lot of work done in this direction. We have very promising algorithms that can solve electronic structural problems. I would say that the problem that we are facing at the moment is the fact that the accuracy that is needed in quantum chemistry calculations, the so-called chemical accuracy, is a very tiny number that is hard to get with a quantum computer. But we are definitely working also in this direction. And hopefully, when we will have slightly larger quantum computers, we will be able also to show advantages in this domain. Yeah, yeah, I have no doubt about that. And yeah, I'm just wondering actually, you know, what I've been thinking, well, you know, as I mentioned in the introduction, CERN, of course, discovered the Higgs boson, right? So, what if we had a quantum computer when we were searching for the Higgs? Do you think we would have found it earlier? What do you guys think? What do you all want to think? Yes, maybe. So, let's put it this way, I want to be non-committal. Let's put it this way, seriously. Okay, what you see here is something that we show in the Visitor Center. We monitor technology and we have made the sort of simulation. How long would it take today to process LHC data if we had the technologies still from the 60 or the 70? And, you know, for example, with mainframes available in the 60s to process the data out of the accelerator today, it would take almost the entire age of the universe. And this is technology that was available 50 years ago, not so long ago. And, you know, even with technology from 1985, the dinosaurs should have started the computation to have results today. So, we see these leaps. There are like a step increase in the way technology helps this kind of research. So, this is the effect that we hope for, we expect from changing paradigms. We went from mainframes to commodity computing, from single cores to many cores, 32 bits to 64 bits, from von Neumann to non- von Neumann applications, quantum computing is the next frontier. So, we expect the same type of effect. It is actually the only way to make the kind of research we do actually progress towards the next challenges or dark matter and everything that we still need to discover. So, my answer is yes. There is, you know, you can see a lot of hope behind my yes. But, I mean, I think this is what happens. Well, you know what, you mentioned dark matter. Oh, my God. You know, this is exactly what I did my M field thesis on dark matter and FRVs. And, yeah, I wish I had a quantum computer when I was studying at King's College, but hey, maybe I should do another degree in a few years time. Anyway, a question for you, Albert, now that I have you here next to me on the screen, somebody is asking, I want to know about research opportunities in high energy physics with quantum computing, speaking of education and stuff. My master's thesis is in particle physics, and I have been active in quantum computing for more than a year now. So, yeah. So, okay. I mean, it depends also where. So, let me start locally. So CERN has this new CERN quantum technology initiative. There is a website, which is simply quantum dot CERN. You can go there and we publish it's, you know, again, it's, we are starting that. So, the website is being populated now with with information and opportunities. But we will put there everything we are doing and all the opportunities we have. We have started already a number of projects. So, there are more projects in the pipeline that we will start. So, going there is a good starting point. But then, you know, this is a, you know, quantum computing and quantum technology is a field that, you know, many universities and many initiatives are working on. You know, Europe, if you want, has, you know, a flagship project of the quantum flagship initiative and US have similar initiatives and programs. The European Commission is launching a number of very interesting projects in the coming years to explore everything in the stack hardware, algorithm, software and to end the stack of quantum computing. So, there will be many opportunities. So, as far as we are concerned, go to the website and we will keep you informed. I guess that universities and institutes will do, we do the same. So, lots of opportunities. Yeah, sounds super cool. Well, and now I'd love to talk actually to both of you about some recent, rather mind blowing in my, for me anyway, research that you too I think I've been doing together on quantum machine learning. We briefly mentioned it earlier and because somebody asked a question about that. Now, if we actually take a step back, right, because quantum machine learning, that sounds quite cool, but also probably a little bit unfamiliar to many people. So, maybe we should start just with, you know, really briefly regular machine learning. So, Ivana, maybe you could give us first, like a brief overview of what machine learning is and how quantum machine learning is different and how it can actually help. Okay, okay. This is really a very interesting question. But let me start maybe from the origin, right? So, we need to talk about the scientific method. This was developed roughly in the 1600s with people like Galileo and Newton and many others. And essentially, it is based on the mathematical modelization of nature. And this was really a success. So, if we are here talking about science is because of the great work of these people. Of course. But with the advent of the computers in the last century and more recently with the big data science, we have now the possibility to train a computer and therefore a machine, right, to recognize patterns in the data and automatically infer a solution for new problems. And all this without a precise deterministic model of what we are investigating. And this is, I really find amazing. Now, coming to quantum machine learning algorithms, in most cases, these are based on concepts that are derived from classical machine learning. Not always, but in many cases. So, we have, for instance, quantum neural networks. We have quantum Boltzmann machines. We have quantum classifiers, as we used in our demonstration. So, coming to the differences, the main difference compared to classical machine learning is that we can make use of the exponentially large hyperspace of the qubits, which can enable, for instance, a more efficient and accurate classification of the data. And in the most specific case for high energy physics, we believe that a quantum computer can also better capture the quantum correlation effects and entanglement that is in the data that may be hard to get classical. And this is because we are modeling quantum with quantum, exactly like Richard Feynman was proposing about 40 years ago. So, namely, modeling quantum mechanics with quantum technologies. Obviously, all this is still under investigation and we are really far from the end of our journey. But to succeed in this process, I believe it is very important to work in these collaborations. And in the case of high energy physics, with an institute such as CERN. Yeah, yeah, that's super cool. And you mentioned Richard Feynman. I mean, I actually have a few books written by Richard Feynman and people listening right now, especially students, if you want to get a super cool overview of physics, read his books or listen to his lectures. He was amazing. He's really probably the best physics communicator that I know of anyway. So, please do. Anyway, back to your paper. I'm sure that some of the people here are not, you know, physicists necessarily. So, yeah, just if anything is not clear, then just do send your questions in the chat, right? And, yeah, back to the paper. Alberto, maybe if you could now give us your kind of view on what the paper is actually about, quantum machine learning a little bit more specifically. Yeah, I can try to explain what the paper is about. I mean, Ivan is one of the main authors of the paper. So, I feel like I'm going back to university in front of my professor and trying to show I understood. Well, you've got lots of people watching you. So, come on, pressure. So, no, very, I mean, let me try to explain very simply. So, this is a classification problem. So, imagine you have your nice accelerator and detector and you smash the protons together and you produce this huge amount of data. Then you want to look for something very specific and maybe for something that doesn't happen so often, very rare. So, you need to really to extract this needle from the haystack. So, one way of doing this, there are quite a number of techniques, but machine learning technique called support vector machine is one of the ways of doing it. Essentially, it's a way of finding some way of separating something from something else in a signal from the background, the thing you want to find from everything else. Now, for simple problems, in general, it's not difficult to find a solution. You have a line, a plane of some type that splits your data into good guys and bad guys. Now, assuming your data is not that simple, you have a lot of features, a lot of properties, a lot of dimensions and maybe the things you want to find are very small. The number of things you want to find is very small compared to everything else. So, finding how to separate the signal from the background is not so easy anymore. Finding this plane is not so easy. Now, plane is a general word. It's not a flat plane. It's a plane that could be in a hyperspace, a plane in multiple dimensions. So, the support vector machine can still be used, but you need to find a way of sort of transforming from the complexity of this hyperdimensional space into something linear that you can work on. So, there are ways of doing this. Normally, in support vector machines, you use functions, mapping functions, the kernels, and the technique works. Now, as the problem becomes more complex and the number of possibilities increases, it is difficult to find the good functions to do this, able to represent correctly the information hidden inside the space. Now, it has been demonstrated already by other people, actually, I think IBM researchers, that it is theoretically true that by using quantum kernels, it is possible to achieve what is called typically advantage, to do something that from a computational efficiency point of view is not possible to do with classic machines. So, the paper takes that approach, that theoretical approach, and actually demonstrates this with a realistic case. So, it's a case of the event that we were looking for. This is a reproduction of something that has already been found with classic ways, but we were trying to see whether it is possible to speed this up. It's a particular X boson production and DK channel mechanism. So, the idea, by using these quantum kernels, it is possible to show in the paper that not only on simplified models, the quantum approach has today the same level of accuracy, but in some cases converges faster. And what is important, what is potentially going to give the advantage, is that by increasing the dimensionality, increasing the complexity of the problem, increasing the amount of data, while the classic techniques will sooner or later fail, this quantum approach will not, because of the way quantum computers work. So, today, this has been demonstrated within the limit of current technology. And by using computers that will certainly, at this point, come in the future, sooner or later we will pass this threshold and do things that today or even the future with classic machines, classic computers, will not be possible to do. So, it's incredible. So, if we can find this kind of advantage, in this kind of classification problems, in anomaly detection, in simulation, then it will change a lot in the way we work. Yeah, for sure, for sure. Yeah, and you mentioned advantage quite a few times. And actually, well, let me ask Ivano, I'm sure you could answer that just as well, Alberto, but I just want to get Ivano's view on this work you guys are doing. And Ivano, at the same time, if you could also explain to our audience here what quantum advantage actually is in simple terms. Well, okay, there are several definitions of quantum advantage. I think it's very hard to give an exhaustive answer to this question. So, let me say that in high energy physics, we know that there are problems for which the classical solution requires an exponentially large number of resources. So, what are these resources? These can be CPU hours or simply memory. So, so much memory that we cannot afford on a classical computer. With a quantum computer, instead, we aim at solving these problems more efficiently. So, this means with a polynomial scaling in the number of qubits, instead of exponential. And this would give definitely a great advantage. Every time you add a single qubit to a quantum processor, you double the power of your processor. So, this is the great, say, improvement in quantum technology compared to classical. Then, for instance, in the case of data classification, the use of the exponentially large qubit space can enhance the classification process that Alberto just mentioned a second ago, but also in jet reconstruction and lattice-gauge theory calculations. We think that one day the quantum computer will enable the solution of problems that are impossible nowadays, even if you use the largest high-performance computer, classical high-performance computer on Earth. Wow, yeah. That is incredible. That's true. So, actually, a question for you, I want to hear from the chat speaking of something a little bit different. Somebody is asking, Ivano, you mentioned about using quantum computing to process the data generated when two particles collide in collider. How long does this process actually take? Doesn't that, and what's the nature of data? Sorry, I'm not sure I'm understanding the question. Yes, basically, the person is asking whether there is any data collapse or data alternation in this process. I hope you understand that better than I do, Ivano. Yeah, I mean, that's a very complicated question. So, let me interpret it. So, the data produced in this experiment are produced really at a very high pace. I think Alberto can answer better than me. And obviously, they need to be filtered in one way in order to be possible to process them, analyze them. I think that the problem of the speed is not an issue here because, as I said, the data comes so fast that the decoherence time of the quantum computer that is in the microsecond or fully soon in the millisecond domain will not limit the analysis of these processes with a quantum computer. But I'm not sure that I'm interpreting correctly the question. Maybe if I can compliment that. I think we need to clarify also the question because it looks to me that maybe this is related to what I said at the beginning, that is my dream or the possibility of getting quantum data out of a detector and feeding that data directly into a computer, which today is not possible. So, the two things today would not happen at all in the same place and at the same time. So, we're not talking about taking the data generated by the collision of the particle and taking this data directly into a quantum computer. So, this data is classical data, is digital data that is then stored in classic digital storage. And then the processes, the classification, the tracking, whatever analysis you want to do is done on this classic data using a quantum computer or a combination between classic and quantum. So, for example, the support vector machine problem of classification we were describing is actually a combination of the two, the quantum kernels are the quantum part and the actual classification optimization happens on a still on classic computer. It's a good way of using quantum computer to accelerate some operations. So, the problem of the data collapsing or the entanglement collapsing is not related to the quantum states of the particles coming out of the detector. So, that has to be clear. So, if that is clear, then I think the answer is easy. So, there is no relation at the moment between the two things. Okay, cool. Well, going back to what you guys were talking about earlier on the specific quantum machine learning, right? I suppose that that is very different from collision analysis, right? And so, quantum, with quantum computers now at your kind of disposition, do you think you can make even more progress with specifically with that type of research? So, yeah, I mean, yes, for sure. So, we have mentioned the value of myself that classification is one of the problems. Any process, any computation where there is this exponential scaling effect potentially is a good match for quantum computers. And this happens in many places. Ivan has talked about reconstruction, jet reconstruction. That is an amazing problem. And when you have this data coming out of the detector, these are essentially coordinates and energy measurements, et cetera. And this data is really digital data. And then you have these dots, you know, imagine, you know, take a detector, it's like, you know, a cylinder and you slice the cylinder and you have a circle in the center, happen the collision, and then you have these showers, the jets of particles crossing everywhere. But what you see is these digital dots. And then you have to reconstruct the tracks. So, normally, I mean, certificate is not very accurate, but I use the analogy of, if you know, these puzzles for kids when you have dots and the numbers, and you follow the numbers and you build a picture attack. So, imagine doing that with billions of these dots and no numbers, okay? So, where do you go? So, it's an extremely difficult problem. It's a combinatorial problem. Every step opens up more possibilities and more possibilities and new showers. So, this is a problem that it's really suitable for quantum computers. If, you know, we need to find the right algorithms, there are possibilities, some application of quantum neural network, quantum graphs, quantum graph networks are a possible approach, tensor network is a possible approach. And there are other places in simulation, you know, using today, simulation is done with Monte Carlo models. Monte Carlo models are very efficient and very precise, but they don't scale well, but the resources don't scale well with precision. If you want to increase the precision and the complexity of the models, you need more and more and more resources. At a certain point, you reach a limit. So, we are experimenting with deep neural networks with the generated adversarial models to reduce the time and the resources. And the step beyond that is quantum generative adversarial networks. Can we reproduce the same probability distributions that we created with Monte Carlo much more efficiently than what we can do today, et cetera? So, and theory problems, you know, Ivano mentioned the lattice gauge problems, you know, quark glue and plasma and quantum Boltzmann theories. So, this is really fascinating and it opens up possibilities that we don't have today. Yeah, for sure. Well, I'm just a little bit mindful of the time here. There are still a lot of questions, but I think we're only going to be able to answer one more now and I'm going to kind of combine it with the question I had for you, Ivano. So, somebody is asking here, how do we monitor, contribute and learn from the CERN quantum initiatives and collaboration? So, this is the question from the chat and just to conclude if you could kind of tell us about the future of this collaboration that you guys are now, you know, entering and what what can we see in the future coming out of CERN and IBM working together? Yeah, of course. So, our goal is to further explore the potential of quantum computing in high energy physics. This is the goal of our collaboration with CERN. This obviously requires different type of knowledge and from, for instance, the expertise on particle physics will definitely come from CERN, but also we will need the development of new resources and new algorithms. So, we have to improve our software offering more modules for high energy physics so that we can leverage in the best way possible the computational advantage offered by quantum computers and this when I say quantum computers, I mean near-term and also long-term quantum computers and long-term quantum computers means fault-tolerant quantum computer that will come one day. All this will hopefully allow for the exploration of new frontiers in high energy physics. We were discussing dark matter, dark energy, all these fascinating fields that are still without an answer and therefore contributing to a better understanding of the universe at all scales. And if in this journey we can also have some interesting demonstration of quantum advantage for real applications that is clearly welcome, so this is also our main goal at IBM. Yeah, and I mean on my side, I mean of course I agree with what Ivan is saying, but specifically for CERN, again I insist a lot on collaboration and this is the real goal. You mentioned there was a question on how do we monitor, so it depends what you mean. So for people to look at again I mentioned the website but we are going to have events and workshops and we are going to make all publications available so it is possible to monitor the research. Then there is another aspect of monitoring, the other way is how do we monitor the impact that we monitor that indeed you know actually what we do is making an impact even potentially. So there are various ways. One thing that we are planning of building for example apart from a specific project is a sort of community-oriented platform where actually we can really monitor the progression and of development of algorithms for high-energy physics and their implementation on the newer generations of platforms. So we can really understand where we are going and what we can do. So assume we start today with NISC computers, we understand how the algorithms behave with the today error mitigation techniques, but then when fault tolerant machines come and we have full error corrections and we have millions of cubes, then we will be able to expand the dimensions of the problems we have and by working by creating this pool of knowledge, database of knowledge with the community for the community we will be able to understand whether we are going in the right direction and what are the areas where it makes sense to invest intellectually and also financially. Yeah absolutely well great you know thank you so much to both of you it's been amazing and also thank you for everybody who joined us today. We've introduced quite a lot of technical concepts. If you guys want to know more first of all you have to subscribe to our channel so that you don't miss any future shows, but you can also go back because in the past we did talk about quantum computing a little bit more kind of on the basic level. If you want to look for example in our February YouTube webinar with Haiky Ryle you will see we talk we explain all these terms a little bit more in detail and then you can go back to this one and listen to Alberto and Ivana all over again because it's going to stay online of course and if you have any comments any questions any ideas do reach out to me and to our experts. I am on Twitter as SciTechCat please do message me whenever you want. Thank you again and goodbye. Goodbye. Thank you. Bye-bye.