@HPC Podcast Archives - OrionX.net - @HPCpodcast-79: Travis Humble of ORNL on Quantum Tech

Episode Date: January 24, 2024

We discuss the state of Quantum Information Science with our special guest Dr. Travis Humble, a global authority on the subject, director of the Quantum Science Center, a Distinguished Scientist at Oak Ridge National Laboratory, and director of the lab’s Quantum Computing Institute. [audio mp3="https://orionx.net/wp-content/uploads/2024/01/079@HPCpodcast_Travis-Humble_ORNL_Quantum-Tech_20240123.mp3"][/audio] The post @HPCpodcast-79: Travis Humble of ORNL on Quantum Tech appeared first on OrionX.net.

I would say if anything in the past five years that pace of development the rate at which we are moving forward has actually increased so from my perspective I would say I'm very optimistic at the moment about the state of the field as well as its development and near-term ability to deliver on some of these key outcomes that people have talked about. Well, unfortunately, some of the materials we are trying to build, we will never be able to simulate, even on today's supercomputers. They are simply too complex in the way that the atoms and electrons are interacting with each other

that our conventional technologies cannot track those systems. But you might expect that a computer that is fundamentally built on the principles of quantum mechanics can actually mirror those systems almost perfectly. My expectation is that the future of quantum is to have a diversity of technologies. We will find that each of these modalities actually fits certain use cases. From OrionX in association with Inside HPC, this is the At HPC podcast. Join Shaheen Khan and Doug Black as they discuss supercomputing technologies and the applications, markets, and policies that shape them. Thank you for being with us. Hi, everyone. Welcome to the At HPC podcast. I'm Doug Black with Shaheen Khan. And with us today

is our special guest, Travis Humble. He is director of the Quantum Science Center, which is a Department of Energy funded partnership comprised of leading academic institutions, national labs, and corporations. He's a distinguished scientist at Oak Ridge National Lab and director of the Oak Ridge Quantum Computing Institute. He holds a joint faculty appointment with the University of Tennessee Bredesen Center for Interdisciplinary Research and Graduate Education, working with students on energy efficient computing solutions. So Travis, welcome. Oh, thank you, Doug. It's a pleasure to be here. Okay, so tell us a little

bit about your areas of focus in quantum and some of the R&D work you've been involved with of late. I've been working in the area of quantum science and technology for, well, about two decades now. A lot of that started with my initial research in quantum chemistry and how to control chemical reactions. But when I came to Oak Ridge National Laboratory almost 20 years ago, I immediately got immersed in the area of quantum computing. And of course, this is a really hot topic at the moment, the idea that we can use quantum mechanics to perform computation. My research in this area has kind of covered the gamut of topics in computing, including both the development of algorithms and applications for quantum computing systems,

software and programming tools that can be used to develop those applications. And then most recently, I've been looking at the development and testing and evaluation of quantum computer systems that are available today. Alongside of my research program, I am also leading the Quantum Science Center's research agenda, and this is broadly focused around the idea of building new types of quantum technologies, especially the development of quantum materials for topological qubits, the development of new types of quantum simulation platforms, and this is using quantum computers to specifically investigate chemical

and material processes, and then the development of quantum sensor platforms, which can be applied to really exotic phenomena, including the search for dark matter, as well as the development and demonstration of quasi-particles and quantum materials. So this position in particular has given me a really broad perspective on all the different types of topics that are under consideration today in the field of quantum science and technology alongside my past research in that area. Okay. Well, stepping back from a 30,000-foot level, I always ask quantum people this. What are your thoughts on the state of quantum and whether the technology is moving forwardly in an encouraging way, or is it somewhat stalled with all the developments and announcements going on? How do you view

the forward movement of quantum? Having been involved in the field for almost 20 years now, I can definitely say that we are progressing in a forward direction. We have gone from hypothetical conceptual ideas of what quantum computers are to today having access to working systems that can be tested. We can try to evaluate their feasibility for solving challenging problems, and then we can even envision how they may grow and scale up over time. I would say, if anything, in the past five years, that pace of development, the rate at which we are moving forward, has actually increased. So from my perspective, I would say I'm very optimistic at the moment about the state of the field, as well as its development and near-term ability to deliver on some of these key outcomes that people have talked

about for quantum computing and other quantum technologies. It is not a simple field, though, in the sense that it is integrative. It does require multiple disciplines working together to try and accomplish these goals. And it is emerging from a long history of physics and making the transition over into a working viable technology. And so the fact that that could even happen on a few decades time scales, not, you know, to speak of the additional time that'll be needed, I think that's really remarkable. And so I'm both encouraged and optimistic about the rate at which things are moving forward. That's awesome. Travis, you mentioned sensing, and that led me to at least three or four branches of

quantum science and technology, quantum sensing for measurement, quantum communication, quantum computing. I imagine all three are part of the projects that Oak Ridge and you pursue, or is this a focus on a particular aspect of these? Yeah, so we have a very broad program in the quantum science and technology field. Certainly quantum computing is at the forefront of this alongside quantum communications. But as you rightly noted, quantum sensing is actually one of the areas that has enormous near-term potential. In a sense, quantum sensors are taking advantage of the physics of quantum mechanical systems

in order to improve the resolution and the precision with which we're making measurements. When we apply that to novel areas like searching for dark matter, which is one of the yet-to-be-discovered particles that could possibly contribute to the state of the universe, we know that we're going to need new types of sensing platforms that push well beyond our current limitations. Quantum sensors are one path forward in that area. And for Oak Ridge and the Quantum Science Center and many others in the field, the development of these new quantum sensors requires us to bring together existing ideas of signal processing theory and sensors and measurement, but now framed in this new framework of quantum information and how it can actually exceed our conventional expectations.

Brilliant. So if we focus on the quantum computing part, there are applications that I see mentioned in the press that look more like a quantum experiment rather than really quantum computing. They sort of seem to lack the programmability aspect that the word computing would imply. Where do you think that is in the current state of things? This is an excellent point. The differentiation right now between a quantum experiment and a quantum device operating is very murky and a bit ambiguous. Oftentimes, we are finding that experimental physicists who traditionally have stood up these systems in their laboratories as point of proof concepts of key

ideas from physics are now making the transition into persisting those systems for longer amount of times and allowing other people external to their laboratory to have access to them. The ability to transition the technology is an entire endeavor on its own. It touches on many of the topics that are traditionally well outside of science itself. So I think it is good to point out that there is some ambiguity right now in what's an experiment versus a product, let's say. But that's also part of the excitement, is that by making these experimental discoveries, we're actually pushing forward the frontier of what we can do with quantum physics, quantum materials,

all these different systems that are out there. And by providing people access to them as quickly as possible, we're simply going to feed back into that cycle of discovery, enabling people to now take those systems and program them or test them or evaluate them in these application areas. So I think you're exactly right that there is not always a clear line between what's an experiment versus what's, say, a product. But the truth of the matter is that's really where a lot of the excitement's happening because we're able to get such a quick feedback cycle in terms of our discoveries of what we can do on these systems versus how we want to design

the next generation. Travis, help us, and I'm sure a lot of people among our listeners, what's a good way or the best way to describe what a quantum computer does and also why it has the potential to be so much more powerful than classical HPC? Yeah, this is always a tricky question, in part because when we talk about computers, conventional computers today, there's a great diversity of them. We all have a general notion that, yes, they're performing some type of computation, adding up ones and zeros quickly for us and doing that with remarkable technology underneath them. With quantum computers, we're actually trying to extend that definition. We're trying to say that in addition to all those

remarkable things we have access to now, there's a whole new set of opportunities that are available by harnessing the laws of quantum mechanics. And so in the way that we traditionally think about adding ones and zeros together, we now think about superpositions of ones and zeros. And that, of course, kind of requires a stretch of our imagination for those who aren't familiar in the field. But what we have found through studying this area is that there are great opportunities when we extend the definition of operations that are available to a computer. I'll give you an example of this. At Oak Ridge National Laboratory, one of the key areas we are interested in is understanding materials,

both their properties and their behaviors, but then also how we can create new types of materials that provide certain designer functionalities. Well, in order to synthesize and characterize those systems, it is an incredibly time-consuming process, and we spend a lot of effort in demonstrating these types of measurements. But alongside that, we want to be able to build computer models of these types of materials that can then guide us in those processes of fabrication and characterization, and ultimately cut down that development cycle in order to get to our sought-after material quicker. Well, unfortunately, some of the materials we are trying to build, we will never be able to simulate, even on today's

supercomputers. They are simply too complex in the way that the atoms and electrons are interacting with each other that our conventional technologies cannot track those systems. But you might expect that a computer that is fundamentally built on the principles of quantum mechanics can actually mirror those systems almost perfectly. And so in this way, by building quantum computers, we can actually create synthetic versions of quantum materials that could not have existed otherwise. And eventually in the future, use that to guide our development of materials that can be used, let's say for room temperature superconductors. And that's exactly the type of material that now has an impact on

how we transmit electricity without loss or enable new types of transportation. Now, that's a long term goal for this type of research agenda. but it's the fact that quantum computers are moving beyond our conventional limits, that we're now able to solve problems using the methods of quantum mechanics themselves that really opens up this space in exciting and new ways. Okay. Now, we recently saw a headline that the biggest current quantum barrier is noise. Is that the same as error correction, which we've heard for years is the biggest quantum barrier? They're intimately connected with each other. Fundamentally, when we try to control quantum mechanical systems, we're having to build something, a tangible object, a material,

or perhaps we're trapping an atom or trying to create a photon. And these quantum particles are not only very small and very sensitive, but the ability to control them and bend them to our will is oftentimes a noisy process. So we try to create an atom at a certain location in a material, and it's actually off by an angstrom or two. And that can have huge repercussions for how that material behaves. So within quantum computing technology, we're actually now focused on how do we mitigate the errors and the noise that's occurring through the processes by which we create these technologies. Now, one of the approaches to that is something called error correction, which is you actually create a reinforcement method that monitors the state of your material or your quantum

system. And then when it detects an error, it will have feedback into how that system behaves to try to correct for that error. Now, the error correction methodologies that are available today and the ability to integrate them into a system that behaves fault tolerantly, that is to say that it can tolerate the presence of errors and faults, is a really complex problem and right at the state of the art of the field at the moment. So we have existing quantum technologies that we can learn from.

And one of the things that we are learning is that we are going to need error correction and fault tolerance in order to scale those technologies up to the sizes and performance that we expect is necessary to try and solve some of these most challenging problems. And it's possible to detect the noise and distinguish it from a proper manipulation of the qubits. How does it know that this is noise and the other thing wasn't noise? It was meant to be. Oh, yeah. This is really subtle. Quantum mechanicals themselves are very sensitive to their environment. And what I mean by this is if I have a material and it's not in an absolute vacuum and not interacting with anything around it, then there's always going to be some transfer of energy and noise and information between that and its surroundings. The ability to distinguish that

type of process from something that I deliberately chose to do to that material is the balance between the error correction paradigm and my straightforward operation of the quantum system itself. So one of the engineering challenges, both at the architecture level as well as the programming level, is to understand how to interleave error correction and the operations that you want that system to do. By performing error correction, I can, like you suggested, identify what were the mistakes that were made. Not necessarily understanding how they were made, but I'll know mistakes that were made. Not necessarily understanding how they were made, but I'll know that mistakes were present. And then by monitoring the operations

that I asked the system to perform, I can now drive it towards a given outcome. Maybe I'm adding together multiple numbers or I'm trying to simulate a quantum material. In all of these cases, I've got to interleave these two methods together. Now, one of the consequences of this is that I am now adding operations to my system's execution. It's effectively, in order to combat noise, I've got to do more operations. And of course, that can add noise to the system as well. So there is a trade-off in these approaches. There becomes a point in the noise and the preparation of well. So there is a trade-off in these approaches. There becomes a point in the noise and the preparation of the material where error correction does improve the system. When we are looking at today's latest results from the field, we are finding that the existing quantum

systems out there are right at that boundary of being able to demonstrate improvement in their operation by using error correction techniques. And that's really exciting because it hints that there is a big jump in the scalability of these systems coming up in the next several years. Like the benefit is growing faster than the opposite of benefit. This is exactly it. There is a bit of a challenge in keeping up with that scaling, but the level of number of control wires that you will need just to program and operate these systems, that's going to increase as well. And so you have to worry about things about the controls and the circuitry and the energy consumption and all of these aspects too. But the fundamental gain from error correction and being able to scale

these systems up, we're just starting to see experimental evidence that suggests this is really feasible. Do you have any perspective on the different so-called modalities, as the industry seems to refer to the word approach? Because it seems to me that if you start with just the fundamental particles, photons, atoms, ions, and as of recently, even molecules, that you could really harness their quantum effects. Is there any modality that looks more promising than others? Or are we still kind of in the discovery mode? I definitely think that all of the modalities under consideration are exciting areas and certainly warrant continued exploration. I am hard-pressed to say that any one of them actually stands out as being the number one contender across all use cases. Truthfully, my expectation is that the future of quantum is to have a diversity of technologies. We will find that each of these modalities actually fits certain use cases. And this isn't entirely foreign from a computing perspective. When I look at a modern computing system, I have specialized

subsystems around memory and algorithmic processing and graphical processing and the communication of data between systems. And that kind of diversity in the technology is something that we have not seen in quantum computing and other areas yet. So I fully expect that we will see these other technologies, all of these technologies, continue to develop and mature and that the real excitement comes when they start to be able to interact with each other. Because now it's a bit like having multiple types of materials for building a new empire. And in this case, that quantum computing empire, I think, will make use of all the different modalities that are out there. So this is almost like I have GPUs, I have FPGAs, I have other custom ASICs, and they all are good

for what they're good for, some more than others. But I, in practice, will need access to all of them to be able to do everything. So in that sense, the quantum computer is not one thing, but will manifest itself in multiple different ways. Is that a fair way of saying it? I think so. I think the long-term architecture for these types of systems won't be made out of a single technology element because there are competing concerns in the functionality. In some cases, I need to very quickly process instructions. So I'm going to want to use the technology that has the fastest clock available to it.

But in others, I want to move information very quickly, and I'm not necessarily trying to process things. So now I need something that can move very fast, but doesn't have these other requirements on it. And of course, this gives rise to the idea, well, now I've got to transfer information back and forth between these systems. So it's certainly, in my vision at least, that the technology, quantum technology of the future, is a collection, very much like you were suggesting, of different components and different pieces networked together. That's not to say that there won't be specialty devices that are developed along the way, but in the grandest version of the future, I do think we need a diversity of approaches. Travis, you're director of both the Quantum Science Center and Oak Ridge's Quantum Computing Institute. Could you give us thumbnail sketches of the missions of those two organizations? Oh, absolutely. So the Quantum Science Center, or QSC, is one of the Department of Energy's National Quantum Information Science Research

Centers. These are a research program established by the National Quantum Initiative back in 2018. And the purpose of the centers is actually to tackle really substantial problems, challenges in the field of quantum science and technology. QSC is one of five that the Department of Energy has sponsored. QSC itself is a partnership of 16 different institutions, including industry, academia, as well as other national labs. It includes almost 300 staff, students, and interns working on a very broad and aggressive research program, partly looking at the development of quantum materials for new devices, the development of quantum simulators for predictive capabilities and scientific discovery, and then the development of new types of for predictive capabilities and scientific discovery,

and then the development of new types of quantum sensors. And what is unique about QSC and the fact that it has all these different stakeholders is that we're able to integrate them and focus on these three mission areas and basically accomplish new types of approaches to quantum technology that wouldn't have been available otherwise. Now, the center and its sister centers that are funded by DOE are all under the umbrella of the National Quantum Initiative, and so they're very high-profile and aggressive efforts. The Quantum Computing Institute that is established at Oak Ridge National Laboratory actually has been around now for just

over a decade. Its purpose is to integrate together internally our capabilities in the areas of quantum computing and computational science. So, of course, Oak Ridge is home to Frontier, the world's fastest supercomputer at the moment. And so one of the key questions that we are asking ourselves is what does the future of high performance computing and supercomputing look like? And how does quantum computing as a technology integrate into that future computational ecosystem? I certainly don't think quantum computers themselves displace all of computing technology. Instead, it's more of a partnership and an integration of these new types of approaches to solving problems with our existing workflows.

So the Quantum Computing Institute, or QCI, actually examines that particular issue, both looking at the current quantum hardware that's available, but then also the types of workflows and application tools that we're going to need in order to integrate this technology in the future. That's brilliant. It really is such a great, great necessary thing to do. What do you think the challenges are in integrating quantum computing into supercomputing? You've, you mentioned about the application area, so that leads into the software infrastructure. But the other part that I really wanted to press on, in addition to software, is training. What is the state of that? Is it necessary to have a quantum mechanics, quantum physics background to be able to formulate the problem into something that quantum computers

can solve? Or is that something that can be abstracted away for more of a traditional programmer types to use? Yeah, this is a great question and really an exciting part of what's happening in the field at the moment. So just like we were talking about earlier, the transition out of the experimental physics laboratory into a computer lab or even data center is a pretty big jump for a technology or an idea. And what comes with that is an additional set of concerns and as well as a lowering of the priority of some of the physics concern that would have been dominated in the laboratory. But in the high-performance computing environment, there are things that you are concerned about in terms of the operational reliability, the power, weight, size, performance of these systems. How do you physically integrate them together? Some of these technologies are very exotic and they come with them, safety hazards that aren't traditional for HPC environments,

as well as requirements on their performance and their behavior. And you need a workforce that can both understand and monitor these types of requirements, but then also operate and facilitate the usage of these types of facilities. So at Oak Ridge, as part of the QCI and something we call the Quantum Computing User Program, we are actually using our access to today's quantum computers to try and train ourselves on what these types of concerns are. So some of this is learning the language of quantum computing, at least at the basic level. No, not everybody has to have a master's in physics or electrical engineering to access these systems, but they do need to understand some of the fundamental principles about what's important and when do

I know if things aren't working correctly. But even more fundamental to that is why do I want to use a quantum computer in these types of facilities. Quantum computers themselves are not commodity devices. They are high performance platforms for solving select sets of problems. And for Oak Ridge, scientific discovery and technology innovation are two of those areas. So by having our experts in those areas look at today's quantum computing systems, look at their existing application workflows, and understand how would I fit this in there? Would this even provide me a performance advantage? Those are some of the key questions, and that fits squarely in the type of work that we are used to doing, and many others are, in terms of evaluating future architectures and how they could accelerate their current

workloads. So I do think that building a workforce that is knowledgeable about the technology, at least at its functional and operational levels, is critical in order to increase the adoption and ultimately the usability of quantum computing in these settings. Travis, another topic that comes up is the idea of emulating quantum computers, both to run applications that are already available but don't have a hardware to run on, and also eliminate some of the pesky problems like IO and kind of management and stuff that is a lot easier on traditional computers. Looks like GPUs can provide a avenue to put a dent into that sort of a solution. There's also the question of quantum inspired algorithms that sometimes lead to

better algorithms, even for classical computers. Would you speak to how these two different avenues can work in parallel to accelerate things? Yeah, so the field of quantum computing is doing more than just advancing hardware and software paradigms, it's actually challenging our own understanding of what computation is and what the limits of computation are. Certainly, some of the most surprising outcomes from adopting quantum computing has been the recognition, in fact, that there are conventional approaches or algorithms to solving these problems that had been overlooked in the past. And these types of quantum-inspired approaches are entirely within reach of our current technology, but it's by examining more broadly the field of computation through the lens of quantum

computing that we've actually been able to recognize them and then ultimately adopt them. So I think just by being a foil for ourselves and our understanding of what computation is, quantum information as a fundamental object is incredibly valuable. But even moving beyond that, we have found that there are problems which are recognized as not being solved with conventional techniques and certainly not solved efficiently. And this is actually a bit more interesting in the sense that over the decades in which we have developed conventional computing platforms, including supercomputers and commodity devices, we oftentimes will encounter problems that we simply refuse to try to solve. Effectively,

we recognize that those problems are too hard. We have no efficient method for solving them, and then we can move on to focused on more feasible solutions with our currently available resources. But quantum now requires us to go back and re-examine that history. What were the problems and decisions that we made about those problems that have led us to where we are today? And how does quantum as a technology, by opening up new opportunities, new solvers, new methods, require us to now redo some of those decisions? And I'll give you kind of a detailed example just to clarify the point. But within the field of chemistry, there's a very frequently made approximation when you're trying to calculate the dynamics of a chemical reaction. And it's called the Born-Oppenheimer approximation. And basically what you're saying is that the rate at which electrons move is much faster than the rate at which the nuclei themselves move,

because there's about a 2,000-fold difference in their mass. And if you make that approximation, you can actually derive a lot of solvers for chemical reactions that are efficient and easy to simulate on conventional computing platforms. If you didn't make that approximation, it actually becomes a much more difficult problem to solve. With quantum computers, though, ignoring that approximation, continuing to keep both the electrons and the nuclei as a single composite system actually is an efficient way of solving the problem as well. So historically, whereas we had viewed this particular type of approach of solving as inefficient or intractable, now by having access to quantum technology, it opens up an entirely new

path to discovery. So I think that both in terms of the quantum-inspired algorithms, as well as in terms of the new types of problems we can solve. Quantum itself is a really exciting area. Shaheen, I'm dying for you to ask Travis, get into your thoughts about quantum parity. We hear so much about quantum superiority, quantum advantage, and when that'll happen. But you've got some interesting points about just bringing quantum up to parity with HPC. Yeah, Travis, we were talking about quantum advantage and when to use quantum. And of course, that's kind of been presented. And as you mentioned, there are problems that are completely intractable and those that can go away faster, even though they can be

done with existing computers. But the idea was that in addition to performance advantage, there's also an energy advantage for quantum computers. So if I can do half the performance, but at one hundredth of the energy or one thousandth of the energy usage, maybe that's sufficient advantage for me to use it. Is that part of your calculus when you look at these things? Absolutely. Absolutely agree with this point. And in fact, I would say that my own framework considers three specific metrics for evaluation of the value of quantum computing. One is time to solution, one is accuracy of the answer, and the third would be the energy consumption. And so like you've just said, if I'm getting the same time to solution and the

same accuracy, but I'm reducing the energy consumption by a thousandfold or even a hundredfold, my quantum technology is suddenly worth quite a bit of money when I compare it in real dollars to operating, say, an HPC system. Similarly, even if I'm consuming the same amount of energy, but if I can reduce the time to solution or improve the accuracy, there is a way to get a value, an advantage from the quantum approach. I think early on, the conversation was dominated around reducing the time to solution because many quantum algorithms, they reach their advantage by reducing the number of operations required. But that's not the only way to win this game.

In fact, I would say computing as a field is actually facing a bit of an energy crisis. When you look at the amount of training and energy consumption required in the areas of artificial intelligence and some of the modeling and simulation problems that are out there, we know that in order to create scalable systems for the future, we've got to get energy under control. And quantum also offers advantages in that area, I think. Exactly. You mentioned AI, and we also read quantum machine learning and the application of quantum computing to AI. One challenge to that and complexity to that is that AI traditionally is known as an area

that needs a lot of data. And quantum computing is not exactly the platform where you want to feed terabytes of data into and out of. So how does that work? In what way is the problem formulated or simplified to allow quantum computers to add value to AI? input-output with a quantum computing system is well recognized in that it is not easy to load data into a quantum computer. I should say it's not easy to load classical data into a quantum computer. And especially for machine learning applications, artificial intelligence, where you're doing training, oftentimes you need lots of these data pairs in order to perform that type of training operation loaded into the system. People are exploring this. They're trying to come up with ways of using autoencoders and other types of methods for compressing that information and getting as much as possible

into those systems. But it does feel like there's a fundamental bottleneck in the conversion of conventional classical bits, ones and zeros, into the quantum representation. One alternative to that is create your information internal to the quantum system itself. If I have a quantum computational model for some process, and it's the output of that process that I am using for my AI training, then all of that can be internal to my quantum computing system to start with. And this is one way to try to reduce that IO bottleneck. It does require me to have a model upon which I want to train things,

but that is actually a conventional challenge within the field of machine learning anyway. So I think there's a lot of value in continuing to explore that area. Some of it is crouched in the technology and the hardware bandwidth limits, but other parts of it may come from novel algorithm development and how to extend those applications to this setting. So final question, Travis, is how do people get engaged with Oak Ridge and what sort of resources you have available that might be a good topic of collaboration? We actually have several different points of contact that you can reach out to engage with Oak Ridge National Laboratory as well as the Quantum Science Center.

Of course, we have a website set up for qscience.org. You can go there to find out more about the Quantum Science Center and our mission as well as email and social media handles. In addition, the Quantum Computing Institute has a website, quantum.ornl.gov, where you can find more about our mission as well as our activities. I would also encourage people to look for the Quantum Computing User Program. This is hosted at the Oak Ridge Leadership Computing Facility, and it's a great way to test and evaluate current quantum computers. So please feel free to check out that. And then, of course, always feel free to send me an email. I'm always looking forward to meeting new people, learning about ways that we can partner together. Brilliant. Thank you so much.

Wonderful. Thank you as well. We've been with Travis Humble. Thanks so much for being with us. Thank you. It's been a pleasure. That's it for this episode of the At HPC podcast. Every episode is featured on InsideHPC.com and posted on OrionX.net. Use the comment section or tweet us with any questions or to propose topics of discussion. If you like the show, rate and review it on Apple Podcasts or wherever you listen. The At HPC Podcast is a production of OrionX in association with Inside HPC. Thank you for listening.