3157: Quantum Meets High-Performance Computing: The Next Frontier in Tech

[00:00:04] What does it take to transition quantum computing from an abstract research concept to a transformative tool, one that impacts real-world industries? Well, this is a question I want to explore today with Michael Gleaves. He is a UK-based AI and quantum expert at a company called Eviden from the Atos Group.

[00:00:26] Eviden is at the forefront of quantum innovation and they're blending cutting-edge quantum technology with practical applications that address optimization challenges and enhance material science and even help reshape industries like logistics and finance and everything in between. And with the recent announcement of their IQM Spark quantum computer installation in France, Eviden is bringing quantum computing a little step closer to businesses.

[00:00:57] Providing physical and remote access to this groundbreaking technology. But what does it all mean? I want to hear more about real-world applications. How does it integrate with classical HPC and AI systems? And what challenges do we need to be thinking about? And is quantum computing on the verge of becoming an operational reality? And how can businesses leverage this technology to gain that competitive edge?

[00:01:24] These are the things that make me want to get out of bed in the morning and record this podcast. But enough from me. Let's get my guest onto the podcast now. So thank you for joining me on the podcast today. For people listening, hearing you for the first time, can you just tell them a little about who you are and what you do? Thank you, Neil. So I am Michael Gleaves. I am the Sales Director for Northern Europe for HPC, AI and Quantum Computing for Eviden, which is part of the Atos Group, a large French IT organisation.

[00:01:53] I've been here for about two and a half years. But prior to that, I was involved in the Science and Technology Facilities Council, one of the big research councils within the UK. And I helped to set up a centre of excellence called the Hartree Centre, which is looking at the similar activities, high performance computing, AI and quantum computing, and helping UK industry adopt those technologies into their businesses. So those are the two experiences I will rely on in order to inform this conversation.

[00:02:23] Well, it's a pleasure to have you with me today. And for people hearing about Eviden for the first time, you are a key player in the quantum computing space and have been for years. So I'm curious, from everything that you're seeing here and how this space is evolving, how has that transitioned from R&D concept to real life operational applications? How's that shaped the company's vision for quantum computing? Because it feels like all the cool stuff is on the verge of happening right now. But of course, you've been in this space for a long time.

[00:02:52] Yeah, so it sits within this large sort of environment within Eviden. HPC is the primary activity. And if I talk you through the paradigms of computing, HPC, AI and quantum, how we see them fitting together, you'll see how these things fit and can be used and utilized in conjunction with each other. So HPC is obviously a long term mature market. It's been around since the 50s.

[00:03:18] And basically, the scientists are building into an application the fundamentals of the mechanics or the simulations into an application. So this could be computational fluid dynamics. It could be weather prediction, or it could be how to develop your next shampoo. These are mature markets and the simulation is well known. It's based on silicon and it leverages Moore's law, which has gone on for 60 years,

[00:03:43] which is driving the computing levels and amounts that you can have on the silicon chips. AI works on a similar infrastructure, particularly in the training models, but relies on GPUs rather than CPUs to do that activity. And rather than the scientist encoding that in the application, it pulls the knowledge from the data and places it within an inference model.

[00:04:07] Now, both of these rely on zeros and ones within silicon, and that is the paradigms that drives those pieces. But classically, these computers suffer with certain types of application, and this is where quantum computing comes in. So Everden has been building emulated quantum computers since 2016 through a technology called Captiva.

[00:04:31] This allows you to build your application, relying on these two quantum mechanical activities called superposition. So instead of a zero and one, it's more like a flip of a coin. So it can be in any state between zero and one and multiple states at once. And then entanglement, which is it can interact with the coin next to it or the coins around it and get an inference from those two activities.

[00:04:58] And it's that different paradigm of computing which makes quantum so interesting. So Everden has been building an emulated quantum computer and then building it into a workflow. So it will work with both the classical computing, HPC, AI, and the quantum computing, because we believe that the first applications of this will be an offload of a small part of a calculation to a quantum machine and then back again.

[00:05:28] And you mentioned Captiva there. And one of the things I try to do on this podcast every day is demystify some of those new and emerging technologies so business leaders can understand a little bit more about it and what kind of value it will offer them. So can you tell me the significance of it and how it empowers businesses to develop practical real world quantum computing applications? Because that's where the magic happens from their side. The actual phrasing of it might confuse them a little bit. So can you just demystify that for me? Certainly.

[00:05:58] I'll try. So Captiva is an emulated quantum computer. So quantum computers come in various forms at the moment. You have these superconducting quantum computers which work on a type of Jefferson Junction. So IQM, IBM, and the Google systems look similar to that. You have ion trap systems. You have annealing systems like the D-Wave and neutral ion systems.

[00:06:23] So there is a big plethora of different functioning types of quantum computing. And what Captiva allows you to do is to simulate these different machines and look at how your application would fit onto each of those platforms. So right now quantum computing is in this, and more terminology, a NISC state.

[00:06:45] So where we've got an intermediate scale sort of quantum computer which has got noise embedded within that, which means the qubits are inherently unstable and only lasts for a short period of time. And what Captiva does is allow you to build your application and simulate that noise onto that system. And do it within a classical machine so you can have this Captiva system in your data center next to your HPC or your AI system.

[00:07:13] And then Captiva then will allow you to submit those applications to a real world quantum computer like an IQM system that we can talk about later on. And I'm glad you mentioned IQM because before you came on the podcast, I was doing a little research and I was reading about your partnership with IQM quantum computers.

[00:07:34] So can you tell me a bit more about that partnership and how does that ultimately enhance your ability to maybe provide those cutting edge quantum solutions that businesses and organizations are expecting now? So IQM is a Finnish-based quantum computing system. They've delivered systems to European HPC centers and they build a superconducting quantum computer effectively.

[00:08:01] We have partnered with them as being able to submit from Captiva to the IQM system for a number of years. And we've actually installed an IQM Spark system in our factory in Angers in France. So we can demonstrate that to customers, but also make it available on a service basis.

[00:08:23] We can also wrap that with consultancy and advisory sort of activities so we can help you transition your application to a quantum computer and then look at how that would perform. So then predict at what point there will be the crossover and quantum computers will outperform a classical machine for your application and what business benefit that could generate for you and your organization.

[00:08:52] And to bring to life some of what you're talking about there and maybe some of those real world use cases, what role does IQM Spark quantum computers play in maybe making quantum technology more accessible? And when I say that, I mean for areas like research labs, universities and maybe even industry players. I mean, quantum computers are really looking at two pieces. Optimization. So can you find a low minimum in a really complex place?

[00:09:18] So the traveling salesman's problem is the one that's classically understood. So an application we've been developing internally in order to demonstrate this to our customers on something that's real world is a thing called Netlist. So when we design and build a large HPC system, one of the big tasks is cabling it. So linking all the computers to the network and doing that in the shortest possible way.

[00:09:45] So using as little copper as possible, but also that improves the performance of the system and makes it easier to cable. Now we do that on a classical system, but it is optimizing for the shortest possible length of cable to deliver the network that you've designed.

[00:10:00] So we have ported that application to the Captiva system and we are demonstrating that on the Captiva application as a real world application of traveling salesman's optimization of quantum computing. In my old days, effectively, we did a similar piece of work with Ocado.

[00:10:25] So this is published work from about 2017, where they looked at routing the robots around a warehouse using a D-Wave annealing system. And we built that original application on a Captiva system, then ported it to the D-Wave. And so that's the method that we used as a sort of DevOps approach to solving that problem of how do you optimize?

[00:10:51] The other area that you see often referred to for quantum computers is around materials and chemicals. So in classical simulations, we have this calculation called DFT, which calculates how reactions occur, how electrons interact. Now that is really computationally expensive in a classical simulation, but works really well in a quantum computer.

[00:11:17] So taking a large simulation and taking the electron interactions and passing that to a Captiva system and then back again might be used for things like drug development, battery development, those sorts of material applications. So those are two of the sort of real world applications where people see the first real world applications of quantum computers making big changes to the world. And this is where it gets really exciting. We're no longer talking about quantum computing in the future.

[00:11:47] I'm curious, how are you helping your customers navigate these, the reality of this and their quantum adoption strategies? And what kind of challenges do you see in making quantum computing more accessible, more affordable? Because we are still at early days. Any challenges around that? I mean, the challenges at the moment is relative to the classical simulation driven by Smorzlodd. There is a huge economy of scale and scope, which has been built up over that 60 years.

[00:12:15] So the amount of transistors that you can get on a classical silicon chip is huge. And quantum computers sit in the sort of 10 to a few hundred range currently. And the noisy intermediate area means that you get some failures in those qubits that exist. So your logical qubits is less than that. So really the challenge is defining your problem that can be then passed onto a real world quantum computer

[00:12:44] and then get the calculation back to the simulation. So we often use mathematical techniques like Hilbert space in order to make that system work. And that's the challenge really. Translating your real world problem into something that can be passed onto a quantum machine and then passed back. Now these things will improve over time because you can see scaling in the size of qubits

[00:13:09] and the quality and type qubits that you're having within the quantum machines. But the challenge right now is making it fit. And how do we help with that? Well, we have, as well as applying the systems, we have advisors and consultants who can help you understand and translate your problems into the real world Captiva system and then test them on these applications that you need to use.

[00:13:36] And of course, stability and fidelity of qubits are essential for accurate quantum operations. So how is IQM Spark addressing some of those challenges? And again, for people listening that are just learning from my conversation today and also expand why it's so critical to the near quantum revolution that we're approaching. So the IQM system is using super cutting systems and it has a special design of system.

[00:14:06] It also uses microwaves effectively in order to hold it in place. So you can see that there's these number of dials using genetics, the shape of the system and microwaves in order to hold these quantum bits in a stable state. This is a huge active area of research. It's still ongoing today, effectively. And there's a number of ways that people are applying to it. But that's the front edge of quantum computing.

[00:14:31] How do you control, keep these systems in play for as long as possible so you can carry out the calculation and then pull that information back out into the real world? And the goal that we're all aiming for is the development of quantum applications. So how does your integration of quantum computing with HPC clusters and noise aware complies? I'm conscious when I say that out loud, some people out there will be thinking,

[00:15:00] great Scott, like Doc Brown in Back to the Future. But how does this ultimately enhance the development of those quantum applications? It is a different language, isn't it? I absolutely agree with you. And quantum is a weird space. So if you ever think you really understand it, you probably don't. So I don't think, I sometimes feel like that myself. So I think the real case is that you've got these applications, whether it's, as I said, whether it's optimization or finding this one, you know, the one application,

[00:15:29] the one best chemical within an environment where these types of use case fit really, really well. And the real job is just to build the application to understand how you're going to pass that out to the qubits and then bring them back. So if you think about like something like a drug development where you've got a set of candidate drugs that fit within a protein space,

[00:15:58] and this is a piece of work that we did at Hartree, you would then build it into a vector effectively and then pass it to a quantum computer and say whether that is good or bad compared to the other systems that you have. And then pick this one best candidate out of the thousand and then take that into experimental space. And it's that improvement of the areas where simulation or AI are not working well at the moment.

[00:16:24] It gives you this new paradigm, a new place where you can try and test these applications and see whether you get significant improvements. And if you do, like getting on that technology trajectory early, scaling up your teams in order to use this technology in the future allows you to gain that competitive advantage that everybody is seeking. So in real world terms, making a tangible difference,

[00:16:51] as that first quantum computer hosted by everyone in the UK becomes accessible, what use cases or industries do you see the first to benefit from a milestone like this? Because it feels incredibly exciting, but it'd be great to further bring it to life with some of those use cases and how you see it being used. So the system is installed in our factory in France, effectively. So it's access as a service.

[00:17:18] So UK customers can interface through a CAPT of a system and then submit, effectively. The applications that I think are logistics, so the routing problem that we've talked about with robots and the net list as an example of that traveling salesman problem. Finance, finance and risk, effectively. So finding out with a portfolio of risk, what would be the optimal position for various systems.

[00:17:47] And then anything that's got chemical or material properties in there where you've got electron interactions, because those calculations are extremely expensive to do in simulations. And there's improvements that could be made in that space. And I think that those are the sort of three areas I think that we have within the industrial space. There is also, you know, the applications around cryptography, which are being followed by nation states, effectively.

[00:18:15] And I tend to focus more on the industrial uses of it based on my background. Well, thank you so much for talking about this in a language that everyone can understand today. And unveiling how we can make quantum computing a reality across businesses and organizations in the UK and indeed throughout the world. But I'm going to bring people back down now. We're going to have a bit of fun with you before I let you go. I have a Spotify playlist that I asked my guests to leave a song to that list.

[00:18:45] All I'm going to ask is what song would you like to add and why? So Neil, I always believe that music takes you back to a time and place. And I was thinking about the time and place I'd like to take you back to. So this is from the early 90s, effectively, when I was at Leeds University in the spring. The spring, you could see the cherry blossom outside on the campus. And I would be turning over Fox Base Alpha to play Nothing Can Stop Us Now by Saint Etienne.

[00:19:14] But a moment there, I thought you were going to say Nothing's Going to Stop Us Now by Starship. So I think you chose the much better... A slightly different genre. Completely different genre. A better tune as well, I would hasten to add. So I'll gladly add that to our Spotify playlist. For anyone listening wanting to dig a little bit more deeper on AI and quantum, where would you like to point everyone listening to that? So if you put... There's a website on the Eviden system.

[00:19:41] So if you put quantum computing Eviden into your favourite search engine, you will get to some more details on the applications and how these systems work. And if you wanted to reach out to me directly, then through LinkedIn is probably the best way. Well, again, thanks for shining a light on this today. And especially learning about this new first quantum computer with both physical and remote access for users. And it's just so important to not just talk about the technology, but the project's implications, use cases,

[00:20:11] and most importantly, the need to make quantum computing accessible and affordable in the UK. Thanks so much for starting this conversation today. Thank you for having the opportunity to speak, Neil. I think as we wrap up this discussion with Michael today, it's clear that quantum computing is no longer a distant dream. It's edging closer to becoming an operational necessity, whether it be solving optimisation problems in logistics and finance,

[00:20:38] to revolutionising materials, science and drug discovery. Potential applications are vast. But don't want to oversell it here. Making quantum computing accessible and practical does require overcoming significant challenges along the way, from qubit stability and integrating quantum workflows with classical systems and so much more. But what I love about Eviden is their commitment to bridging this gap

[00:21:06] through initiatives and partnerships with IQM, for example. But over to you, what do you think about the future of quantum computing? Will it indeed become an integral part of business operations sooner than we expect? Please, let me know your thoughts. Don't forget to share this episode with anyone who you think might be curious about the evolving quantum revolution. Let's keep this conversation going. And techblogwriteroutlook.com. Please connect with me on LinkedIn at Neil C. Hughes. Love to hear your thoughts on this one.

[00:21:36] But I have indeed taken up far too much of your time. So it's time for me to get out of here now. I'm going to listen to Nothing's Gonna Stop Us by Starship and St. Etienne. Compare and contrast the two. What are you going to do? Thanks for listening, everyone. Speak with you all tomorrow. Bye for now.