From Exascale to Quantum Advantage with Bert de Jong
Welcome back to the New Quantum Era, the podcast about the people, ideas, and technologies shaping our quantum future. I'm your host, Sebastian Hassinger. Today, we're exploring the intersection of high performance computing and quantum science with someone who's been at the forefront of both. My guest is Bert DeJong, director of the quantum systems accelerator at Lawrence Berkeley National Laboratory and a longtime leader in computational chemistry and scientific computing. Over nearly three decades in the US national lab system, Bert has seen and helped drive the evolution from classical exascale computing to today's race for quantum advantage.
Sebastian Hassinger:We'll talk about the shift that brought him into quantum research, the role national labs play in the US National Quantum Initiative, and why QSA has bet on advancing multiple hardware modalities, superconducting qubits, trapped ions, and neutral atoms in parallel. Bert shares his perspective on what it'll take to move from prototypes to systems that deliver real scientific advantage, why error corrected devices with a 100 logical qubits may be closer than we think, and how collaboration between scientists, engineers, and industry partners will shape the next five years. If you've ever wondered how the quantum ecosystem moves from breakthrough experiments to practical technologies and what the road map looks like from the national lab vantage point, this conversation will give you a clear and exciting picture. Let's dive in.
Sebastian Hassinger:Hey, Bert. Thanks for joining me.
Bert de Jong:Hi, Sebastian. Nice to meet you.
Sebastian Hassinger:So you are a director of a National Quantum Centre. And you've got a sort of interesting background. So I really wanna hear about the NQI Centre. But can you start by just giving us a little bit of background on on how you got to where you are?
Bert de Jong:Also, I've been in the National Lab System now for coming up twenty seven years. Spent time at Pacific Northwest National Lab and for half of my career, and now my second half has been at Lawrence Berkeley National Lab. So I'm a computational chemist by training. My real driver underneath everything is I wanna solve scientific problems. And so my long early part of my career, I focused a lot on HPC.
Bert de Jong:I needed to run bigger and bigger simulations, so I was running large computational chemistry simulations, developing codes that would scale on exascale platforms to actually be able to do these simulations. So when I came to Berkeley, I started to think a little bit more broadly. It's like, okay. Classical computing, how much further can that scale? And we have to solve problems that are going to be bigger and bigger.
Bert de Jong:So then I started to look at alternative technologies. So quantum computing really was very nascent at that point, something that could be a pathway to provide more computational power than classical computers. And AI was there at the same time too. So
Sebastian Hassinger:What year would what year would that have been?
Bert de Jong:Oh, that would have been 2014 ish, so ten years ago.
Sebastian Hassinger:Okay. Okay.
Bert de Jong:Yes. So I started to work more heavily in quantum at that point in time. It's like, can I use a quantum computer to solve real scientific problems? What are the problems to making that happen? How can I make a quantum computer do better things?
Bert de Jong:So started to work a lot on algorithms, understanding the software that is needed to actually run on these systems, and actually figuring out where the challenges are. The noise, can we actually find ways to correct some of these challenges? So it really has been, I'm I'm a tool builder at heart too. I wanna solve science problems, but I realize I need to build the tools. Mhmm.
Bert de Jong:And to me, HPC, quantum, and AI are tools.
Sebastian Hassinger:And we
Bert de Jong:need to advance them. So so for the last
Sebastian Hassinger:You you were you were you approaching chemistry from a computational perspective right from the beginning of your career?
Bert de Jong:Yes.
Sebastian Hassinger:Yeah. I actually am isn't it? I mean, there's sort of there's sort of a dividing line between the lab chemists and the computational chemists.
Bert de Jong:Yeah. So I perceive what I do as a virtual experiment. But to give you a little bit more of tidbit, even before I started to get my, degrees in in physical chemistry, computational chemistry, I actually started with a b bachelor's degree in chemical engineering. So I had a very practical view of things, and but I didn't like the rule of thumb, and that's why I wanted to get more and more into understanding the fundamentals. And then if you become a computational chemist, you pretty much are a non card carrying physicist because you use quantum mechanics to do these things.
Bert de Jong:So Right. Effectively, life has been quantum mechanics.
Sebastian Hassinger:Right.
Bert de Jong:And, naturally, when you start thinking about what tools could be doing quantum chemistry the best, it would be a tool that is, by nature, quantum. It's really what Feynman said. Yes.
Sebastian Hassinger:Yeah. It's exactly what I was gonna say. It falls into that category. Right? If you wanna simulate nature, you need nature is not classical, so you need a quantum computer.
Bert de Jong:Exactly.
Sebastian Hassinger:That's that makes perfect sense. And and obviously, the, you know, the application of classical computing and high per performance computing and then in national labs, the exascale supercomputers, you would have sort of a front row seat on the the the performance challenges. Right? The the scaling issues around the size of the system that you can simulate. And then the trade offs of the approximations, the the sort of the scaling back of the the system that you're trying to simulate to try to make it fit into something as small as an exascale classical computer.
Bert de Jong:Well, I do not underestimate the exascale computers we have at Oak Ridge and and Argonne. We spent a lot of time, I would say, years at Berkeley and at all the national labs to really develop the the codes and the the systems to be able to use them to the max of our ability. That was done through the exascale computing program, but was funded out of DOE for for a long period of time. It has been a very productive environment. But, again, the chips are getting smaller and smaller.
Bert de Jong:We're running it there also into almost the quantum nature of these chips. And so there is a question of what's next. And that's when I started to think about what's next, quantum came as a pathway, and so I started to pivot a lot more to quantum ten years ago.
Sebastian Hassinger:Yeah. And so the the National Quantum Initiative, I think, was passed in 2017 or 2018, and it established for the DOE and the NSF a series of of centers, National Quantum Initiative centers, which are all, I think, anchored in national labs. There's sort of a national lab that the center of each of those centers. Lawrence Berkeley Lab was was I think there's QSA is is the NQI center. Right?
Sebastian Hassinger:The quantum systems accelerator. So did you join the QSA right when it was formed?
Bert de Jong:I had a small role that it was primarily led by Irvin Sadikhi at that point in time. And over the last couple of years, as leadership changed, there was an opportunity for me to take the deputy to rule, so two years ago. So I was a deputy for a year. And then at that point, the the director that was out of Sandia left to go to IARPA, and I was appointed the director. So that's about I've been doing this role for a little bit over a year now, which has very
Sebastian Hassinger:It's exciting.
Bert de Jong:It has been a very exciting time because, again, I'm a computational person, and three quarters of our center is experimental. And it's actually fascinating to learn how these technologies are advancing. The research that is being done, it it has been a very exciting and rewarding, I would say, two years because as a deputy, I already had
Sebastian Hassinger:to learn
Bert de Jong:a lot of new insights.
Sebastian Hassinger:Yeah. Yeah. Yeah. You you were, whenever, being being trained for the director role, I guess, in deputy. So so the the QSA is does the focus area or areas of the QSA sort of differentiate it from the other NQI centers?
Sebastian Hassinger:What do you see as as its role in the overall NQI?
Bert de Jong:So the QSA has taken a very interesting approach early on, and I would say most of the centers focused on a specific technology or a specific direction. One of the things that the quantum system accelerator did from the outset is say, we're trying to not choose a lane. So that we start with superconducting trapped ions and neutral atoms because those are at that point in time, again, we talk about twenty twenty one, those are the three technologies that have the highest potential to deliver a technology that could be used by the Department of Energy to solve real scientific problems. And so the research scope has been literally in that, moving those three fields forward. What you then quickly see is there is also a lot of relationships between the different technology areas.
Bert de Jong:For example, the trapped and neutral atoms have very similar challenges. A lot of research ended up being focused in, for example, integrated photonics to miniaturize, to actually avoid taking all the tools that the telecom community is using and creating bottlenecks. So but the technologies were able to move forward very quickly. So the neutral atoms, for example, is a great story within the quantum system accelerator. I would say if you look at 2021, superconducting was ahead, trapped ions was a little bit behind, and neutral atoms were still in a very nascent state.
Bert de Jong:Literally, you know, over the course of the center so far, we have delivered a system that is 256 cubits, 256 atoms, do both in analog and and digital operations and be able to do real science with them. The cool part is that a lot of the technologies that have come out of that then, through a very close partnership between the neutral Adams team and a company like Quira allowed them to actually develop as a, I would say, one of the leading companies right now.
Sebastian Hassinger:Absolutely. I mean, everybody's looking forward to their next system. I think it's going to be named Gemini. That's the word I the name I've heard tossed around. And and, I mean, based on the the work from Micha Lukan, who's leads the Neutral Atom Thrust for the QSA.
Sebastian Hassinger:Right? That role. Right. So he's also co founder of Qera. The expectation is that that Gemini device is gonna have something like 48 logical qubits that have some degree of error correction, which is really exciting.
Sebastian Hassinger:Because that's starting I mean, that's coming out of the gate with several dozen logical qubits, which the general consensus is that we get to 50 logical qubits, they're all capable of being interconnected with one another. That's a system that's not classically simulatable in any it's kind of extraordinary to see that kind of enormous leap forward as long as the Gemini system can actually do that. I think that's it's very exciting. It's very exciting for sort of this, you know, the next phase of the NQI and and of the QSA that we're at the that that beginnings of that logical qubit era.
Bert de Jong:I would say not just for the NQIs, but also for the the whole quantum ecosystem. Yeah. And that's a big challenge big change we've seen in the last couple of years. Right? We see the first demonstrations of of logical qubits that can at least break even or slightly do better, and that is something we could not have really focus have focused on five years ago because the systems were not of a scale that we could even consider that.
Bert de Jong:And and in addition, there's a lot of theoretical computer science development and just fundamental quantum mechanic development that has happened to better understand how we can use and work with error codes. And so Mhmm. For a so we're currently in a recompete situation. If we get another five years, yeah, error correction is going to be a much bigger scope within
Sebastian Hassinger:in the
Bert de Jong:center since we we'll be able to scale up our systems both in the neutral atoms, trapped and superconducting qubits. Right. Will be using different ways of error correction at that point in time.
Sebastian Hassinger:But still I was gonna say there's been that sort of progress towards error correction and logical qubits across all the modalities that you listed off. So do you think the second phase of the QSI QSA would continue that focus across all three modalities?
Bert de Jong:Yes. So we will continue with all the three modalities. We've taken taken slightly different approaches to scale. Neutral atoms is yeah. As you kinda pointed out with what is thinking about, we have a similar road map relatively speaking for the neutral atoms.
Bert de Jong:Trapped ions, we are looking at not just increasing the number of qubits, but also figuring out how we can take more advantage of degrees of freedom that might be available in the system instead of doing a two to the n, do four or five six to the n type simulations using different degrees of freedom. That allows us to scale too. Do you mean by
Sebastian Hassinger:do you mean more than just two levels in the system? So sort of QDITs or
Bert de Jong:So the other thing that that we already demonstrated as as something we already have done as early demonstrations in the current quantum system accelerator period is actually demonstrating that we could use both the fermionic and bosonic degrees of freedom Okay. Got it. The same time. Got it. So combination of those can get us a much bigger
Sebastian Hassinger:Right.
Bert de Jong:Computational space without doing huge numbers of of ions. Right.
Sebastian Hassinger:Right. Right. Right. That's interesting. And that's because of the the fine grained control within the ions.
Sebastian Hassinger:Yeah. The electron spins. Right. Yeah. Yeah.
Sebastian Hassinger:Yeah. That's really interesting. Is that similar in some sense conceptually at least to to what's happening in super com is superconducting with things like cat qubits and dual rail where there's there's bosonic modes as well?
Bert de Jong:I think it's a little different in how we are actually going to complement those two degrees of freedom. I I would say in the superconducting qubit side, what you see is actually people are using those bosonic degrees of freedom, those cat codes or the GKP state code that we have done in a actually hard coded into a qubit now to actually deal with error correction more than adding more degrees of freedom.
Sebastian Hassinger:Okay.
Bert de Jong:That's a different different focus
Sebastian Hassinger:is I see.
Bert de Jong:Getting more computational space over getting better computational space Yeah. Which is the challenge with superconducting qubits.
Sebastian Hassinger:It's interesting. I mean, the DOE National Labs have such a strong focus on science. It feels like you're uniquely positioned to contribute to this stage of quantum computing because there's there's so much scientific knowledge that's required to to even approach how to build these systems and control these systems. It's really way more science than engineering at this stage. Do you think that's that's accurate still that we're sort of in this this fundamental science phase?
Sebastian Hassinger:Or it's playing more a stronger role than it does in something like a mature technology like classical computing?
Bert de Jong:I would say we have definite there's still major scientific questions on how we can build better quantum systems. But what I would say is we see also a big trend towards engineering challenges. Right? So we can make so let's say we have trapped ions. We wanna be able to control the laser light better.
Bert de Jong:If we do that with modulators and other things, it's just large scale, physically large scale to do that. So integrated photonics is a pathway to do this. So Right. Integrated photonics, sure, there is some science to it, but it's a lot of engineering challenge. How can we actually fabricate these kind of things and integrate them into chips?
Sebastian Hassinger:Right.
Bert de Jong:I would say superconducting, it's an engineering challenge by now to get higher fidelity. How do you get more perfect materials? How do you get more perfect Josephson junctions? How do you do things like dual rail or CAD qubits in a more accurate way, a more perfect way so you can get better qubits and be able to run simulations on them, with higher accuracy? And that has to all be then combined with the error correction story, of course, and the science.
Sebastian Hassinger:Right.
Bert de Jong:What the labs to me bring one of the aspects the labs bring is is they understand the DOE science mission and and know what the problems are that have to be solved. They also have a lot of facilities to do the engineering part. For example Mhmm. The Quantum System Accelerator is a partnership between Berkeley Lab and Sandia Lab. So Sandia has a lot of fab facilities.
Bert de Jong:They make trapped ions. They actually build that trapped ions on space. They actually fabricate those integrated photonic We
Sebastian Hassinger:had Dan Stick on the podcast a year or so ago. Yeah. Talking about the big enchilada.
Bert de Jong:Exactly. So big enchilada is one of the products that come out of the quantum system accelerator. So those kind of capabilities so Berkeley Lab has entities like Molecular Foundry, the advanced light source that can actually probe, let's say, a superconducting qubit to the molecular scale to see where the errors are coming from. And we have, actually, in the molecular foundry tools to actually make chips and really understand the process of making chips. Those are engineering challenges.
Bert de Jong:We have also a very large team focused on on controls. All of these light sources and also the capabilities that we built for DOE at large, there is a lot of controls issues there. So we have a team that has pretty much built their own open source superconducting qubit control system that works actually with trapped ions too. That's called Cubic.
Sebastian Hassinger:Cubic. Right.
Bert de Jong:That comes out of the Quantum System Accelerator two. So there is to me, that is an engineering problem. So there is a lot of engineering problems that are coming up right now. BIOB is still continuing to do some of the fundamental research to just build better better
Sebastian Hassinger:Yeah. Yeah.
Bert de Jong:I think it's a combination of the two, and this is what National Labs are exactly good at.
Sebastian Hassinger:Well and you you mentioned, you know, sort of bringing that end user perspective to the the the use cases that these devices, as they start to match and surpass the capabilities of classical systems, the first use cases they're gonna be applied to will be scientific computing. Things like chemistry, putting your chemistry hat back on. You really have the the most qualified sort of user base in The United States in the national labs and the affiliated universities for for driving those initial applications of this emerging technology. Do do you do you sort of have an instinctive sense of of where that first impact might be where you we clearly we carry out an experiment where we clearly go, well, there's no way we could have done that classically. Yeah.
Sebastian Hassinger:I mean, do you have a perspective on that?
Bert de Jong:Definitely. So even though I'm a computational chemist, early on when I started to work in the quantum computing, I partnered with high end physics and nuclear physics. Turns out that the problems they have, sure, they have slightly different ideas and and energy scales, but the fundamentals on how to solve these problems are not that different. So between the three areas, high end physics, nuclear physics, and chemistry and materials materials is another area that is going to be important. We will be able and I honestly believe that within the next five years, we're going to run simulations that will be a challenge on the classical computers.
Bert de Jong:And the fun part for us is people say, well, yeah, we need fully fault tolerant quantum computers, but that's going to be ten years out. We don't need that, and that's the key part. But I've been, my instinct says that if we can get to a quantum computer that is, by order a 100 logical qubits that have a fidelity of 10 to the minus seven, knowing where our algorithms are and where we see the number of operations are going to be, we can get reasonable results, release more reliable results out of a quantum computer for high end physics, nuclear physics, MSD materials, and be able to deliver a quantum scientific advantage. Right. Is it the quantum advantage?
Bert de Jong:I don't like that word.
Sebastian Hassinger:I Yeah.
Bert de Jong:I talk about the scientific advantage of using Yeah. Quantum computers to solve real scientific problems.
Sebastian Hassinger:Right. Well, it's when it when it can produce as a tool, it can produce results that would be very, very difficult, if not impossible to do by other means. Right? That's that's to me, that's the benchmark.
Bert de Jong:Exactly. That's the the exact benchmark. And if you look at it, when you look at most of the fields, they are making approximations so they can make it fit on a classical computer.
Sebastian Hassinger:Right.
Bert de Jong:And what we know is that certain degrees of freedom that we do not include in our classical simulations are critical. And quantum computers tend to be more natural to solve some of these problems too.
Sebastian Hassinger:Right. So then the first sort of advantage will be that there you're not having to make those trade offs that reduce the accuracy of your final results. You'll end up with a more realistic simulation of the system that you're trying to simulate. That's really interesting. That's really interesting.
Sebastian Hassinger:And you mentioned, you know, you sort of said that you're a toolmaker. When you when you think about these progress towards these early use cases, are you also thinking how to sort of bridge the gap between the the experimental physicists, the systems engineering, the super specific domain expertise that goes into algorithm design, and and then making tools that somebody who's a computational chemist or a material scientist can use without having all of that background in quantum computing?
Bert de Jong:So I can now put on my other hat. So my earlier so before I started to lead the quantum system accelerator, and I still continue to, I started to lead other smaller centers that are focused on quantum software, quantum compilers, and quantum algorithms. So most of my funding has been in that realm. So we actually developed a end to end software stack in one of the programs we called AQC, and now we have a new one that that is, that started it's about a year old now that is called MachQ, so the speed of quantum. And so we're really building critical components that could easily plug into the software stacks that are being developed by the commercial entities or open source software stacks.
Bert de Jong:So we went away from building a whole end to end stack, but rather building the critical missing capabilities. Mhmm. And one of those is is a fully open source, quantum compiler. We call it Biskit, b q s k I t. Slightly different than the the Biskit you would eat.
Sebastian Hassinger:Yeah.
Bert de Jong:So that one, actually, we had have been developing for the last eight years or so that, it can now handle up to a thousand cubits without too much issue. But now the big transition again, what we've seen again, but the error correction. So our next phase is to actually build that compiler out so that it can actually build you fully logical
Sebastian Hassinger:Right. Well, and and I've been thinking about that. That's gonna be quite a challenge for compilers because the definition of what a logical qubit is and how to do logical operations is gonna be extremely different from one vendor to another, at least in this early stage. Right?
Bert de Jong:Yes. And that Yeah. We we encountered that same problem when we were building our compiler from scratch because at even at that point, six, seven years ago, there was also not a lot of consensus of what it would look like to run on a quantum computer. So we've been dealing with this forever, and that's okay. That's Yeah.
Bert de Jong:Part of the fun. If we in the classical computing world, it's now all pretty much, well, diverged again, but it became all x 86.
Sebastian Hassinger:Right.
Bert de Jong:So it's pretty standard sets of of building the tools. Now all of a sudden, the the community in the last, what, five, ten years has come moved to GPUs or FPGAs and all those other techniques. So we see, again, a diversity and a more challenging environment even for compilers there. So, but that's okay. You bring in the computer scientists that know what they're talking about.
Bert de Jong:And and that's the one big thing of any of the projects that I lead and any of the teams that I lead is the key part is diversity and teamwork. So my programs generally have computer scientists, mathematicians, and domain scientists working closely together to really move either the software stack or the technology forward to get to a point where we can actually deliver a scientific advantage or a scientific capability to the Department of Energy.
Sebastian Hassinger:Yeah. Yeah. And I guess the NQI even has you know, the National Labs have a long history of collaboration across the academic research world, but the NQI even sort of sharpened the focus on industrial collaboration as well. There's industrial members of, I think, all of the NQI centers, if I'm not mistaken.
Bert de Jong:Yes. Yes. And and I think that's going to even become more and more important as we see now a real growth in these companies. They are starting to produce systems, sell systems, but most of them are still VC funded. So their road map is pretty short.
Bert de Jong:That's two, three years, and they need to deliver. And this is an opportunity I feel for the National Quantum Initiative Centers is to not just think two, three years, but look at what's next. So what is beyond these two, three years that needs either scientific or engineering development to a point that we pretty much can hopefully derisk some of these companies' development long term road maps going forward. And and and I I come back to that point. This is why integrate the whole integrated photonics that was done at QSC has been interesting because going forward, once we start for another five years, we have Quera, Quantidium, I n q, Atom.
Bert de Jong:They're all very interested in getting their hands on that kind of a technology because it's
Sebastian Hassinger:kind of story.
Bert de Jong:Just beyond the Right. Road map that they have, but they know they need to get there.
Sebastian Hassinger:Right. And that's I mean, that's that is the classic role of public sector funded science and research. Right? I mean Exactly. It's that over the horizon kind of investment of resources that's very difficult to do in the private sector for obvious reasons.
Sebastian Hassinger:So, yeah, that's really fascinating. So so if you think about that that all of the complexity of the the areas that you the QSI QSA is focusing on, those multiple modalities, the phase we're going into with error correction and logical qubits starting to emerge, and this cross collaboration between private sector, public sector, academia. What do you see? You know, those the as you said, the the, recompete phase has started as part of the reauthorization in QI that everyone's hoping happens soon. What do you sort of hope is sort of the, you know, five years from now, you're looking back at at what you've accomplished at QSA?
Sebastian Hassinger:What are sort of the top three kind of bullets in your mind of of what you wanna be able to rattle off and point to.
Bert de Jong:Are you talking about what we, five years from now, what we might have accomplished? Goal, honestly would be Bert. I'm not prognostic. We have a very clear vision, and that is we will deliver quantum technologies that are about a 100 qubits. They will be prototypes.
Bert de Jong:Right? They will not be full fledged industrial system. Our target is to live prototypes that are about a 100 logical qubits at 10 to the minus seven accuracy.
Sebastian Hassinger:Amazing.
Bert de Jong:Using those to then actually start to do these early scientific demonstrations that could have a path to scientific advantage.
Sebastian Hassinger:That's fantastic.
Bert de Jong:That's the goal. And that's an extremely and I understand that's a very ambitious goal, and but we need to get there. We have had a long time where it's another five years, another ten years.
Sebastian Hassinger:Yeah. Yeah.
Bert de Jong:Think the time is now that that five years is actually not a far fledged horizon. It's a realistic five year horizon. Yeah.
Sebastian Hassinger:It's funny how there is a qualitative difference between five years from now when it's sort of a wild guess versus five years from now, but I'm really confident it actually is five years.
Bert de Jong:And I think the industry sees that too. And
Sebastian Hassinger:I agree. That's the exciting part. You look at all the roadmaps and they're all in the five year horizon. They're delivering a number of logical qubits that would be, again, very difficult, if not impossible to simulate. So, yeah, it's an exciting exciting time.
Sebastian Hassinger:So this is fantastic, Bert. I really appreciate your time. It's you know, I mean, I I think the QSA has done amazing work over the last five years, and I'm I'm very much looking forward to the next five.
Bert de Jong:So do I. If you look at the partners that we have, and and our mix is really we have a lot of universities, leaders in the universities that are doing the fundamental science connected to the kind of the engineering side of the labs and the application side. It has been an a fantastic five years already, and I think another five years, we really can put quantum on the map, so to speak, for the Department of Energy.
Sebastian Hassinger:Great. Well, thanks so much for your time, Bert. Thanks for joining us.
Bert de Jong:Thank you.
Sebastian Hassinger:Thank you for listening to another episode of the podcast, a production of the New Quantum Era hosted by me, Sebastian Hassinger, with theme music by OCH. You can find past episodes on www.newquantumera.com or on blue sky at newquantumera.com. If you enjoy the podcast, please subscribe and tell your quantum curious friends to give it a listen.
Creators and Guests
