Beyond Binary

Story by:

• Mika Ono- [email protected]

Published Date

We're looking at ways we can train the trainers, where we can bring some teachers out to UC Riverside for a week-long workshop to learn methods for teaching quantum mechanics. I think the way to get kids engaged in this is to find some of the amazing things about quantum, where you can talk about quantum mechanical tunneling (where a particle can go through a barrier) or this famous problem of Schrodinger's cat. You might have heard about that, where the cat is both alive and dead at the same time. These types of problems engage kids, and they're like, "How can that be? That doesn't seem like reality to me."

Rao: Absolutely. Are there examples from biological processes that might be accessible to a larger audience, like photosynthesis? Or I've read that the retina has enough sensitivity to detect just a few photons. It seems quantum sensors are out there in the real world, not just engineered systems, but natural, existing systems.

Cybart: That's an interesting observation. I think in the biological sciences there are a lot of examples, especially if you think of your own brain, its neurons and different behaviors.

Rao: In computing we represent things as zeros or ones based on a certain state. In the world of quantum computing, what's the analogous way in which you go about using this physical phenomenon to represent the state of an underlying construct?

Cybart: A transistor, like you said, has a zero and a one, where a qubit would both be zero and one at the same time. So if you're trying to factor large numbers, for example, a classical computer would use brute force to go through and do several calculations to get to all the factors of the number, where a quantum computer would just do one iteration and have all the factors.

Rao: And hence the possibility of doing it with less energy and in less time?

Cybart: Less energy, less time. And that's the famous problem that is called Shor's algorithm that has people worried about security, because if you're trying to figure out somebody's passcode or so on, the quantum computer could supposedly do this in one calculation.

Rao: When you teach quantum physics at the high school level, or even at the freshman level, since these things are rather invisible, is that a challenge?

Cybart: It's absolutely a challenge. Even at the undergraduate level, it's a challenge, because you're trying to teach these concepts that go against human intuition. One of the first things I like to establish is that matter - things like electrons, protons and neutrons that kids know are parts of atoms - can be represented as a wave. So you teach the duality of matter versus a wave. Is it a particle? Is it a wave. Is it both? And so on. And once you start showing them things like the de Broglie wavelength of something large like a baseball, you can start to break this feeling they're embedded in the classical world around them.

Josephson Junctions

Rao: Educate us a little bit on Josephson junction devices. What are they?

Cybart: A Josephson junction is a quantum mechanical circuit that operates much like a transistor, but it has special properties in that it can operate and switch between different states with very low energy, almost zero energy. You essentially look at the difference between quantum mechanical wave functions on different sides of a junction. These junctions have to be very small, just a few nanometers. The Josephson effect is interesting. It's a recent discovery - the 1960s is recent for someone like me.

Brian Josephson was a graduate student at the University of Cambridge, and he was taking a class by Nobel laureate Phil Anderson. Phil Anderson was from Bell Laboratories, but he was doing a sabbatical in Cambridge teaching this class on many-body physics. Josephson was in his class looking at one of the problems, a calculation for tunneling between two superconductors separated by insulator. It was a quantum mechanical calculation in a paper by some researchers, and they left off a term related to the superconducting electrons that would move from one side of the superconductor through the barrier.

Rao: They left it off because it was a mistake?

Cybart: They left it off because they thought the probability two electrons would go through the barrier was 10 to the -10 times 10 to the -10: "It's a tiny number, so we can neglect that." What Josephson said was these two electrons are paired together and they move together, so the probability is only ten to the -10. Josephson wrote a paper on this, and he's the sole author. Nobody else wanted to be on it, I guess. He had a hard time getting it published. The Nobel laureate [John] Bardeen, famous for the superconductivity theory, rejected his paper as the reviewer. [Josephson] was giving talks and getting into arguments with Bardeen about this simple prospect that they tunneled together instead of one at a time. At that time, Phil Anderson went back to Bell Labs, and he said to his postdoc John Rowell, "Hey, why don't you look for this effect? Why don't we build [a way to test it]?"

Rao: Experimentally actually verify.

Cybart: "Let's do it; let's build this thing." And that's what John did. In those days, he took a glass microscope slide. He evaporated lead on top of it, let it oxidize, put another strip of lead on top of it, and measured the electrical properties between the two. And, sure enough, he sees this current that looks like a Josephson current - that's what people call it now.

A picture (below) shows the measurement system that was used to measure the first Josephson junction at Bell Laboratories. This system was John Rowell's system. It has a sweep box he made out of this old resistor and this probe. The probe is a piece of an old switchboard from Bell Laboratories that has the old phone plugs the operators would [use to] connect the calls. John built this probe to make his first Josephson junction. Then he measured out the data and plotted it on graph paper.

Rao: Josephson was a theoretician? He was not an experimentalist?

Cybart: He was a theorist. That's correct.

Rao: That's all he did.

Cybart: Exactly. [How John did] the measurement… is just so different. It's before calculators, before computers. He had this Josephson junction cooled down to liquid helium temperatures, and he had a bar magnet. He just moved the magnet and wrote down the coordinates he measured with the ruler. That's how simple the experiment was.

Rao: Sometimes those early experiments are the most accessible, right? It's on a scale that one can relate to.

Cybart: Exactly. He wrote a paper, and that was even controversial. People were like, "Oh, how do we know it's not leaking? How do we know it's a Josephson current?" There was one more prediction that Josephson made: if you measure this current inside of a magnetic field, it'll show these oscillations. When John showed the oscillations, then everybody knew it was a Josephson junction. John Rowell had a [mentee] who was my Ph.D. advisor, Bob Dynes, later the chancellor of UC San Diego, [which means I'm] three people away from the discovery of the Josephson effect.

Rao: Absolutely. Going back to CalIT2, Bob was the chancellor of UC San Diego at the time CalIT2 was established. Then he went on to become the president of UC a few years later. That's an amazing connection. Thanks for sharing that story.

What is the implication of [the Josephson junction]? Does it mean you can switch more rapidly, at lower energy costs?

Cybart: Absolutely. You can switch very rapidly. In in the '80s people thought that Josephson junctions would replace CMOS [complementary metal-oxide semiconductors] in computers because you can operate them at hundreds of gigahertz. You can really transfer the information very fast. But silicon kept being scaled up and scaled up. And Josephson junctions are very difficult because you have to cool them down to a low temperature.

Rao: What about applications in computing, sensing and communications?

Cybart: There's been a recent surge in interest in Josephson junctions again. I think during the '80s, people like [at] Bell Laboratories and IBM worked very hard at trying to scale up to larger numbers of circuits, but then they tapered off in the '90s and into the 2000s as silicon devices were starting to improve very rapidly. And now silicon devices use a lot of energy, especially when you scale up billions and billions of devices per chip. Josephson junctions use orders and orders of magnitude less energy - about six orders of magnitude, so a million times. When you start to add in large numbers of those circuits, the energy costs are very beneficial even though you have to cool them down to a very low temperature.

Rao: And what happens if it is not cool enough? How does the Josephson junction fall apart if the temperature is not maintained?

Cybart: If the temperature is not maintained, the superconducting material will actually change into a different state. It'll become normal material, and you lose the quantum mechanical information.

Rao: So it's really important to hit these targets.

Cybart: Yeah, the temperature's very important.

Superconductivity

Rao: You mentioned the criticality of having the right temperature in order for this material to be superconducting. What is superconducting? What happens when it is not superconducting?

Cybart: Superconductivity you can think of as a different state or a different phase of matter. When you cool a material down, the electrons that usually carry the charge or the information in the material pair up. I think about this when I'm driving down the road. If you're going down the highway and you're just one car driving around, it's hard to move around and change lanes. But suppose you see a big truck and the truck's just going through the highway, and you get really close to the back of the truck. They're making a big path for you, and you follow through it. That's what the electrons do. They're pairing up and finding ways to move through the material without any losses.

Rao: They're not bouncing off.

Cybart: Exactly. There are no collisions. They not bouncing; they're just moving fluidly through the system. They all can move with that.

Rao: Very rapidly.

Cybart: Very rapidly. And there's no resistance or heat lost.

Rao: And they flow around in loops or circles? And you can flip the direction in which these things flow? What's the degree of control you have once you are in the superconducting state?

Cybart: Absolutely. We recently were working with our congressman, Representative [Mark] Takano, and he sponsored a community project that we're doing at UC Riverside to set up one of these dilution refrigerators. One of the things that'll be unique to Riverside is that we're going to allow people to get time on this instrument. We'll have technical staff who are part of CalIT2 and the new quantum center that we're just establishing, and they'll be able to take the quantum circuits and help people cool them down so they can go in and test their devices.

Rao: So it lends itself to shared use.

Cybart: Absolutely. Typically, for shared use, you can coordinate a lot of experiments at the same time, load them up, and then it may take several days to cool the system down to temperature. But once you're down at temperature, one team could come in and measure their circuits and then the second team could come in and measure theirs.

Rao: That's interesting. You patiently wait until it actually gets down to the temperature you want it?

Cybart: That's right.

Rao: These institutes, the Cal ISIS [Institutes for Science and Innovation] were set up essentially 25 years ago, and every one of these institutes more or less invested in advanced lab facilities, including clean rooms. But we are always looking to see with an eye on the future how this notion of creating shared infrastructure serves the evolving scientific disciplines that have emerged. Shane, any thoughts on where you see quantum computing fitting into CalIT2 at Riverside?

Cybart: At UC Riverside I think we have a lot of unique things we can bring to the CalIT2 institute in working with San Diego and Irvine. In particular, we are a big agricultural school. One of our biggest patents and [revenue streams] is the Cutie. It's a tiny little tangerine-type orange. If we can somehow apply quantum and AI and these new technologies to agriculture, where we can be more sustainable and preserve resources and [achieve] better management of land and so on, we can work with the other UC campuses to develop a new type of program in precision agriculture.

Rao: That's wonderful. You mentioned that you got some support from Congressman Takano for acquiring some new dilution refrigerators. Tell us a little bit more about it. Is that something of strategic importance to the community, not only the campus?

Cybart: Yes.

Rao: How did that come about?

Cybart: We worked with the congressman to do something called the Inland Empire Quantum Initiative. The idea is that in the Inland Empire, UC Riverside is really the only R1 research university. There [are] no other universities that have the kind of facilities, the graduate studies and the things we have. Because of the intricacies in quantum, you really need that infrastructure. For the most part, Riverside doesn't have a lot of the tools that you need to actually do experimental quantum. We have a lot of researchers in our physics [department], that do theoretical studies in quantum. But we really wanted to have some infrastructure where students can get their hands on the tools and become quantum engineers, where they know how to cool down the electronics and read out the electronics. We worked with the congressmen to get the tools that we need installed in a shared facility, where people from the whole community can come in, people from San Diego and Irvine can come in, and get access to these machines and be able to do experimental quantum mechanics.

Rao: So it's happening? You got the resources and you're in the process of acquiring the [equipment]?

Cybart: It's absolutely happening. We're working on getting the lab renovated. We hired a research professor who is going to run that facility and be able to take chips and different materials from the different researchers who want time on the machine. We hope to have that launched in early 2025.

National and International Quantum Scene

Rao: Tell us a little bit more about the interrelationship between the massive new investments we are seeing across the board in the CHIPS Act area and quantum computing. How will the two play off of each other?

Cybart: I think, in the CHIPS act, most of the support went through the Department of Commerce, but there's also some support that goes into NSF [National Science Foundation] and the Department of Defense. A lot of these initiatives really embed quantum in with that. In particular, the Department of Defense established multiple hubs across the United States to do infrastructure in these different research areas and quantum was one of them. NSF also is doing a lot of funding in quantum to [help] the US to have a workforce that's quantum literate.

Rao: From a communications perspective, noise is an important part of what you confront, and you come up with coding techniques and other kinds of techniques to overcome noise. Is this something that is a fundamental aspect of what you need to be studying?

Cybart: Yeah, noise is something that's really holding people back in quantum computing right now. There's been tremendous progress over the last 5 to 10 years in identifying the sources of the noise. Essentially with the qubit, you're trying to maintain a quantum state, a super superposition of these two states. The bit itself is a one and a zero, and then you may have another bit next to it that you're trying to couple to that, and any kind of noise can disrupt that or collapse it into one of the states prematurely before you can actually do your calculation. I think we've identified many of the sources and a lot of it comes right from the materials themselves.

Rao: What's the startup scene like today for quantum applications computing?

Cybart: The startup scene definitely seems like it's peaked, where now it really takes a lot of revenue to get into this area. Around 2010 or so, there [were] a lot of startups coming out of some of the big quantum groups, especially UC Berkeley - there were some good researchers there - Yale and so on. Now, the big players are IBM and Google. Amazon has gotten into it as well.

Rao: Internationally, which countries seem to be making the most progress and perhaps investing the most in quantum?

Cybart: China's investing quite a bit into quantum, but it's not clear what their progress is or what they're doing. Everything is not really open and public. It's the same with the US: IBM shows some things, and Google shows some things at meetings and conferences, but there's a lot of information we don't really see. There are a lot of other companies and national labs in the US, so I would think the US and China are the big investors.

Rao: What about EU? Do they have focused programs in quantum?

Cybart: Absolutely. EU and Japan have always been strong in superconductivity.

Rao: Do you get into things like biological applications? Is this part of what you do? Photosynthesis is essentially a quantum phenomenon.

Cybart: I don't know much about the biological because, most of my background is material science and physics, but people have thought of using quantum computers for developing new drugs and for understanding how different proteins fold and things. I hear about these applications.

Rao: The application I will never forget for the rest of my life is somebody calculated the carbon footprint of a loaf of bread, and it's pretty substantial. A very significant part of that comes from the liquefication of nitrogen. So to make urea, which is typically the fertilizer they use, you liquefy nitrogen, and the energy cost of that is a huge part of the carbon footprint of a loaf of bread. Why is it like that? Well, the Haber-Bosch process, from what I read, is what is used, 100 years from when it was first discovered. Searching for the right catalyst to make these processes more efficient will have a huge impact on how we'll feed ourselves without destroying our environment.

Drug material discovery, massively parallel material discovery on extremely large scale, seems to be something that lends itself very well to quantum computing. I keep looking for other applications - traveling salesman problem, the Grower algorithm, logistics, the complexity of how logistics works in the modern world, computational support for it. But it still feels to me from where I sit that quantum computing hasn't proved itself to be useful as a utility just yet. It's over the horizon.

Cybart: I think that's what this recent National Academy study came to: there needs to be more investment, there needs to be more progress, and error correction and things like that. And we'll see what happens in the next probably five years or so. We'll know how that's progressing. But you have these other computing methods that are looking to eclipse that.

Other Advanced Computing Methods

Rao: What's the distinction between quantum computing and neuromorphic computing?

Cybart: Neuromorphic computing is inspired by your brain and how your brain actually comes about solving a problem. It's very different than quantum. Quantum computing you can think of as having both ones and zeros at the same time and running one operation to get to the answer [as] opposed to having to go serial through like you would with CMOs. Now, neuromorphic computing is more parallelized, and it's more thinking about learning patterns like your brain would. One of the examples a lot of people use is a cat. If you show somebody different pictures of cats, your brain characterizes it just by, "OK maybe it has some little ears and some whiskers, and that's a cat. That's a cat. That's a cat." You can keep showing your brain different pictures, but it'll realize what that is. It's categorizing things. Colors are another example. There are many shades of blue, but when your brain sees something that's at some particular wavelength, it's like, "OK that's blue." It puts things in bins, and that's really a neuromorphic-type computer.

Rao: Is that closer to commercial use than quantum computing?

Cybart: Neuromorphic computing is in commercial use, and it does a lot of image recognition because it's trying to mimic your brain, and there are some different types of circuits. You can do it with CMOs, and it's extremely power-intensive. But for some types of problems like image search and recognition, it seems to get the job done. There are other circuits that people have been working on, something called a memristor, which is a type of circuit that is a little more efficient than the CMOs circuits, but it also has some power problems. Then there's also superconducting logic. Superconducting logic, because it's a superconductor, can handle electricity without losses. It's another leading candidate for a neuromorphic large-scale process. I think what researchers are really struggling with is how do you scale these circuits up to be the size of the brain. The brain has so many synapses and a lot of interconnectivity. There's hasn't been a really good system developed yet to be able to mimic that, but I think that's going to be coming soon.

Looking to the Future

Rao: So where do you think we'll be in five years?

Cybart: I think in five years neuromorphic will probably overtake quantum. I think that's going to be driven by AI, because AI will really benefit from these neuromorphic type circuits. And I think quantum will still evolve, and it certainly solves certain types of problems, but the engineering involved with the quantum seems to be a bit more challenging to me than the neuromorphic, so it'll probably overtake it in terms of advanced computing.

Rao: If you let our imagination soar, where do you think we'll be 25 years from now?

Cybart: Oh, 25 years…

Rao: That's a long time, right?

Cybart: With AI and the rapid advance of that, maybe we'll have computers designing new types of computers for us and all these modalities integrated together. So you'll have quantum computing; you'll have neuromorphic computing and low-energy supercomputing.

Rao: One of the issues we worry about today is sustainability and can we produce enough energy without destroying the environment that we rely on. From this perspective, do you think quantum computing offers more promise or more challenges?

Cybart: I think it offers a lot of promise, and one of the problems that I I've heard about is in fusion. I think fusion is a very clean energy source, and if we can control it and master fusion, then we basically have the power of the sun right in our own hands. The calculations to control fusion are very intense and require a supercomputer, like the one here in UC San Diego, to control. If you have much more computing resources like quantum or neuromorphic to tackle some of these problems, I think we'll figure out ways to control fusion and other energy sources.

Rao: So then we have infinite energy practically.

Cybart: Yeah, I think so.

Rao: Thanks, Shane, for spending the time and sharing your thoughts. We are of course looking forward to creating new collaborations, not only amongst the three campuses that form CalIT2, but also other partners in the community, including startups with a focus on applications, the full stack from materials discovery to software programming environments that will activate this much larger ecosystem. I hope we'll continue this conversation and come back and share more thoughts in the future.

Cybart: Thank you, Ramesh. It was great coming down here to talk to you about quantum, and I look forward to working with you through UCR.

Learn more about CalIT2 and its UC San Diego/Qualcomm Institute, UC Riverside and UC Irvine divisions.

Share This: