The video features an insightful exploration of the complexities associated with quantum mechanics and, more specifically, the challenges of building quantum computers. A seasoned physicist with a strong theoretical background admits the difficulties in bringing the theories of quantum mechanics into practical computational devices. The discussion introduces an expert in the field, Patti Lee, who is actively working on developing scalable trapped ion quantum computers to realize the full potential of quantum computation.

The conversation expands on various approaches to quantum computation, such as superconducting qubits, neutral atoms, and photonics, each with its unique challenges and methodologies. Lee explains the process of trapped ions and highlights the involvement of electric, magnetic, and electromagnetic fields to manipulate quantum bits. Visuals and real videos of qubits in action bring clarity to the complex nature of quantum operations. Furthermore, the discussion touches on programming languages and how quantum computers are already being used by hundreds of users worldwide for exotic computations.

Main takeaways from the video:

Please remember to turn on the CC button to view the subtitles.

Key Vocabularies and Common Phrases:

1. quantum mechanics [ˈkwɒntəm məˈkænɪks] - (noun) - A fundamental theory in physics describing the nature of matter and energy on the atomic and subatomic levels. - Synonyms: (quantum theory, wave mechanics)

I myself learned quantum mechanics, you know, something like 40 years ago.

2. intrepid [ɪnˈtrɛpɪd] - (adjective) - Fearless; adventurous (often used for rhetorical or humorous effect). - Synonyms: (bold, daring, brave)

And our guest for this conversation is one of these intrepid scientists.

3. superconducting [ˌsuːpərkənˈdʌktɪŋ] - (adjective) - Relating to a phenomenon in which a substance conducts electricity without resistance at low temperatures. - Synonyms: (zero-resistance, conductive, cryogenic)

And so a very popular one is the superconducting qubits.

4. entanglement [ɪnˈtæŋɡlmənt] - (noun) - A phenomenon in quantum physics where two particles become interconnected and the state of one instantly influences the state of the other. - Synonyms: (interconnection, link, intertwining)

And as long as you can just pick two, two that are best for operation and entanglement, those then assign that 0 and 1 and you can do your computation there

5. decoherence [ˌdiːkoʊˈhɪrəns] - (noun) - The process by which quantum systems lose their quantum behavior and become classical due to interaction with their environment. - Synonyms: (dissipation, conversion, transition)

And what are the things that you see as hurdles to realizing that vision? And I think there are typically a bunch of words that people throw at this fault tolerance, decoherence and so forth

6. quantum supremacy [ˈkwɒntəm suːˈprɛməsɪ] - (noun) - The point at which a quantum computer can perform a task significantly faster than a classical computer. - Synonyms: (lead, dominance, ascendancy)

Now, maybe not up on the details, but I do remember some pushback from an initial announcement of so called quantum supremacy, where there was a statement, we've finally done it would take a classical computer ten thousand or a trillion years

7. fault tolerance [fɔːlt ˈtɒlərəns] - (noun) - The ability of a system to continue operation even if part of the system fails or malfunctions. - Synonyms: (reliability, robustness, stability)

And what are the things that you see as hurdles to realizing that vision? And I think there are typically a bunch of words that people throw at this fault tolerance, decoherence and so forth

8. coherence [kəʊˈhɪərəns] - (noun) - The quality of forming a unified whole, in physics, the property where waves are in phase or maintain a fixed phase relationship. - Synonyms: (unity, consistency, alignment)

For trapped ions, for our system in particular, the coherence times is on the order of seconds, so maybe like 10 seconds or so.

9. microwaves [ˈmaɪkrəˌweɪvz] - (noun) - Electromagnetic waves with wavelengths shorter than radio waves and longer than infrared light, used in radar, communication technology, and heating. - Synonyms: (radiowaves, microwaves, electromagnetic radiation)

Sometimes you can do that with microwaves as well.

10. cryogenic [ˌkraɪəˈdʒɛnɪk] - (adjective) - Relating to or involving the production and effects of very low temperatures. - Synonyms: (freezing, subzero, cold)

And it's actually also under, it's in cryogenic temperature.

Crafting Qubits - Harnessing Quantum Mechanics for Computation

I myself learned quantum mechanics, you know, something like 40 years ago. And I learned it from the Nobel prize winner Norman Ramsey when I was an undergraduate student. And since then, I've pretty much immersed myself within the, the subtle intricacies and qualities of quantum mechanics, really ever since. And when it comes to quantum computers, actually a real one that people are trying to make, I understand reasonably well, of course, the theoretical ideas underlying their operation. But if you came to me and tasked me with actually building a quantum computer, I assure you I would fail miserably. Now, in part, this is because in my particular physics upbringing, my training, I was so focused on theory, on the mathematics, that somehow I basically sidestepped just about every hands on laboratory course. But even so, even if I did have some experimental training, there is an absolutely monumental challenge to translate the theoretical ideas of quantum mechanics into an actual computational device. You know, not animation, not visuals, but an actual device that can leverage the powerful, delicate qualities that quantum mechanics provides. Now, many companies are trying to do this, and some of these companies are claiming great success that we are heading toward the day when quantum computers really will reach their potential. And our guest for this conversation is one of these intrepid scientists, and she's going to illuminate for us where things stand and how far we still need to go to realize that promised version of quantum computation.

So I'm pleased to introduce Patti Lee, who is the chief scientist of hardware technology development at Quantinuum, where she leads the team that is planning and en route to building a scalable trapped ion quantum computer. Thank you so much for being here. As I'd like to focus on taking the ideas of quantum mechanics and the ideas of quantum computation, actually turning it into a device that would carry out these processes. And so what I wanted to begin with is just some of the various approaches that people have tried and many are pursuing to, to build a quantum computer. And I gather the basic idea is you've got to find some physical structure that realizes this notion of the quantum bit, these bits that can be in a blend of, say, 0 and 1. And just as a kind of list of the various approaches that people have tried, together with the locations and the companies that have been involved. Here's just sort of a sampling. Can you sort of just illuminate us, you know, in just a minute or so for each, what it is that they are doing in these approaches? What are they building to try to create these quantum systems?

Yeah, so there are actually a lot of different objects that are quantum mechanical or have quantum mechanical Behavior and anything that has quantum mechanical behavior, you can pretty much make it into a qubit. And so a very popular one is the superconducting qubits. And so these are superconducting devices called Josephson junctions. They're pretty well known and they have, you know, very quantized states that are, when you build the device, you can predict and you use those quantized states and you pick the two most lower energy state usually, and you can use that for computation. And they're sort of like your 0 and ones, these lowest energy states. That's right. And in some ways you can. A lot of these quantum objects have many, many different states. And as long as you can just pick two, two that are best for operation and entanglement, those then assign that 0 and 1 and you can do your computation there.

And so that's for the superconducting case. If we move on to this other example, the one just underneath it, the neutral atom. What is that? How does that. Neutral atoms. The same thing. So atoms have lots of different electron configurations. In high school chemistry we learn all these different orbitals, electrons have spins. And so again you can pick usually a ground state levels for your computation and you can use these atomic interactions, usually with these Rydberg interactions, so that when you put two atoms together, you can entangle them. And is there a particular species of atom that people tend to focus on? Does people typically like the Group 1 atoms? Because it's all the electron structures look like hydrogen. So it's very simple. Usually has a ground state, it's nice and clean. And you know, we like the hydrogen atom because we know how it works.

How about the next here, photonics. So photons are quantum mechanical objects, as you shown, with the little particles of light. Yes, it's particles of light, but they have wave characteristics as well and they will interfere with each other. When just taking two photons and trying to interfere them, they, you know, putting them through two double slits, you get interference pattern. And so they are perfectly valid quantum objects. And so you can assign either a, you know, zero or one as a there's photon or there's not a photon, or you can use a polarization, whether it's polarized vertically or horizontally, as your qubit or different properties of your photons. And is that a fundamentally different way or are they basically the same? In other words, when I think about neutral atoms or superconducting circuits, I think of them as being at a given location. When I think about photons, I'm hard pressed to think about that. Photon is sitting still. That's right. So that's the challenge with photons, is that now these photons, they're always going somewhere. And so you have to keep them going somewhere. You can't store them, like just hold them in place. Unless you just chase after your quantum system. That's right. So sometimes you may have to put them into some, you know, fiber optics or some circuit where they just keep running around and they're easy to lose, you know, as they sometimes can get absorbed by the material. And so there are different ways of handling that, but you can definitely compute.

So there are a bunch of others. But I'd like to now really just jump ahead to the example that you actually use in your own work, which is this notion of trapped ions. So can we just drill down a little bit more on that and try to teach this theorist who never touched a piece of equipment, what it actually takes in your particular case to create one of these systems that can carry out quantum computation. And I know that we have some visuals, but can you first give me the 30,000 foot overview so I can just get the idea and then we can look at how you actually implement it?

Sure. So trapped ions, really are you just to think of them as atoms with charge? Right. So we take a neutral atom, fine atom given to us by nature, and we remove an electron and we have an ion. Sometimes you can remove X ray electrons. But basically that system, atomic system, has its electron structures that are well known. You can characterize them and every single one of them is identical. Right. Because they're given to us by nature or their atomic properties. And so they operate like, again, is there a particular atom again here, that. Again. Now we want the group two atoms for our trapped ions, typically so that once we remove an electron, that electron structure looks kind of like hydrogen. We like to keep it simple. As physicists, it's the easiest way to operate. And so that's. And the nice thing about these trapped ions is they're often naturally used for atomic clocks. Best atomic clocks are aluminum ion clocks, the same thing. And so these ions have very nice, very long coherence times naturally. And so it's. It'll maintain that quantum memory for a long time as we do computation.

All right, so the basic picture then is starting to form. You have one of these group two atoms. You strip off an electron, you have this ion. I gather you're going to tell me now that it has charge, you can manipulate it more easily because you can use electric, magnetic, electromagnetic fields to do so. How should I think about the 0 or 1 structure here is it again, excited states of this. It's electron. Electron excited state. So it's electron. So if you look at the electron structure, you know, you get the ground estate manifold, the lowest energy state, lowest energy state. That's like your zero or something. Yeah. And usually there are these hyperfine splittings, as in the electron spin and nuclear spin interact with each other. So you can find two states that are very stable. They're used for atomic clocks. And that's what you call, and you just label it 1 and 0.

And as long as you can manipulate it, and typically we do that with lasers. Sometimes you can do that with microwaves as well. And as long as you can control it, there's your qubit. So let's then take a look at some graphics that will allow us to dig a little bit deeper. So tell us what we are seeing here. Yeah, so this is a ion trap. You see those electrodes on the surface there? You apply both radio frequency and static voltages to those electrodes. So there you saw earlier with a neutral atom coming through, literally a hole on the device itself. And we can strip out the electron using photoionization with the laser. So you just zap it with a laser, kick out an electron, and now have this. Now we have a charged particle. Now, if we apply the right fields on those electrodes, right voltages on those electrodes, you generate the field, and that will capture the ion.

And so the yellow structure there that looks like it's containing the ion, that's the potential coming from this electromagnetic field that you have created. Electric field to hold it in place. Yes. Actually, it's just electric field because these are charged particles. This is why neutral atoms, you can't capture it like that because there's no charge. But with charge, it's easy to do that. Okay, so that is step one, I gather. And then what are we seeing here? So with these electrodes, you can see there are lots of little electrodes on the ion trap. And so we can actually change the voltages on those electrodes and move these ions into different regions in the device. And there we just moved it into the laser, what we call the gate region, where we can shine the laser on it, on the ion, and then perform a quantum operation. So you can put it in a superposition of these, any arbitrary superposition you can define for this.

So you can put that trapped ion in one of these superposed states and then allow it to start to carry out operations that are in this quantum domain. Parallel computations of. That's correct. Yeah. You start out with putting a qubit in the superposition. And a key thing is described in this graphic. So can you just take us through this? So even so, in slice superposition. You need to actually entangle two qubits. And just quickly remind us what entanglement is. I'm sure most people are familiar, but just a quick explanation. Yeah. So when you have two different quantum objects, right. They may have their own quantum mechanical description. Right. Quantum state and quantum superposition. But when you. But they have to interact. Interact somehow. So that when you entangle them.

Then all of a sudden their quantum mechanical state is dependent on each other. So this is some of the. Like a Bell state. Where the qubits or electron spins always correlated. That is in form of entanglement. And so to do quantum computation. You have to be able to entangle your qubits. And you have to do it in a controllable manner. So that, you know, you can actually form these massively entangled state. Using all of your qubits. So that when you're computing, you have that two to the n number of states. And so can I think of it that in essence, in the language of quantum computation. entanglement involves this profound sharing of the information. Between these members of an entangled state. So that they're no longer just acting independently. Which is more what a classical kind of computation would do. This guy's doing his work. This guy's doing their work. What you want is them to be working together. In some deeply enmeshed manner.

To leverage the full power of quantum mechanics. And that's what you do through these entanglements between the trapped ions. That's right. And if you think about it classically, Right. If you do a classical operation between two bits, right. You may. Then they are dependent on each other during the operation. But then afterwards, then when they go and operate with another bit. Then those become sort of separate entity. In quantum computing, they're always entangled. As in, if once they're entangled, if I take one qubit. And then go entangle it with another qubit with a third qubit. That third qubit is also entangled with this first qubit that it has never met before. Right. And so these are thus how you get the power of quantum computing. In that exponential sense. Compared to classical. Compared to a classical system.

Now, these are sort of nice visuals, animations that give us a feel for what's going on. But you actually have realized these states in an actual physical device. So what is it that we're seeing here? So that is an ion trap. That is the structure, those electrodes that are holding the ions in place. And so that is microfabricated device. You make these structures and if you connect them up to with the right voltage control and like I described you, ionize the atom and those ions can be moved around sort of above the surface of this device. And so are there many ions that are influenced by this device or is it 1 per ion? How does it work in terms of actually realizing it? The device are just a bunch of electrodes, so they can hold as many ions as you're clever in designing your electric field that you can support those ions.

And so this is an H1. This is actually our H1 system. It's currently holding 20 qubits and users are programming it and moving these ions around and doing quantum computation with it. Now, no doubt these are quite delicate quantum systems, so I gather. And I have an image here where you actually put these types of objects. So what are we seeing here? Yeah, so you're looking at that device is actually inside an ultra high vacuum chamber and it's actually also under, it's in cryogenic temperature. Can you give us a sense of scale? I mean, how big is this thing? The vacuum chamber is kind of like a football size. And so it just sits in the lab. The cryo system is not like the chandelier, but it's actually just flowing and cooling the device itself. It keeps the vacuum clean. So typically our vacuum pressure is about 10 to the minus 12 torr. So these are really, really low vacuum.

And the reason we want to operate under ultra high vacuum is we don't want all these background gas and all these different molecules and atoms come in and knock those, your nice clean qubits out of the trap. And where is that device? I mean, is it sitting nearby? Are there any of those? It's sitting in our lab right now. Colorado and Minnesota. Yep. And is this the kind of thing where you have many of those and you're doing various experiments with each? Is this like a one of a kind? We have many of those. So we have a few sort of what we call commercial quantum computers that we let users use. They're fully operational, can do all the computational operations needed. And we have several of those devices that people can access online. We also have a lot of test beds where we're testing out new devices and new ways of operating these quantum computers so we can build better and bigger quantum computers in the future.

Now you mentioned that lasers are an important part of the ways in which, for instance, you can ionize these atoms and affect them in various ways. What are we looking at in this image here? Yeah, so that's what is surrounding the vacuum chamber in our lab are a lot of optics and lasers. All of the quantum operations are actually driven by laser interaction with ions. And so these ions have these natural energy levels that are given to us by nature. We characterize them. And then now we have to go and build lasers that are. That control these atomic transitions given those. The wavelengths that are given. And so that's why you see a whole rainbow of colors and spans from the infrared to the ultraviolet through all the visible colors, because there are a lot of atomic transitions here. We use ytterbium ions. And so ytterbium ions do have some ultraviolet colors. And then there are also. We also have barium ions for what we call sympathetic cooling.

That is to keep the ions around and laser so we can laser cool these systems while the ytterbium ions holding quantum qubits and quantum memory and performing the quantum computation. So you tune the frequencies of these lasers based on your understanding of how these particular atoms respond to the presence of light. Photons. That's right. That interact with them. And that gives you presumably, a precise set of tweezers to move them, a precise set of kicks to change their state. And that gives you the capacity to manipulate these with fantastic precision. That's right. That's right. So, I mean, we don't have the tweezers in the sense that tweezers are typically used for neutral atoms to hold the ion, to hold their neutral atoms in place, because neutral atoms have a charge. So their handle is an optical tweezer. In.

By the way, my tweezer reference is totally poetic, but I'm glad you interpreted. But we do have these very sort of narrowed laser beams that hits the ions precisely. And yes, they are absolutely tuned to the transition. So they're driving exactly the coherent operations needed to perform these entanglement and operations or putting them in position. And then when you put this into further practice, and I gather. Is this animation or is this. What are we seeing here? You're looking at actual ions, actual qubits running around in a device. So this is our H2 system that looks like a racetrack. And why race track shape. Is there a particular reason for that? It's just a matter of how we pack them in and how we arrange the qubits. Right now, this H1 and H2 are sort of more linear type of scale linear system so that the ions are rotating or stored in this linear fashion. And you can see the ions are running around the racetrack in that video.

That's an actual video of ions. We're seeing real qubits. There on the left is a device. You can see all the little electrodes on the device that control these ions. And so just by changing the voltages on those electrodes, we can make those ions move around. And so how many qubits are we seeing involved in this? So this device, the picture, the video you're seeing, right, there are 32 qubits. And so the thing is, that number of qubits is not. It's really how we manipulate, how we control those electrodes. So this device is actually under beta testing right now for more than 50 qubits. What's interesting about more than 50 qubits is that it becomes. The Hilbert space becomes so large that you can no longer simulate that even with the most powerful supercomputer in the world. So when you say Hilbert space, you're talking about the possible configurations in the quantum domain that these particles can be in is just so numerous that you don't even have the capacity to do a classical simulation because it's just too big.

Yeah, yeah. And so that's a really exciting regime, right, because now we can actually do computation, you know, at some more than 50 qubits, you know, when that's. That's fully operational, that a supercomputer can't calculate. And so this all sounds extremely promising, you know, for the field. But if we can turn to some of the challenges that you're facing and actually implementing this in a manner that, you know, I don't know if the goal one day is that everybody has a quantum computer on their laptop, but presumably the goal at some point would be that everybody who had an interesting problem that would be amenable to a solution with quantum computation would have a stable, large quantum computer out there somehow in the cloud that we could access. We're not there yet, presumably, but this is the direction that you're heading.

And what are the things that you see as hurdles to realizing that vision? And I think there are typically a bunch of words that people throw at this fault tolerance, decoherence and so forth. Can you sort of take us through what these hurdles are and where you think we are in the journey to surmount them? I would say the biggest challenge is actually driving down error rates. Right. So these physical systems have errors. That's coming from some of it is Physics, fundamental physics, a lot of it is in more instrumental environments that we just simply cannot control them. And just concrete. When you talk about an error, you mean when you ultimately measure it, you wanted it to be a zero and it's a one or so. I mean, is it that level of. Yeah, but every operation, every entanglement operation or within the working of. Within the. Yeah, within your circuit introduces error. Right. Because we can't do it perfectly.

And so right now, for example, in trapped ion, on our trapped ion system, we're able to get 99.9% fidelity. That means, if you want to think about it classically, we do a measurement one out of 1,000 times, you get a wrong answer. But of course this is in the quantum mechanical space, so it's a little bit different, but that's the idea. So there's always some limits. And I would say 99.9% is probably best. I mean, that sounds best in industry right now. Yeah, but I guess. But it's not good enough because you want 100,000 operations in your computation. You can't get there. And so how do you get better? So obviously you can work on it on a physical level, but there's this concept of fault tolerant quantum computing, which is taking a bunch of physical qubits to construct a logical qubit.

So just want to understand that seems to be a key idea. Yes. Because normally when we think about quantum computation, you think about each qubit as carrying the information that you want it to carry. But I gather you're saying that you can take many of these qubits to actually carry the information of one. And by having many, presumably you can have redundancies and you can have error correction built into it. And is that the way forward? Is that the way to minimize these issues associated with error? Yeah, and I think, you know, classical computing does that all the time. Right. When we pass, when we're sending packages, packets and or, you know, you write some, you know, your image on the CD or something, a lot of times bits gets lost and there are ways to recover it. Right.

And so you can build that in, you know, with redundancy. With quantum mechanics, it's a little trickier because there's no cloning theorem, so you can just straight copy it. So just quickly tell people that you, I mean, the no cloning theorem, what does it tell us? So if you have an arbitrary state, my quantum state, you can't copy it. Right there. quantum mechanics forbids quantum mechanics. Right. Just the theory itself says you can't do it. Right. And so you can untangle it. Right. And create something that's an object that's related to it, but you can't copy it exactly. Right. Then you can just go measure that and preserve original. So redundancy. You have to be much more clever in the quantum realm, but people have found ways of being much more clever.

Yes. And there's this whole field of quantum error correction, fault tolerant quantum computing, that describes how you encode logical qubits. There are many protocols to do that, and they apply in different ways. And so the idea is that once physical error rate hits a threshold, if it's below a threshold, then you can just add more qubits to form these large logical qubits and get better and better error rates. So what ratio are you dealing with now? I mean, how many of these qubits do you have? Encoding the information in this logical qubit depends on how good you want the logical qubit error rate. Yeah. And where are you now? Where are we right now? We just demonstrated with Microsoft just last month that we can get it to 800 times lower error rate by creating these Bell states using logical qubits. And the encoding rate is about six to one or seven to one. Six or seven of these physical ones to one logical one.

That's right. But it's very system dependent. Right. If your error rates are higher in other systems, then you may need more qubits to get to the same error rate. But in the example you gave before, when you had, say, 50 qubits, should I then think about that more like seven logical qubits? Because you need to build in this quantum redundancy. So, yeah, so it depends on what kind of computation you want to use. If the error rates at those N50 qubits are enough for your computation, then go ahead, use the physical qubits. Right. But if you somehow you need deeper circuits or better error rates than what the physical qubits can offer you, then you have to construct these logical qubits to get better performance.

So, yes, now, but then your 50 qubits now becomes, yes, more like, you know, seven or fewer, you know, logical qubits depending on the construct. Right. But it will give you better error rates. And so when people talk about, you know, when we have, you know, whatever the number is, a few hundred qubits, the number of operations that will be happening is on par with the number of particles in the observable universe. And therefore that sort of beyond any thing that you could ever imagine requiring, we would really need to scale that up probably by some factor of 7, 8, 9 or 10 or whatever it is to have this truly fault tolerant version of that. Possibly more. Right. Depends on how much error what the error rates you need. If you need Shor's algorithm you might need 10 to the minus 10 error rate. Then you need a lot more physical qubits to construct logical qubits to reach that kind of error.

Now when you mention that people can actually log on and make use of the system, what sort of interface does the system use? How does one talk to the quantum computer? Yes. I mean I'm from the old days. The last time I really did serious computing it was with punch cards. So this is not really what. That's a bit of an exaggeration but so how do you talk to it? Yeah. So there are actually several programming languages that people that's available out there. We use what's called quantum assembly language is what are our quantum computers accept. But you can use other languages and use compilers to compile it to the proper for the proper input for our systems.

But the idea is it's just look like Python code. Right. There's really. Yes, you can use Jupyter notebook and there are standard. This is how you entangle this one these two qubits. This is how you put it in a superposition. You can define its rotation angle and your gate operation and you can define which qubits goes with where. You can define when the measurement happens in our system. You can even take the outcome of your qubit measurement and then change your programming so it's an if statement and change your programming so if it's zero then you do this. If it's one, you do something else. And how many people make use of this? We have hundreds of users around the world and programming.

So for example, the Microsoft demo they're forming logical qubits. All they do is program lines of code and our machine just runs it and produces that result. So you can even construct logical qubits that we're using just programming lines of code. So it's very exciting because anybody you can go and actually write program to generate these very exotic quantum states that has never existed in the world. I'm definitely tempted to do this. Yeah. And no doubt will. So this is a huge excitement. But I want to talk about one final piece of physics before we end which is this other issue that goes under the name of decoherence which of course is intimately related to a lot of what we've already spoken about. But when anyone hears about These qualities of the quantum world superposition up and down qubit 0 and 1, or entanglement.

Distant objects somehow sharing information. We all, I think by this point have acquired the correct intuition that these are really delicate systems. And if you do anything to them, typically any kind of interaction, you are in danger of losing that pristine quantum quality which is required for the quantum computation to do what you want it to do. It's worse than that. If you just leave it there, it will by itself, it will interact with its environment and decohere. So what's the time scale in your quantum computer for maintaining that so called quantum coherence? Not allowing it to decohere into a more classical like state. So that depends on the system. So for trapped ions, for our system in particular, the coherence times is on the order of seconds, so maybe like 10 seconds or so. Right.

And there are other groups out there that have demonstrated, you know, putting these ions in the system and for half an hour they can maintain coherence. But like in yours, when you say seconds, that, does that mean that whatever computation one does in terms of computer time, it has to complete within a number of seconds or else it just is going to fail. If you're using physical qubits, that's true. If you're using logical qubits, the nice thing is that then you can go back in quantum error. Correct. So you sort of recohere the system with your knowledge of where it was when it decohered. Right. So the way quantum error correction works is that you have these ancilla qubits that goes in and do a syndrome check. So the trick is you have to check if an error occurs, occurred in your system in your logical qubit without actually measuring your logical qubit. Because if you measure it, then you, it's gone, you've destroyed it. That's right.

That's right. And so, but you can check and see, you know, is the collective property that an error occurred. And then if an error occurs in one physical qubit, then you can go fix it. Right? Yeah. And so that's, that's what we, that's what error correction does. And it extends that coherence time. Right? So yes. So our logical qubits has much longer coherence time than our physical qubits. And so just for the final part of our conversation here, just to prognosticate a little bit, where do you see this going? I mean, presumably you're working on all these challenges to increase the coherence time, to decrease the error rate and so forth. What is the end game?

Well, certainly, you know, we want to be able to do these very complicated computation, right? But right now, I think the resources required for some of those really complicated computation, or even Shor's algorithm, you know, or quantum chemistry applications, those require a lot more qubits and a lot lower error rates than what we have today. But what we do have today is that we are already getting into this quantum advantage regime where we can run calculations that the largest supercomputers in the world cannot compute. Google had demonstrated it, there are groups in China that has demonstrated it. And our system, when it has 50 qubits, will be able to do that as well. Now, maybe not up on the details, but I do remember some pushback from an initial announcement of so called quantum supremacy, where there was a statement, we've finally done it would take a classical computer ten thousand or a trillion years.

I don't know what this statement was. And then others came along and said, well, actually, if you have a more clever algorithm on the classical computer, kind of do it in a couple days, right? So there's this thing in classical computing, it's more of the techniques, right? Yeah, because techniques to simulate this gets better and better. And some of the tensor network methods, you can say if people find clever techniques, they can do that, right? They can simulate it better. But Google recently this year came out with a larger chip, right? And the thing is that if you have just add one qubit, all of a sudden you have like twice as many possibilities. So the company gets harder very quickly. And so they came up with a larger chip with a better, better error rate than before and, you know, performed the same thing. And now that is guaranteed that the supercomputers cannot keep up.

Now what sort of computation was used as this benchmark? Was it something? Right now it's random circuits. You randomly throw together this circuit and go, okay, quantum computer. You run it. How long did that take? And what's the answer? And then classical computer, supercomputer, can you compute that as well? But when you say random circuit, is it a challenge that is in the sweet spot of a quantum computer to kind of, you know, stack the deck in your favor, or is it really something that one would care about? Right? So random circuit is more of a benchmark, right. To show that our quantum computers can do things that classical computers cannot. But do you have a useful algorithm, right. A useful problem that can be encoded and fit onto that system and compute it in that regime. And right now it's not clear what that is. And I think that's where we need the theorists and we need the application engineers to go in and start programming and trying out and apply to their application.

So we're in this sort of middle regime where the known algorithms we know we can't get to, but we are better than classical computer. And so what's in that computation space that we can actually take advantage of? And that's where the sort of new research is, because now we have a tool that never existed before. So if we were to have this conversation, say, at the World science festival in 10 years from now, no doubt I won't be in this chair. But if we had that computation, we had that conversation, do you think it will now be hands down that we have achieved quantum supremacy in a completely convincing manner and that we are really leveraging the dream that these systems can do what they've promised us to do? Yeah, I think so. And I would say that we are in that quantum, you know, quantum advantage regime. It's just that we don't have the applications yet. And in the next 10 years it'll definitely get better.

I think there are some of the, I would say condensed matter physics applications and maybe some of high energy physics simulations that I think are in that sort of 50 to 100 qubit regime. And in the near term, you know, with error rates that we can reach in near term to be able to do those calculations that we can't do currently on classical computers. Well, it's all very exciting and I look forward to maybe having that conversation in 10 years. So thank you so much for explaining this all to us. Thank you, Sa.

Science, Technology, Innovation, Quantum Computing, Quantum Mechanics, Computational Chemistry, World Science Festival