Loading summary
Transcript61 lines
- [00:00]
A
The idea is that if classical devices could mimic quantum devices, then we wouldn't need quantum devices. So going back to the digital twin, those qubits, you're right, they're notoriously noisy, they're very unstable. That's the beauty of it because that's kind of at the core of what quantum mechanics is, that you can get
- [00:19]
B
to 80% accuracy with Shor's on factoring the number 21, is that right?
- [00:26]
A
We took this knowledge that we gained from our digital twin and implemented it on the IBM platform and got to 99% accuracy on Shor's algorithm.
- [00:35]
B
Can you start by introducing yourself and giving give your background insofar as it's relevant where you studied, what you studied and how you came to quantum elements. And then we'll talk about quantum elements, use of large scale digital twins and AI to make noisy real world quantum hardware useful.
- [01:03]
A
Absolutely. Thanks for having me, Craig. I studied physics and chemistry back in Israel and focusing on nanoscale computational devices using quantum dots and other components. And these were kind of the early days of, you know, those kind of nano scale computational devices. I did my postdoc in Switzerland at ETH Zurich, kind of diving deeper into the field, Started my career in the industry as a research scientist and very quickly switched to business dev and product and took the product that actually developed as a scientist to the market and that was quite a thrilling and interesting adventure and switch to a couple other companies and built my first startup for a few years was a very interesting journey, but maybe for a different podcast and a few years back alongside my two amazing co founders, Professor Daniel Lidar from usc, who is really one of the world leaders in quantum error correction and quantum algorithms, and Professor Amira Kobe from Harvard, who is a leading experimental scientist in quantum. We realized that in order to deal with those very unstable systems, meaning those quantum devices, you need to be able to use the ability of, to augment those system using classical devices, meaning creating a digital twins of those systems and best inbreed classical tools that are used today for software development in order to control them, accelerate their development and ultimately enable end user applications. On quantum.
- [03:06]
B
Yeah, and when we spoke before, I was asking whether this is
- [03:14]
A
when you
- [03:15]
B
say using classical hardware, whether this is using classical computers to run quantum algorithms, which there's a lot of work in that field, but from my understanding you're. You're building digital twins of quantum hardware and then using AI to optimize those that hardware and, and you have some pretty interesting results. Is that right? And if so, can you explain that.
- [03:50]
A
That's absolutely right. The idea is that, you know, if classical devices could mimic quantum devices, then we wouldn't need quantum devices. On the other hand, in order to allow AI and other software components to accelerate the development of quantum, you need to give those developers a platform that mimics those quantum systems. Right. As we all know, quantum systems are not an extension of classical devices. They're completely different species and completely different technology. So what we've done is we really went back to first principles and solved those equations that describe those quantum systems. We solve those equations on large classical computers, you know, supercomputers. So now you have a very realistic representation of any given quantum hardware, and it allows you to experiment and solve and develop different layers of the software stack that are specific to the hardware that you're interested in. Meaning if you are sitting in a lab developing your own system, you can use this digital representation to solve problems on how the system works and how to optimize it, et cetera. On the other hand, if you're sitting in a company that would like to develop an application, which hardware will you choose? It's very hard. They're very expensive, they're very scarce. So if we're giving you a realistic representation of the system, you now have the freedom to experiment, develop and augment different system and understand how to work on them and ultimately execute your application with certainty and confidence.
- [05:48]
B
And when you say build a digital twin, can you describe that? Is this pure math or is it something more tangible?
- [06:01]
A
So there are two components to building a digital twin. There is the physics part and there is the algorithmic part. The first thing is looking at the system in front of you and being able to describe the math that is behind its working principles. Specifically, in quantum systems, you need to be able to describe what is the modality in front of you. Are you using superconductors? Or maybe you're using ions? Right. There are many different ways today of realizing quantum devices. Just to put it in perspective, There are about 70 different quantum hardware companies out there, each one of them with their own kind of unique flavor. So the first thing is to look at the system that you're interested in creating this digital twin floor and, and saying, okay, what are the governing principles of those systems? What are the equations that, if I solve them, will give me a reliable prediction of how this system behaves. And in the same way that you look at any physical system, think about the spring. Right? In order to understand how a spring works, you need to be able to solve the spring Equation of this specific spring that you're interested in. So the same principle. Now the other thing is, okay, you're now able to solve those equations, but they're very expensive computationally. So how do you do it efficiently in a way that you can enlarge them? So once again, to put it in perspective, there are some tools out there, scientific tools, that can solve one or two qubits of those digital twins. But what can you do when you're able now to solve 50 or 100 of those qubits? You can now tap into every problem that the industry is tackling and provide the right tools because you solve the system at scale to a point where it's now very meaningful. So these are those two components, the physical representation, the ability to scale it reliably.
- [08:10]
B
And when you talk about solving for individual qubits, I mean, quantum is notoriously hard to understand. Qubits are hard to understand. Are you talking about does the digital twin simulate individual quantum qubits? Or are you talking about optimizing all of the things around, you know, for sampling or for temperature control or for all the other, you know, isolating or eliminating noise, all the other things happening around the qubit, or are you actually doing something with the qubit?
- [09:03]
A
That's a fantastic question. The third piece that comes into play when you need to provide users with the digital twin is the ability to also take into the account the environment around this system. Right? What is the noise that is affecting the system and how does it evolve over time as you are operating this machine? So let's take an analogy from a different world and then come back to quantum. Let's say you want to develop your next generation airplane. The kernel that you're using in order to understand how the air is flowing over the wing of this new generation of airplane that you want to design is solving flow dynamics, right? You want to understand how those particles are moving across the wing. So you start up, you start with a single particle, you kind of solve the equation. And now the secret sauce is how do you enlarge it to represent millions of those particles, you know, that are flowing over the, the, the wing. And if you want to take it to the next level, what does, what happens to the flow where the airplane is stalling or maybe when there is some non laminar flow or some issues with the weather, you know, kind of you have some other phenomena that are environmental now that are affecting this airplane. So going back to the digital twin, those qubits, you're right, they're notoriously noisy, they're very unstable. That's the beauty of it because that's kind of at the core of what quantum mechanics is. It's, it's a statistical system, but it's also very hard because we need them to be quiet enough in order to do computation. So what we are doing at quantum elements, we're solving the equations that not only governs how those qubits are working, but also what are the environmental conditions that are prohibiting them from working. Working correctly with low enough noise and we give the user this kind of recipe, or all of those components that are now jeopardizing the performance of their qubits. Now that they have at their disposal all of those different knobs that control the performance of their qubits, they can go ahead and develop the mitigation in order to make them work correct correctly. And that's exactly the word that the industry is using. Air mitigation, air correction, air suppression. These are all tools that are built on top of the engine that we're bringing. And because we're able to do it with such high fidelity and accuracy, and it's such a large scale, it now uses the users are benefits benefit is that they can now do it at scale with very high certainty.
- [12:09]
B
And you this is tuned to each particular hardware or the, the modality, as you put it, of the, that the hardware is using, whether it's superconducting qubits or trapped ion qubits. So what data do you need from the hardware provider to build the digital twin? It seems to me it would be not only the hardware, but where the hardware is positioned in what kind of a room and what kind of a building and all of that.
- [12:52]
A
This is exactly it. And it really speaks, your question really speaks to the crux of what we're doing. So there are two aspects here as you go up in the software stack. If we take the classical analogy, if you're developing an app, you need kind of some sort of a translational layer to the hardware that is underneath it, right? You can be very up in the stack. It can be very abstract. As you go deeper, you need to know more and more about the hardware to a point that if you're developing an operating system for, let's say GPU or cpu, you really need to understand all the intricacies and architecture of your system. So same with quantum. When we are working with hardware manufacturers, we would usually have a very intimate relationship with them. We will work with them very closely to refine this model so it will truly represent Everything that is going on with their system because it is used by their own developer in order to refine maybe the architecture or the design or how those qubits are working from an operational point of view. But when we are working with customers who are upper in the stack, that are maybe interested in application, the models can be more abstracted. And in that regard we don't need much from the hardware manufacturers. If we take another analogy outside of quantum, let's say you're developing Formula one simulator. You have two paths. As the guy who's designing the simulator, you can go to McLaren and sit with engineers and ask for all the design files and everything. And they would say, okay, we need to sign an NDA and we need to have some relationship. But you know, if you accomplish that, your model is not really true to life. And probably the people using it will be the Formula One drivers of, of McLaren. Right. But if you're, there is another path. You can borrow a McLaren, run a couple of laps on, on the track and kind of record the acceleration and the drift and everything and use that to have a more maybe abstract model of your car that will be used by novice drivers. Or maybe people are interested in developing a self driving car, not necessarily the next generation of McLaren. So you can see that you can tap into different users, Personas and user, different users and participants in the quantum workforce based on what their needs are and how accurate this twin needs to be.
- [15:43]
B
So yeah, so I mean, you're, you answered this question, but maybe in, in more layman's language, how, how do you validate that your twin is a true representation of say, IBM Heavy Hex Device or, or something like that.
- [16:03]
A
So let me give you an example. I think one of the most famous algorithms out there is Shor's algorithm. It's an algorithm that allows you to factor a number. So what we've done is we went to the IBM platform as users and kind of looked at the published information that they have on their website on the performance of their system. We know what superconducting qubits are, we understand what the models are. So we wanted to see the relationship between how those systems are behaving versus how Shor's algorithm is behaving in terms of its accuracy. Meaning when you're executing Shor's algorithm, how accurate is the result? So we went and executed shor's algorithm on IBM system and got to about 80% accuracy. And we've suspected that there is something in the relationship between those qubits and we know how to solve this relationship. It's called crosstalk, meaning when those qubits are influencing each other, in an ideal world, every operation that you do on one qubit should not influence the other. So what we've done is we've simulated thousands of times those IBM qubits with different crosstalk combination. Because if you had to do it on hardware, you will have to manufacture a new QPU with new crosstalk. Right. And then we realize that there is this sweet spot where we can inject a process called air suppression that can give us the best performance on Shora's algorithm for the specific IBM system. We took this knowledge that we gained from our digital twin and implemented it on the IBM platform and got to 99% accuracy on Shor's algorithm. So if the highest reported accuracy on Shor's in literature is about 70% when we submitted the circuit, as is on IBM, we got 80% when we implemented, based on the knowledge we gained from our digital twin, the remedy for this crosstalk, we got to 99% accuracy. So you see that the ability to close loop, to validate and to, to get the benefit of using this technology can be very quick. On the other hand, when we're working with a hardware manufacturer and we are working with one of the large public companies very intimately, we're literally sitting with their scientist and gauging how accurate is our simulation and digital twin against their experimental result and refining the model based.
- [19:02]
B
Yeah, that's, that's fascinating. How, how sensitive is that 99 to drift in the underlying hardware parameters? I mean, if IBM recalculates its chip tomorrow, do you have to relearn everything or, or, or are these optimizations applicable? Regardless,
- [19:31]
A
there are a few layers here. When you look at qubits at performance, you're mainly looking at what is known as decoherence. How quickly are those qubits moving away from their optimal conditions? Yeah, these numbers are publishable and they would usually change based on the calibration process that each of those manufacturers is doing, usually on a daily basis. There are other more kind of intimate parameters that will govern the performance of those qubits, which we can extract from just by using the system as any other user. So to your point, the model is the same model, but the noise and performance parameters have to be updated daily in order to keep track of how the system evolves day over day. And that's the benefit of using this kind of very intricate physical understanding of the system is that once you solved it from first principle, the only Thing left is to optimize the daily parameters that change due to environmental conditions. You know, things drift and temperatures. We need to remember those superconducting qubits are sitting in temperatures that are colder than space. Right? So it requires just a millikelvin degree change where everything starts to drift and change. So yeah, that's how we do it.
- [21:05]
B
Yeah, yeah. And so what, which noise channels and hardware parameters are you modeling? I mean, or, or, or is this all approximation?
- [21:18]
A
So these are models, these are physical models. They're known as master equations. They're really solving the quantum system from, from first principles, right. From describing it from the physics side of things. There are a few noise models that are very important. The first one is known as T1, T2 and T Star. These are looking more at the rate of noise accumulation or decoherence in those systems. We can do that. There is one over F noise. That's another one. There is crosstalk. And crosstalk is a very notorious one. Think about this game of Whack a Mole, right? You're dealing with one guy and you think you're done. You calibrated this qubit, but then you realize that in the process of calibrating this one, the other guy went out of calibration because they're influencing each other. So we know how to model that. And it's a very important one because when you have a few qubits, then, okay, you tap this one and you kind of think about the knobs of, you know, this analog machines. You're turning them and you kind of, you have the secret sauce, you're the expert. But when you have 100, it becomes a bit harder. When you have a thousand, well, forget it. So it probably would lead us to the next topic, which is how is AI kind of tapping into.
- [22:53]
B
Yeah, that's, that's right.
- [22:55]
A
And the last one is, is leakage. Leakage is a, is also a very interesting one because you have to remember, you know, we're going back to a bit to quantum physics. Most of those qubits are using what is known as a two level system. You kind of move between, let's say, let's call it a 0 and a 1. It's a bit more than that, but a 0 and a 1. But sometimes there is a 3 and there is a 4 and you know, those things are unstable. So that's another thing we can, we model.
- [23:21]
B
Yeah. And for listeners who are not quantum experts and for me who's not a quantum expert. So you showed that, that you can get to 80% accuracy with shores on, on factoring the number 21, is that right?
- [23:51]
A
Yeah. So we can take it from 80 to 99%.
- [23:54]
B
To 99. I'm sorry, to 99. Yeah. 80 was, was, was where the industry was at. Can you. Yeah. Explain what factoring 21 means?
- [24:11]
A
So I know it sounds like a very small number. Right. My nine year old can probably factor 21, but it's more of a way to demonst or to illustrate the fact that Quantum can do something very unique. It can solve those factorization problems that ultimately tap into encryption in a very efficient way computationally. So the reason we are all relying on RS encryption is because if the bit string is big enough for classical devices, it will take the age of the universe to factor it. And by doing that to break the encryption quantum computers are working in. You can think about it as a parallel way. It's almost like we lay in front of you all of those possibilities and the system will converge to the accurate solution in one go. And the way we've to demonstrate that is, you know, Shor's algorithm is kind of solving this problem in an effective way. And that kind of gave the industry or this field the acceleration and the confidence that we can now use quantum computers to deal with real world problems. The problem being still, that takes us to the kind of where the state of the industry today is that to develop those algorithms at scale, you need very unique capabilities and approaches. So we're still at very early stages, meaning those numbers are still small. But I'm pretty confident that as the hardware evolves and the number of cubits grow, we will start seeing companies that, and, and scientists that are solving shores or factoring shores on much larger numbers. Getting us closer to the point where Quantum can really start tapping into the encryption industry and disrupting it in the way that we think it will.
- [26:37]
B
Yeah. Okay then, and, and where does, how is AI applied? Am I, I understand the digital twin concept, but then how are you using AI and are they standard models or are you building bespoke models? Are they generative models?
- [26:59]
A
So I want to talk about here now first, and then we can maybe think about the next steps. One of the things that AI needs, and I'm sure your audience can appreciate that, is large scale curated data labeled data that is actionable and can be used to solve a specific problem. One of the things that we can agree on after a 30 minute discussion is that Quantum is scarce, expensive and exotic, meaning that each modality has its own intricacies. They're not easy to come by. And even if you go ahead and dedicate a machine to train specific AI model, the next day you can wake up and this generation of machines is already history and you need to repeat the process again. What you. The advantage that you have with the digital twin is that you're not bound by any physical hardware. You can not only create a representation of the current state, but also of future state to a system scale that is big enough that allows you to generate massive amount of data to be able to address some of the issues that the industry is facing today. The most present and critical issue that we're dealing with is noise, right? And errors. So one of the things that we are doing, and we're gonna release it very soon with our partner, is the ability to do large scale error correction using a digital twin. Now, if you want to develop, let's say you're a scientist in a company that is really interested in now dealing with errors on a large scale, and you want to develop a decoder that is solving those errors not only for the current generation of your machines, but also for the future one, you need to be able to generate a lot of labeled data that will allow you to train this AI based system and decoder in a robust way. So we're giving you the platform to do it very economically and in a very robust way. One of the things that I didn't mention about using those large scale simulator versus hardware is people sometimes hear this notion of shots. In a quantum computer, you need to run multiple shots. Sometimes it's 10,000, sometimes it's more. Why? Because when you run a quantum computational cycle, you get a snapshot into the statistical representation of the accurate answer. Meaning you need to run enough cycles in order for the answer ultimately to be statistically significant enough to represent the right answer. That's how kind of wild quantum is. When you're running our digital twins, one experiment gives you the full picture of all the statistical representation of your answer. It's called the density matrix. It gives you the whole universe of solutions. So when you have this power and at scale, with the right architectural configuration of errors, you can now as the end user say, okay, you know what, I have a new idea. I think ML, AI, reinforcement learning, whatever. If I have enough labeled data that is representative of the system that I'm interested in, I can now build the engine that will allow me to solve those future problems. So think about us as this development platform that gives you the flexibility to generate and augment and use large scale data for your AI needs. And that's one of the things that excites me the most.
- [31:18]
B
And on the, on the labeling, what exactly are you labeling? What's an example of, of the labels that you're applying and how do you curate them?
- [31:32]
A
So think about, you know, if you know that you need to do those, let's say, error correction cycles. In error correction, you're doing a process that is known as syndrome extraction. You want to understand what are the syndromes that are governing your errors and how to deal with them. If you, as the guy designing the system, can change the knobs of your hardware virtually, see how it taps into the output of the error, you now have a direct link between hardware vulnerability and error accumulation. So now you've connected the dots, right? So if your cost function is, I want to minimize the number of errors and you're now curating the data in a sophisticated enough way that allows you to start connecting the dots between the governing principles of what generates the error and how the errors are being, are manifested on this large scale. You now have the ability to start minimizing the governing components that ultimately are jeopardizing the performance, performance of your system. So because, you know, one, one of the things that we always say about the quantum system is the moment you ask the quantum system a question, it collapses. Right? It kind of the, the notorious thing about quantum is you cannot stop at a given point and say, okay, what is your state? If you're asking the quantum system, what is your state? You've, you've killed the process. The process will not continue. You cannot say, okay, what's your state? And then, okay, you can now continue. But you can do that with a simulator. So it gives you this superpower to stop at every given point, understand the underlying mechanisms, and continue. This allows you to be very selective on the parameter space that you are using, how you're training, how you're labeling. You almost have the full recipe of what is going on, with the caveat that obviously you need to make sure that then you go and test it on hardware and et cetera, but at least you have the control knobs at your disposal.
- [33:53]
B
Yeah. Can you sort of walk us through the pipeline from a user uploading an algorithm through the simulation engine to the AI training on those runs and proposing a remedy for the hardware and optimization for the hardware.
- [34:18]
A
So if you join our platform today, you will be asked a few questions. We will ask you, what kind of experiment are you running? So that's for an example. Move away from error correction and let's Talk for a second about error suppression. We'll describe what error suppression is in a minute. So what kind of experiment are you running? Are you connecting to a physical machine, or do you want to create your own virtual machine? If you're connected to a physical machine, we can extract the system parameters for you and kind of feed them into our model. What kind of modality are you using? Let's focus for a moment on superconducting qubits. And what is your system size? Let's assume you want to use a grid of five by five. Now, we will give you a canvas. You will have a canvas, a digital canvas with those qubits. You can now control the connectivity between those qubits, and you can literally go to each qubit and update the system parameters, the frequency, the T1, T2, the leakage, the crosstalk, all of those things that I talked about. And from that point on, this system will behave as a physical system. You can upload a circuit. So there is a tab. You press the tab, you upload your quantum circuit. And now you can run this simulation. And we allow you to inject pulses into your circuit to deal with the different noise sources that you have. So now think about the following process. We give you the tools to create your machine. We give you the canvas to run your circuit in. And it's a very fine detail, but it's important. The circuit runs in the time evolution domain. It means. It means it behaves in the same way it does on hardware. I'm saying that because most all of the simulators out there work as discrete systems. They're not tied to a physical entity. So now you can run this circuit and you can inject different remedies. So now let's think about a model that you can develop. You can think about the iteration and the placement of those injection pulses. And now start running multiple experiments, you know, thousands of iterations, where you're optimizing location shapes and components, where all of the time you're monitoring the fidelity of your circuit, the performance of the circuit, and when you're hitting the optimal point, you know, now that you've kind of closed the loop on this cycle from that point on, you can now start thinking about training a model that will find the next iteration when conditions changes to hit the same performance level and again and again and again. So it gives you the flexibility, it gives you the canvas to do all of this work. And obviously there is a huge cost benefit here. When you're working on real quantum hardware today, you know, if you go to the leading providers, you would probably pay a few thousands of dollars an hour to run a circuit. You can do it in a much more economic way on our platform. And when you're gaining your confidence, then obviously you go to the hardware, but you understand what you're doing and you understand the root cause for the things that work and do not work.
- [38:01]
B
And again, can you give
- [38:04]
A
sort of
- [38:04]
B
a before, I mean, you did earlier and the first time we spoke. I mean, we did, for example, the 80% to 99% on factoring 21 was Shor's algorithm. But can you give some other before and after results?
- [38:25]
A
Yeah, absolutely. One of the things that the industry is moving towards is logical qubits. It means when can we take those physical systems? You know, qubit is the analog of a transistor in classical devices, but the problem is they're very noisy. So you need an ensemble of them in order to do a logical computation with a low enough noise level that they can work at scale. Until not long ago, the only system that was able to show that, at least on superconducting qubits, was the Google system. Using our system, our digital twin, and our augmentation capability, we're able to take an ensemble of qubits from IBM and connect them in such a way that they now perform as logical qubits, meaning they can now go ahead and do meaningful computations as you scale the number of those systems. So that's one very interesting achievement that we were very proud of because we were the first to show it outside of those systems. The other one is thanks to our capabilities, we're able to push the boundaries of the performance of Rigetti hardware to a point where they can show 99.9% fidelity on their qubits, because we're able to optimize all the governing conditions of this system. And now obviously, it serves to a point where you can squeeze more performance out of this system. Another example is that we showed that we can significantly improve the performance of a given algorithm using our capabilities to a point where if before that, the noise level were just too high. Once you add our capabilities, the circuit is at the point where it can really perform and give you the results that you're expected to get.
- [40:53]
B
We talked last time about claims of quantum advantage, and those claims are routinely challenged by running the same algorithm. Classically, does a digital twin change that verification of claims of quantum advantage? I mean, can your platform be used as an independent arbiter when someone claims that they've reached quantum advantage? Or there are reasons that it can't be neutral. Your system,
- [41:37]
A
I, I think not only that, it can, it is used and we're using it internally to benchmark performance and different algorithms and different approaches on whether or not you're able to get to whether it's logical qubits or below threshold or quantum advantage. Now that being said, every system scale has its limitation, right? Let's take an approach, maybe from, or an example from classical devices. Let's say you want to develop the next GPU. GPUs today have tens of billions of transistors in them. The design tools by the leading companies, whether it's Synopsys or Cadence, as advanced as they are, they cannot simulate the whole GPU at once. You cannot run finite element analysis of 60 billion transistors at once and etc. So there is a finite limit to the number of the scale of the system that can be used. But can those simulators and those design tools predict the performance and tell you whether or not this design is suitable to hit a certain benchmark? Absolutely. So that's exactly the same approach here. We can benchmark and we can give you the tools to develop. We're not saying, and we will never say that we are replacing quantum computers. We are accelerating the development of quantum computers and accelerating the development of quantum algorithms and giving the industry the best in breed tools, combining digital twins in AI to do that. And one of the things that you can check along the way is whether or not a specific configuration and a specific approach can get you to a point of demonstrating quantum advantage.
- [43:34]
B
Is, is there a clear line between classical simulation of quantum hardware and what you actually need quantum, a quantum device to do? I mean, at, at some point, what you were referring to the, with the. Well, the classical simulation has its limits. Is there, is there a clear line or is it sort of as companies push simulation that that line keeps moving?
- [44:16]
A
Well, I have to say we keep on surprising ourselves with the scale. When we started this journey, we thought we will be capped at a few qubits. And then we've developed a new solver that pushes the boundaries. We can now simulate above 100 noisy qubits. The reason I'm specifying noisy is because those qubits take into account noise models and interactions, etc. And this number of 100 is very critical because it taps into quantum error correction, specifically Quantum error correction distance 7, which was claimed to be the cap where classical devices cannot simulate quantum hardware anymore. We think we can push the boundaries beyond that. But there is Definitely a limit. Whether the limit is 200 or 300 or at that range, it, you know, yet to be seen. But the important piece is how do you participate and provide those tools as the industry grows? And how do you use those tools furthermore to orchestrate things that quantum hardware just is not able to do? You know, I think the days of running quantum design and quantum development softwares on quantum computers is still very far off. It means that there is no escaping from the fact that classical devices will govern and will push the develop of quantum computers in the foreseeable future. And that's the thesis behind everything we're doing. It's not a question of whether or not they will participate, is it's clear that classical devices are the tools to push quantum forward. The question that keeps us awake at night is how do we provide the industry with the best tools in order to do that in the most effective way?
- [46:27]
B
Yeah, is, I mean, you referred to it already, but is there a limit to, to, in terms of cubit count depth and noise complexity in, in what you can simulate or what you, in your digital twin?
- [46:46]
A
I, I think it's fair to say that with the current tools, we're probably looking at the hundreds of cubits, not more than that.
- [46:56]
B
Yeah, yeah. And you, you mentioned the modalities you're working on superconducting and trapped ions, or you mentioned when we spoke that you're working on photonics. What are the other modalities that this applies to? And, and how portable is a digital twin between different hardware families? I mean, can you apply lessons learned in one to another architecture?
- [47:32]
A
The underlying physical models are quite different. So if you want to do a decent enough job, you need to understand the physics very well and you need to be able to represent all the governing noise models in a way that you know your results will be accurate enough. So we have to be very selective on the modality that we're picking at the moment. Like you mentioned, superconducting qubits, ions. We're also looking into spin qubits because we have a lot of domain knowledge into that. This is still a modality that's kind of in its early days, but they're moving actually pretty quickly as you go up the stack, moving away from tools that are used by the hardware designers to tools that are used, for instance, dealing with errors. Then you can be a bit more abstract and then there are components that you can migrate from one modality to the other. So in that regards, it's pretty easy for us to provide quantum error Correction platforms for the different modalities in a much shorter time frame.
- [49:03]
B
Yeah, and who's your primary customer in this ecosystem? Are they the hardware vendors, the algorithm developers, or the end user verticals like pharma or finance?
- [49:21]
A
We are currently moving along the whole software stack. In quantum, everything that you're doing has to be hardware aware. Classical devices, you know, a zero is a zero and a one is a one. Once you figured out how to manipulate those transistors from that point on, you might have different flavors of your operating system, but the underlying physics is the same, the underlying operations are the same. In quantum systems, everything has to be hardware. Where, meaning the way you're calibrating and operating the system, what kind of pulses you're pushing, then how you run circuits, how do you deal with errors on the circuits, and then how do you deal with noise and error correction on the application layer? It all has to be hardware. Where. So in that regards, the users are users that are participating. Our users are users that are participating today in all of those efforts. Meaning we have some users that are literally designing their next generation machine. They want to understand how to optimize their qubits and their gates. There are users that are interested in eliminating errors or noise on the circuit. They're using the platform to execute circuits and reduce the noise. And there are users who are interested in error correction and developing new decoders. They're using the platform in order to do that. And very quickly I think you will start seeing end users using this platform in the combination of AI to be able to accelerate and improve the performance of their algorithms.
- [51:08]
B
And on the hardware side, I would guess that most hardware providers are using the platform. I mean, if you're showing this kind
- [51:23]
A
of
- [51:26]
B
optimization, why wouldn't they? But, but does that, how does that affect the dynamic, this sort of competitive dynamics between hardware vendors? I mean, if it becomes pretty clear, I would think, who, which modality or which hardware architecture is giving the best, the best performance, at least in the digital twin.
- [52:03]
A
That's a wonderful question. And I think it talks into the flexibility of what we can deliver and how we're doing that. So obviously one approach is to come in and say, okay, give us all of your secret sauce. You know, you've developed this system and you've poured billions of dollars into your, and now tell us everything you know. It might work for a handful of customers, but we come from the industry, we understand how sensitive it is and we appreciate that and we are grateful for our customers. Another approach could be, this is the general model that represents your system. You know, just because the physics is the physics, we're giving you all the knobs, you go ahead and you tweak this model for your system. You don't need to tell us how it works. We don't, we actually don't care. We will give you the best engine to now go and run this twin at scale for your development. So we're giving them the canvas and the mechanism and now they're going it and doing it in their own proprietary way and we're out of the picture. And this way we kind of have this firewall between their secret sauce and their know how and their IP and our capability. And I think it works very nicely.
- [53:33]
B
Yeah. In what you're building, if, if everyone optimizes their hardware and then optimizes the algorithms and then optimizes for specific use cases, at some point, have you guys optimized yourselves out of, out of demand? I mean,
- [54:02]
A
When, you know, we, the world luckily is always in pursuit of better and faster and bigger. So if this day will come, we'll probably have a follow up podcast on how does the life of startup executive looks after he's retired and won the big prize. But I think we have a lot of challenges. Until this day they will come.
- [54:29]
B
Yeah. And, and then, you know, I hate to ask timeline questions, but I always do because people always want to know what, what's your sense in that? You see this progress of, of when quantum hardware is really going to be useful in real life cases, not just tests.
- [55:01]
A
I think there are a few axises that are moving in parallel. That gives me confidence on the timeline being closer than we think the axises are. Noise level of the system, number of qubits, and the number of qubits that are needed in order to run a specific algorithm. In all of those axises, we're seeing an acceleration beyond what we expected even a couple years ago. So to put it in kind of more tangible terms, if we thought previously that we need 20 million qubits in order to run a specific algorithm with a specific depth and size, with a specific noise level from those qubits, the noise levels that we can tolerate today are higher. The number of qubits that we need is smaller, and the size of the algorithm that we need is smaller. Having said that, I don't think there will be a step function. I don't think we will wake up one day and Quantum will solve all the problems that it should solve. What I think we will see is that Quantum is adding benefit Sooner through some hybrid approach. It is taking a portion of the questions that it's very good at solving from classical systems, solving it, feeding it back to the classical systems and allowing us to accelerate in this way to a time horizon that I think is within a couple of years.
- [56:53]
B
And you're using digital twins and AI to optimize quantum systems. But there's a lot of hype about AI and quantum. I get pitched all the time on using quantum to run AI models. Again, it's very confusing to the layman. What claims about that combination do you see as genuinely promising and which ones make you roll your eyes and, and dismiss?
- [57:38]
A
Well, I'm an expert on how AI can accelerate quantum development. I have to put a caveat. I'm not an expert on quantum AI, meaning AI that is running on quantum systems. But I think we need to be very careful when we are extrapolating from our knowledge on classical devices into quantum devices. And I think there are still some proof points that we need to see before we can declare victory on AI running on quantum systems. Yeah.
- [58:27]
B
And then finally, what are the milestones that we should be watching for in the next three to five years to assess advances in the quantum trajectory?
- [58:42]
A
One obvious is the noise level. We gauge those performance metrics in terms of fidelity today. Best system fidelity. And we have to be very careful here. Usually people will report single qubit gate, which is a bit. It's not a bit. It's much easier than two qubit gates. Right. When we're, when people are talking about qubits and quantum system, they're talking about entanglements, this ability to connect qubits. So two qubit gate noise levels are still higher. So we'll see those numbers become better. So 99.9 and then another 9, and then another 9. So that's one metric. The other one is the qubit count. How many qubits do you have? Physical qubits. I'm not mistaken. IBM is targeting, I think, 10,000 qubits by the end of the 2030. So that's kind of the rough number. So hopefully they can track that number. That would be a huge achievement. And the third one is the translation of physical to logical qubits. Currently there are strong indications that we need a few tens of them, maybe 70 or 50. If we can get to a point where we need a few of them, maybe less than 10, it means that those 10,000 physical qubits can now be a few thousand logical qubits. That's a huge achievement. So I would say those through three axes, noise count and number of logical qubits would be the metric to watch and obviously to keep on watching the progress on real life applications and quantum advantage that hopefully will come pretty soon.
- [60:35]
B
Yeah. Okay. Well, this has been fascinating. Also clarifying, you're a good communicator because this stuff can quickly become too difficult for a layman to follow. Okay, so let's leave it there. I'm going to end the recording. Unless. Is there something I haven't asked that you want people to know about?
- [61:06]
A
No. I had a great time. Thank you for having me, Craig.