Transcript | Designing Quantum Computing Building Blocks for the Future with Quantopticon

Building better quantum computers requires improving multiple elements of the machines. What if there was a quantum CAD tool that could help develop higher-fidelity qubits or effective quantum memory? We take a look at designing high-performance components with the future of quantum computing in this episode of The Post-Quantum World. I’m your host, Konstantinos Karagiannis. I lead Quantum Computing Services at Protiviti, where we’re helping companies prepare for the benefits and threats of this exploding field. I hope you’ll join each episode as we explore the technology and business impacts of this post-quantum era.

Our guest today is the CEO and cofounder of Quantopticon, Mirella Koleva. Welcome to the show.

Thank you, Konstantinos.

It’s great to have you here. We’re going to be doing a little bit of futurism today, because we talk about accelerating the development of quantum components. They touch multiple aspects of quantum information science. We’re going to take a look at some of the ripple effects this might have on the industry and dive in to what you do at the company. Do you want to give us a little background on how you got started in quantum?

Back in 2014, I was writing up my Ph.D. thesis in photonics at Imperial College London when I first learned about the latest research of my cofounder, Dr. Gaby Slavcheva, which was in a different domain of quantum optics and solid-state physics. She had developed a very rigorous, theoretical framework connecting classical electromagnetics with quantum optics, thus being able for the first time to accurately describe light-matter interactions on a nanoscale. Her work was just gathering momentum, as she had shown through her collaboration with experimentalists at Imperial College that her model’s predictions were indeed correct. It was a very exciting time and, at that juncture, it was dawning on both of us that this model could be instrumental in facilitating the engineering of next-generation quantum photonic devices. I wanted to be part of this exciting effort.

At the time, Gaby had implemented the model specifically for describing quantum systems with either four or six energy levels and was forced to have to rewrite all of the equations and recalculate all the input values every time a different quantum system was being considered. It was a monumental challenge that I set about to solve, but eventually, I wrote a piece of additional code that automated the whole process for quantum systems with an arbitrary number of energy levels. This is how we first penetrated the quantum field, and this is how the company got started.

Before we talk about the company, I was curious about what you just said: Were you trying to achieve some kind of better-defined qubits, or were you opening up the door to things like qutrits or anything like that?

The aim of this was to broaden the applications of our modeling software as much as possible so we can apply it to any problem.

OK. Tell us a little bit about the company.

We believe in a photonic future for quantum computing, and although there are several horses in the race, our expectations are that the photonic modality will prevail with the rest in the long term due to the numerous advantages of this type of system, including room-temperature operation, long decoherence times, compatibility with telecom infrastructure and harnessing the mutual semiconductor industry.

However, quantum photonic technology is still in its infancy, and there are many practical problems that need to be overcome before quantum photonic computers become a reality. The major challenges are developing quantum memories that can store the photonic qubit until it is needed and improving the performance with existing components.

Now, as an analogy of what we at Quantopticon are doing today, I always tell this story about how Intel designed the famous 8086 microprocessor chip in the ’70s, and of course, it went on to have a lasting legacy, giving rise to the x86 classical computing architecture. But it took the design engineers many months to do this using manual calculations, as the proper software tools did not exist yet. Photonic quantum computing is in a comparable situation today, as the appropriate computer-aided-design tools needed to guide the design process are mostly lacking.

Quantopticon is a software startup building a rapid design-optimization platform specifically for quantum photonic component systems and devices such as qubits, single-photon sources and quantum logic gates. We’re developing a first-of-its-kind quantum mechanical and electromagnetic computer-aided-design tool for quantum photonics which is a culmination of 20 years’ worth of academic research work. Our flagship product is called Quantillion.          

That brings us naturally to the very next question — I was going to ask about Quantillion and who you envision its intended users are in the space right now.

Quantillion is specializing in modeling solid-state quantum systems optionally embedded in optical cavity structures for greater quantum-state control and quantum light-emission enhancement. Quantillion computes complex light-matter dynamics occurring throughout the spatial extent of the device — and I should point out that unlike many quantum optics modeling tools out there, it can render microscopic extended devices. It can render very realistic simulations of the device. Examples of these extended devices are photonic structures on chip and the integrated architectures of quantum logic gates.     

So, it could, in theory, help you build the modules that would go into a quantum computer one day. Let’s say you wanted to have just a few qubits on one module, optimize and then build them out. Would that help you model if they’re efficiently being created that way — what the best number of qubits be? Would you be able to see that kind of granularity?

This is a little bit further into the future for us. At some point in the future, our aim is to be able to model our whole quantum photonic integrated circuits, but we are starting with baby steps with individual components, and then moving on to larger and larger systems.

For those tiny, little components, here’s a good time to start digging into future impact: What types of qubits would your software help you model and maybe optimize?

Quantillion can model material qubits, which are basically optical transitions in neutral systems, localized spins in quantum structures, gate-defined quantum dots, spin qubits in silicon and 2D materials as well — for example, transition metal dichalcogenides and hexagonal boron nitride. We can also model photonic-qubit rotation upon interactions with material-spin qubits.

So, it’s not a narrow audience here. Almost anyone in the industry could probably benefit from using this.

Yes. It’s designed to have very broad applications.

That’s terrific. Are there goals for how these qubits might change? Are you hoping that this tool will lead to improved fidelity — length of coherence times: T1, T2? Is it like that?

Absolutely. We want to address all of these characteristics and maximize indistinguishability, purity, brightness, overall transmission efficiency of single-photon sources, as well as the coherence times in fidelity of material in spin qubits.

Going back to the analogy you gave earlier, is this just about the calculations saving time, or is it also about saving a whole lot of time in the lab? Do you imagine that this would save rounds of revisions on silicon, like, “No, we’re not using that one?” “No, let’s try again. Back to the drawing board.” Are you hoping that in the end, you would get a little closer to something that’s like a beta board being ready to be tested?

Absolutely. We want to streamline the design-optimization process so that we can get to better manufacturability, to large-scale manufacturing and making quantum computers and quantum communication networks a mainstream technology.

So, I’m sure you’re keeping track of all the qubit fidelities that are being talked about out there. Are any of the leading companies or any of the leading research teams using this tool that you could talk about?

We don’t have any quantum computing startups using them, but we are collaborating with a few academic groups — with the University of Oxford and the Technical University of Munich — on a European Space Agency project, and we’re developing superlative single-photon sources for this particular project. The work that we’re carrying out is for satellite-borne quantum communications, actually.

It’s going to be exciting if you start to see this tool being mentioned in papers, because we have an abundance of them now: “I got this qubit to work as well.” It would be nice to see that the tool is used in that space.

Shifting to quantum memory, could you explain about the type of quantum memory that the tool helps develop and how that works, for our listeners to understand what it is?   

Quantum memories can be built on different types of architectures, and Quantillion can help to develop solid-state quantum memories that will be based on semiconductor quantum dots or color centers in nanodiamond impurities in silicon and 2D materials, which I already mentioned. The other kind of quantum memory that we can help to develop is silicon nitride waveguides on the chip doped with either erbium or ytterbium ions. This latter architecture is actually favored by BT, the U.K.’s biggest telecommunications provider.

I used to work at BT years ago. That’s funny — that seems to come up every once in a while. And, of course, ytterbium, that’s what’s used in trapped-ion today until they move to something else like barium-137. We got to talk about that on the show before. If you improve quantum memory, what kind of impact would that have on the industry if we’re just going a little forward in time here, extrapolating out?

Making robust and reliable quantum memories will find uses in memory registers in quantum computer processing units and will be, obviously, a big step toward photonic quantum computing or universal photonic quantum computing. Quantum memories will also enable building quantum repeaters that will significantly increase the reach of terrestrial quantum communications. This will grow quantum communication networks to practical sizes that will allow many more organizations and businesses to benefit from them. For example, I see some of the first clients that this will attract as financial institutions and healthcare providers.

You mentioned that with repeaters, that’s going to be part of the magic to quantum networking. We haven’t figured it out how to do that just yet. What impact do you think quantum networking will have on the industry as a whole if we can get it to work better?

Quantum memories can actually help in quantum networking through utilizing protocols of photon exchange between qubit nodes, and one can achieve distributed quantum entanglement and cross the states along the network for quantum computations worldwide — so, a quantum internet.

Yes, and there’s trickiness there. There’s the no-cloning principle, which makes it very difficult to copy a quantum state and move it on efficiently, so we still have to get over those hurdles. Otherwise, you’re limited by, let’s say, the length of a fiber. This is going back to my BT days, but, yes, you could send things like UKD — we can move photons around — but then it becomes an issue when you try to extend them on and keep the information quantum.

This software took about 15 years to bring to market — 15–20. Are there any other offerings you’re working on that might be on the horizon — closer than 15 to 20 years away?

As I mentioned, we’re working on enhancing our capabilities for modeling single-photon sources as part of the European Space Agency project, and we’re actually specifically adding to the model another model for nonclassical noise, which is very important for being able to faithfully model single-photon sources. A little bit further on the horizon is implementing a full description of three-dimensional photonic geometries with embedded quantum structures. Three-dimensional simulations are naturally very computation-intensive, especially since we render the structure in exquisite spatial temporal precision down to attosecond and angstrom level. We will want to speed up the code as well to bring the runtime within more practical time frames. Further plans are to make Quantillion accessible through the cloud and build out a vast utilization of the results as well as add on a module for extensive data analysis.

This is software you just download or run locally, then?

It’s stand-alone software.

If you go to the cloud, do you anticipate optimizing for GPUs or something like that?                       

That’s definitely something that we are considering. We are considering harnessing HPC as well — basically, every kind of technique we can get our hands on in order to improve the runtime and to streamline the code and to make it run faster.

Would it be a whole new service model — like a monthly fee or something that includes compute time plus license and all that, all bundled up?

We are still figuring these things out on the business side, but yes, we’re working toward a monthly subscription model or charging based on compute time.

That makes good sense. So, you’re working with companies very deep down in the hardware stack here with these projects you mentioned. Based on what you’re seeing behind the scenes and these development days, how do you see the next three to five years looking for our industry? Do you have any predictions, maybe even some that are counter to what other people are saying about the timelines that we’re expecting to see?

I have some expectations counter to some other experts in the field in that I think that photonic quantum computers will prevail. But to be more conventional in this question, we anticipate that the quantum photonic players will move to true deterministic single-photon sources in the near future. Techniques like spontaneous parametric down conversion and four-wave mixing as a source of single photons will be replaced with solid-state quantum emitters — color centers and diamond or semiconductor quantum dots.

Now, quantum dots, it’s still not as reproduceable and definitely not optimized still. However, we’re working on this by developing novel repeatable and scalable ways of generating single photons out of semiconductor quantum dots. We’re developing a model for emission at 930 nanometers, but once we prove this new methodology, we will move on to wavelengths in the telecom range, which is what PsiQuantum is interested in, for example, and the other photonic quantum players as well, as this is the wavelength that will enable interfacing with telecom infrastructure.

Yes, we’re all interested in PsiQuantum. We want to know what’s going to go in that big, black box that’s going to make a million qubits, as they promise. When you talk about shifting to this other photon sources, is it just to make it more room temperature, more affordable, etc., or are there going to be benefits to things like coherence and fidelity, or do you anticipate, because it’s going to be cheaper and easier to make them, that you can do better error correction because you could have so many more of them?

We’re actually planning on addressing all of these challenges, and we plan to optimize them in all the ways that I’ve described before. We want extremely high fidelity. We want long decoherence times. We want all of these characteristics to be optimized. We’re setting ourselves a big challenge, but we believe that with the capabilities of our software, we can actually achieve this. This is within our reach.

I’m a big fan of removing some refrigeration from the process. That’s the only way to get these machines and rooms easier and racks or things in the future, definitely. Has anything surprised you from these newer projects that are going on — some innovative results that change your thinking about something you should pivot to?

Well, obviously, I’m very intrigued by Xanadu’s recent announcement that the Borealis quantum chip took something like 36 microseconds to complete work that would take 9,000 years on the traditional computer. One has to always be cautious and scrutinize claims like this carefully, but at least I’m glad that photonic quantum computing is making headlines on the subject of quantum advantage.

Do you feel like quantum advantage should be that a machine like that can do something faster, like Borealis, or do you feel like it should only be quantum advantage if you could do something practical and useful fast — a known use case like optimization or something that people actually want to do, rather than an experiment? Because there’ve been other claims: In China, there’s a photonic claim that they did something fast. These keep coming up from time to time, and then they get attacked by other members of the industry. What’s your take? Do you think it’s like moving the goalposts to say it has to be useful? Should it be useful, or should it just be proof that it’s faster?

We have to have proof that each type of modality is producing results faster than a traditional computer. We all have to go through this. Overall, obviously, we’re all moving toward a goal of a universal quantum computer being able to carry out any computational task of any importance — hopefully, toward the benefit of humanity one day, to resolve some of the biggest challenges of humanity. It’s all a gradual process, and we’ll get there step by step.

Yes. Thanks so much for coming on. I enjoyed this, and I hope that your software can lead to some more efficient machines in the very near future. Thank you.

Thank you very much, Konstantinos. Thank you so much for having me on your program.

Now, it’s time for Coherence, the quantum executive summary, where I take a moment to highlight some of the business impacts we discussed today in case things got too nerdy at times. Let’s recap.

Mirella Koleva was working on quantum photonics in 2014 when she met the eventual cofounder of Quantopticon. Her soon-to-be partner was working on a framework that could describe light-matter interactions. The two realized this model could help create next-generation quantum photonic devices. Mirella developed code to automate the use of the system and broaden its application to different types of problems.

Quantopticon believes that photonic quantum computers will prevail as a leading type of architecture for reasons including room-temperature operation, long decoherence times and compatibility with telecom equipment.

The photonic systems have to overcome some engineering challenges to become a prevalent technology. One issue is effective memory that could store quantum photonic qubits until they’re addressed. Another is improving the performance of support components.

To help accelerate the solving of these challenges, Quantopticon created software that can help design qubits, single-photon sources and quantum logic gates. It’s called Quantillion, and you can think of it as quantum CAD. It can render very realistic simulations of devices such as photonic structures on a chip. In the future, they imagine being able to model complete quantum-photonic integrated circuits.

For now, modeling individual components is possible. The software can model material qubits, which are optical transitions in mutual systems: You may have heard of cold atoms, as an example. Quantillion also helps with localized spins in quantum structures, quantum dots and spin qubits in silicon. The software is designed to be useful to many types of systems.

A lot of criteria go into qubit quality. Quantillion can help with improving coherence times, purity, brightness, efficiency and fidelity. The hope is to improve the quality of components and speed up production.

Quantopticon is also using the tool to develop satellite-based quantum communications for the European Space Agency, showing its versatility.

Quantum memory is still in its infancy, and Quantillion can help develop solid-state and ion waveguide versions of memory. Improving quantum memory could help universal photonic quantum computing but also will enable quantum repeaters to grow quantum communication networks to usable sizes. Interconnect may benefit from these networks too, letting systems act as one large quantum computer from a distance.

Quantopticon hopes to have the Quantillion tool available in the cloud eventually, with even better visualization capabilities. For now, it’s a download. Check out the link in the show notes if you’re interested.

That does it for this episode. Thanks to Mirella Koleva for joining to discuss Quantopticon and their software, and thank you for listening. If you enjoyed the show, please subscribe to Protiviti’s The Post-Quantum World and leave a review to help others find us. Be sure to follow me on Twitter and Instagram @KonstantHacker. You’ll find links there to what we’re doing in Quantum Computing Services at Protiviti. You can also DM me questions or suggestions for what you’d like to hear on the show. For more information on our quantum services, check out Protiviti.com, or follow ProtivitiTech on Twitter and LinkedIn. Until next time, be kind, and stay quantum curious.