Bosonic quantum error correction with Julien Camirand Lemyre

The New Quantum Era, a podcast by Sebastian Hassinger.

And Kevin Rowney.

Welcome back to The New Quantum Era. This is Sebastian, and Kevin is with me for a change. Finally, thanks for rejoining. Kevin, good to have you back.

Good to be back, man.

The fans have been clamoring for your return. We've got a great episode today. We were made aware of a research paper, out of a group. It's actually a startup called Nord Quantique in Sherbrooke, Quebec, Canada. They're a spin out from the University of Sherbrooke, which is quite a good school for, quantum computing research.

Yeah. Interesting paper. We'll we'll post the link, for the for the team. But, this whole domain of of quantum error correction continues to be just thriving in so many areas. And I think, you know, fascinating both science and physics and math.

So let's let's dive in, but I think this could be a good one. Welcome back to the new Quantum Era. Hey, it's Kevin Roney. And today we have our guest named doctor Julien Camirand Lemyre. He's the cofounder and CEO of Nord Quantique. We'll be digging into the topic of quantum error correction and also about the missions of of North Quantic and other interesting market developments. Welcome. It's a real pleasure to be here, Kevin, Sebastian.

Hi. Yeah. Thank you.

Hey. So, we usually begin our podcast by asking our guests about their pathway through their arc of their career that led them to quantum computing. Not always a common path for a lot of people trained in physics, but, tell us about, your journey into this interesting domain.

So, I like to say that I was born and raised in quantum. So starting to be introduced to to the field. So I was doing, quantum physics, during, basically, my whole studies. But at the bachelor's degree, I was fortunate enough to to be working, with a lot of the quantum systems. At first, it was really on quantum material, trying to understand the exotic properties, the exotic magnetic properties, of some, of some materials that were were new to the field.

That's great. Thanks thanks for joining us, Julian. We really appreciate you, spending time with us. And so you were at the University of Sherbrooke. So note Pontic is is a spin out then from from that work that was started there?

Definitely. So, North Pontic is a spin off from, Institute Quantic. So Institute Quantic is a is a very famous, Quantum Institute that is affiliated with University of Sherbrooke. And, before NorthCon six started, so we spin it off from, a famous lab in superconnecting circuits and and device, which is the lab of Alexandre. But this is a theory group.

And so is it fair to Sorry. Sorry. Sorry. Sorry. Go ahead. No. It's fine. Go ahead.

Thanks. So is it is it fair to to say that, the North Quantic, one of its major market thesis, its its direction is the exploration of the applications of bosonic qubits. Is that a is that a fair summary?

Definitely. So the idea that we, we we follow so much since we, spun up from from there was really trying to use these systems. So these bosonic systems, to try to make qubits, get that can be, quantum error corrected in and by themselves. And the idea behind this is, of course, trying to use quantum error correction to build better qubits, but also to to make systems that scales better as you add more of and more of these cubits together. And this is really the purpose of, trying to to to build these cubits, but also make it in in a system that can scale afterward.

That's really cool. And so, Alexandre, you mentioned, leads the Institute Quantique at University of Sherbrooke. He's part of the the Yale gang, right, along with with, Jay Gambetta and Chad Righetti, and he's the that group that sort of started up, under the the supervision of Gurvin and Sholkov and Devore, and and really was the birthplace of the of the transmon, sort of the first really, really, sophisticated design for for super connecting qubits.

Right? Exactly. So Alexandre was there, during that time. So he was one of the co inventor of the transpods. You're right under that topic. Of course, he continued doing a lot of great works, here at Institute Quontig, and we're fortunate enough also to have a close collaboration on some of these bosonic systems that we are studying.

So this is I think an interesting area for us to maybe delve down a little bit lower level on the technology and the science. Because I mean, I think for a lot of our audience, they've heard of course of the Transmon system and you know of course, cubit architectures. So they're very familiar, I mean, almost intuitively with the fermionic notion, right, of the way cubits are. But I mean, you know, in your paper and so much of where Norton Quontic is going, you're exploring how trans bonds, right, the the core, fundamental architecture for so much of these advances, is instead set up to operate in a way that creates bosonic qubits. So could could we just get into some more of that, the detail for the benefit of our audience in terms of that that contrast?

And definitely, Kevin. So there's, 2 questions in your question. 1 is about, like, why Bosonic codes and why should we care about these? And the other one is about the architecture and how different or how similar it is to other architecture like, Google or IBM that are also pursuing, the Transmon approach. So I'll start with the first one.

So interesting. And, you know, I I guess one of the perspectives that I maybe is due to our audience is that only some of these quantum architectures can realize this sort of like dual use approach, right, of implementing both fermionic and bosonic systems. And so, it's a Transmon is one of those unique architectures. Kind of a kind of a compelling difference. Right?

So exactly. So and in our case, we don't use Transmon as as the core ingredient. So, if if you if you compare our architecture to to different ones, so more in super connecting circuits, for example. On superconducting circuits, you'll often have a bunch of resonator or strip lines that are are used to, send microwave photons to control transmons. This is how the architecture is usually built.

So the transmon is still responsible for the the two level system then?

The transmon in our case is is used, yes, as a two level system. So we have some ways that we control the the system, like in any super connected qubits where we access other levels. But the point here is just that we use this transmon just to to drive the cavity, and that that's the point of it. You can see this that using this transmon enables us to do some operation, that will be universal, means that we have access to all the setup gates Okay. That we need to control our, super connecting resonator or cavity.

But even for the for these bosonic cubits, I mean, it's it's multiple levels of energy. It's not just

Yeah.

Exactly. The standard 2. Yes. Right. Yes. Right. And and

that goes into effect.

I'm I'm just assuming that gives you the headroom, right, for a lot more redundancy that can be embedded for quantum error correction in in one Mhmm. Transmon. Yeah.

That that exact exactly the point. So we now leverage, like, these many levels that you are describing. We're able to leverage this to encode some bosonic states, that have quantum correction property. So we we can now, run a set of pulses on these on these, photons that will, provide us the means to do quantum error correction on that system.

Wow. Yeah. Just amazing. And I just I have the impression that you guys must be doing them some pretty detailed custom work right down at the hardware, right down with the firmware. You're not taking somebody's off the shelf QC in doing this. Or, I mean, don't don't you have to sort of build a lot of this ground up?

Yeah. So a lot of of what we did was actually some some of what we did was not existing, once once we started, so we had to Yeah. Work on it. We had to work art also and everything that is related to control. So how we control the the states, how we improve the control. And this is still a very important topic to to our hearts. So we're working hard on on trying to push the quantum error correction limits, as high as possible in these systems.

I was gonna ask about control. So is it still, like, room temperature, like, microwave control that you're sort of shooting down into the dill fridge and and manipulating the cubits. It's so that's still the same as as other, like, transmon or other fermionic based superconducting cubits?

Yes. Exactly. So the the platform that we are using is still a super connecting plat a super connecting qubit platform. We like it because it provides us a lot of control and also, like, super connecting circuits are fast to operate. So this is something that, to really really like about this.

And, like other, superconducting systems, do you have sort of the the the syndrome detection and then correction loop sort of, from, you know, sort of classical quantum loop that you're gonna need to to perform to to carry out the error correction?

So on the so let's say we refer to the paper that we that we just out. This is something that we actually didn't need. So the quantum error protocol, that we've implemented at Das Paper is autonomous in the sense that, we run a protocol, so we do send signal to the chip. But the results of this is a, autonomous quantum refraction protocol. So we don't need this feedback loop.

I mean, I think that's a great bridge to to your paper. I mean, that was, I think, one of the main subjects that we wanted to cover during this interview. I mean, there's some pretty interesting results. I mean, not just the autonomous error correction, but, I mean, it it feels like there's, a a rich, array here of of results explored by this this technique that it sounds like where Quantique is a is a pioneer on. Can you give us some of that just summary of the top line results that, really thrilled you from from this research?

Yeah. Definitely. I think one important point we just mentioned. So we've been able to implement an autonomous quantum error correction protocol on-site this this bosonic qubits. But what is more exciting for us is actually what the the results are.

So interesting. And and this is, this is excellent. Sounds like advancements in error control both on your bit flip and on on phase error. Right? It's like you're Mhmm. You're really demonstrating, significant advances both, both frontiers.

Yeah. Exactly. So the this 24% is actually average over, the different states, that that would, that would, compose that that part. So it it indeed includes both bit and phase flip, in that case. So this is an average between the the two rates.

Wow. And and again, done autonomously and on top of all of that, it sounds like you've got, some new results with respect to QEC parameters as well. You were able to, look at the operating parameters of these systems and make new advances there. Can you can you comment on that in more detail?

What are you exactly do you do?

The the the results around QEC parameters that was talked about in the paper. It it it sounds as if you've got, new results on, yeah, essentially fine tuning these systems to get them to, yeah.

Yeah. Exactly. I mean, so this is something we we this is something that, yes, we we work with. So, there's a lot of parameters to play with inside these quantum error correction, protocols. Started with the vanilla protocol, from theory doesn't always cut it, so we need to optimize.

Yeah. I mean I mean, Sebastian, it just feels like this is really a a standout approach that could produce, some pretty interesting commercial advances. I mean, I I don't know the landscape entirely of of the different competitors in the mix, but it it does seem as if there's there's real novelty

here. Yeah. The the autonomous nature is really exciting. I mean, the the idea of being able to detect and correct errors without that round trip to the the to the room temp and the classical computing in some cases, it that's really, potentially huge advantage. I mean, I'm curious, like, when you when you look at at other, schemes for producing logical cubits, there's often sort of a ratio.

Yes. I mean, so it's it's a bit different in the case of, of Poisson codes. Although when you're talking about the architecture, this is what matter in the end. So, I mean, you have these qubits, what we need to perform something interesting. The way we we like to think about this right now is that now with these qubits, we also have a no a knob, to turn, which is this quantum refraction of on single single devices, single units, where we can, reduce error rates on on these qubits before having to to concatenate it into one of the codes that that you've mentioned.

Okay.

For thinking about, like, concatenate thing in into a code. And there's also other opportunity. There's, also other codes that are, like, only applicable to those next systems that that we are looking into. And this is this is something that will be coming up in into our architecture as soon as

I see. I suspect that that would be at the another follow on, paper, perhaps with the follow on conversation as as Yeah. Reports arrive. Definitely.

That's cool. So so am I right in thinking that, like you said, you're you're trying to optimize the performance of single qubits. And in a sense, that's almost an error mitigation. I mean, it there's a there's a fuzzy line between error mitigation and error correction. There's there's some error correction going on in the the value of the qubit itself, but error correction in the broader sense, in the algorithmic sense, like a surface code or an LDPC, that's, as you said, concatenating together the results from multiple qubits.

When we think about bosonic codes, something that that we we like about these systems is is not only the fact that they can be error corrected, but this also calls in for the different opportunity when you think about the the full scale architecture. And you were talking about, these these other codes, like surface codes or or QLDBC codes or or or name your favorite code here. Bosynix codes also have the opportunity to be used as, elements within these, higher level code. So with this, we call concatenation. And what's interesting in that case is, before because we are doing quantum error correction on single of these these qubits, we can also have better knowledge of which qubit is faulty and which is not.

That's really

So, it's a it's a multi tier architecture you could almost do. And and it it opens up, you know, room for optimism that we could be sometime soon with results from your group approaching the quantum, the critical quantum error correction threshold where there would be sustainable operations for a logical qubit that could operate in a relatively error free amount of time for a long, long interval. I mean, it it it feels like there's, I don't know, a route towards some really huge achievements here.

Yeah. Exactly. This is this is what we are working towards.

So try to

There it is.

You make one of our refraction and and, yeah, that's the mission statement here. So, yeah, we're really, trying to push push this limit. But we take it in in to a different approach. So we're not trying to concatenate Qubes that we have at hand first. We're really trying to push down the error rates, on on on these Yes. Bosonic system.

Right.

And of course, as we speak, we're we're we're also working with the more complex architecture in the involving many of these, these physical units.

So have you focused primarily on, like, you know, t 1, t 2, and and, single qubit gate fidelities, or are you already looking at sort of 2 qubit gate fidelities as well?

All the of the above. So who we are working into all these directions. The numbers will be out there at some point. So this is something that is, for internal r and d right now. In the paper, what the focus is on, really quantum error correction benchmark.

Got it. And you mentioned the Transmon, being part of of the sort of the universal set of operations. Is that how does that compare you know, how do you implement a universal set of operations in a in a bosonic mode versus, the fermionic, the CO so pure pure fermionic operation with transplant?

I mean, for any quantum system, what you need is an Indiana universal set of gates, to to be able to do whatever you want. In in our case, let's say that the first task that we are up to is, encoding the system into these bosonic codes. And to do this, we we need a gate. We need 2 things. So we need a gate.

Interesting. That's cool. It's always fascinating to me how, you know, the the different sets of operations and, quantum unitary operations that need to be sort of, exploited to get to that universal set of operations.

Oh, they manifest. Yeah. And they are Yeah.

Exactly. Exactly. Yeah.

Pretty cool. Yeah.

Yeah. It is. It's exotic. You know, so what in terms of, you know, this research paper, what do you sort of see as your sort of next steps? Is it further exploitation of the of the quantum error correction or, or you're moving on to to other topics?

I'm both. I'm I'm both. Both. Of course. So, of course. Yes. Exactly. So, so what we showed in the paper is 24% increase. So, we need to do better, and we're working on it. Some of the aspects involve having better hardware.

So it's a

little bit foundational work on a, yeah, key question. So, I I don't wanna push too hard, but could you give us an approximate timeline where the next big wave of results are coming out? Because, you know, we're we're interested in in knowing, more about the future here. It seems seems very promising.

I mean, the summer let's say, the summer will be important for us. So we we are working with these systems right now. We are working toward achieving results, when it will be out, you you'll witness it at the same time as us. I guess. Okay.

Fair enough. As as as CEO should say

it. Yes. Exactly. Precisely.

Well, that's really exciting, Julian. We've we've really enjoyed, hearing more about your approach. I think, you know, if you can deliver some level of hardware, you know, autonomous air correction down at at the at the qubit level, that's obviously an incredible, boom to the field and and, exciting things for it to come from Norquatique. So thank you very much for joining us.

Thank you so much for your time.

Thanks for having me. It was a great pleasure to to discuss with you these important topics. It's our pleasure. Thanks.

Wow. That wasn't that cool, Sebastian? That's a feels like feels like breakthrough research. I mean, I don't I don't know the landscape entirely, but, the the novelty of this approach, right, where, you know, a a transmon circuit, right, is sloshing these Cooper pairs back and forth. It's this harmonic system. And, of course, everybody realizes that you can create a quasi particle in those systems that exactly represents.

Yeah. That's old facts.

A a cubit. That's that's the old news. Right? But but in this this whole approach, right, I mean, there are sort of you could use, in in some parts of this system, transmons as classic cubits and other parts as harmonic oscillators that create an entire bosonic, you know, a quasi particle with, like, you know, essentially infinite dimensional number of

Yeah.

Of energy levels. And that's and that's the pathway for redundancy they do for this quantum error correction. So I

get the the redundancy, but but, you know, just take us back to sort of basics for a second, Kevin. Like Yeah. So so a a typical transmon, your run of the mill transmon super boring. Yeah. It's from it's a fermionic system. Right?

Those Well, I mean, let's let's be careful because I mean, really,

I mean,

what what trans bonds are doing is, you know, there's this, it's Josephson junction. It's, essentially set up to in the superconducting mode. These not electrons, but Cooper pairs are jumping back and forth, and there's a harmonic, oscillation of the system. The the the standard way in firmware you can set up those oscillations is to make it so that the quasi particle that that is created from the en mass movement of these Cooper pairs, is, you know, essentially the same as isomorphic 2. Right?

And and an electron state of

Exactly. Exactly. Right. And so the the the spin state of those electrons is the the, theoretical framework that you've learned in Right. Quantum information science to sort on which quantum computing rests.

the And and photons are are bosonic. Right?

That's right. That's right.

It's it's the microwave photons in the system that they're treating as,

That's right. I mean, right. I mean and right. And they're, it's a tricky thing in this context because often, again, these these swarms of these Cooper pairs, these paired electrons inside a superconducting state are are are producing quasi particles. And so that we it it those those things are not really, subatomic particles or quasi particles, but they have, you know, isomer who

They exhibit the papers. Right. So

just like a subatomic particle. Yeah. Right.

The the weirdness of superconductivity

Isn't it now?

Never ceases to Yes. Yeah. I get the point. It's been we've been obsessed with it since 1911 or whatever.

Yeah. I mean, there's already a huge set of abstractions to to just to digest on computer architecture. Right? I mean, you know, like, down at the CPU and up to assembly and then set, you know, up to operating system. But here also, I mean, there's, like, numerous different layers of, you know, what what are we talking about in terms of a Yeah. A particle. Right? And Well, then you're making terms of the the architecture. Yeah.

And and putting it in terms of computer architecture, that's sort of my strongest, you know, resonance with this result is that, essentially, they're implementing error correction at the quantum hardware level, whereas previous approaches have required, a round trip for all error correction to the the room temp classical computing. It's one of the the critical sort of unanswered questions is can we implement, error correction without incurring so much overhead from the classical part of the error correction that you can't actually realize the, the performance advantage of the quantum Right.

Which which seems like a question which has now been positively answered by this really interesting result. So Yes. I I can't wait to see the next wave, of of their work, man.

It's gonna be cool. Yeah. Absolutely. No. I I mean, I we we I'm glad you sort of probed on timeline, and we didn't really get we got a very, as you said, sort of cautious appropriate response, but I'm dying to see, then get get a system into the market that people can start playing with. That would be really, really exciting.

Yeah. Such exciting times.

Yep. As always.

That's great.

Cool. Good stuff. Right? I appreciate this time. Yeah. It's fun times. Looking forward to the next, next one, Sebastian. Thanks for Likewise. Setting this up.

Thanks, Kevin.

Okay. That's it for this episode of The New Quantum Era, a podcast by Sebastian Hassinger and Kevin Roney. Our cool theme music was composed and played by Omar Costa Hamido. Production work is done by our wonderful team over at Podfi. If you are at all like us and enjoy this rich, deep, and interesting topic, please subscribe to our podcast on whichever platform you may stream from, And even consider, if you like what you've heard today, reviewing us on iTunes and or mentioning us on your preferred social media platforms.