Episode 357 - Quantum Engineering with Jelena Vučković

Transcript

[Music] Welcome to Zero Knowledge. I'm your host, Anna Rose. In this podcast, we will be exploring the latest in zero knowledge research and the decentralized web, as well as new paradigms that promise to change the way we interact and transact online. [Music] This week, Germo and I speak with Yelina Vauovich, professor of electrical engineering at Stanford. She is also the lead at the nanocale and quantum photonics lab. In this episode, we dive into the field of quantum engineering.

Yelena provides a clear introduction to the various platforms used in quantum technologies, explaining how these systems are modeled and developed. We explore the different branches of quantum engineering from quantum communication all the way to quantum computing. Given Yelina's expertise in photonics, we dedicate part of the conversation to examining how light can be used as a platform in quantum systems and cover the unique advantages and challenges this approach presents when compared to other quantum platforms. This episode takes us in a slightly different direction from our usual topics, but as you'll hear, the implications for cryptography are significant. Our focus in this episode is on what quantum technologies can make possible in the realm of quantum cryptography, an area we hope to explore

further in future episodes. Now, before we kick off, I just want to let you know about ZKS 13, which is happening in Toronto on May 12th. ZK Summit is our bianual event where you will learn about the latest in ZK research, the newest applications, and find out who are the most important players in ZK today. It's also a wonderful way to get to know the larger ZK community. So, if you've never been to a ZK Summit, I hope you'll join us. To check out the speakers, program, more info about the event, and a link to join, please visit zksummit.com. I've added the link in the show notes. Now, Tanya will share a little bit about this week's [Music] sponsor. You might already know the ZK

jobs board, a place where teams can share open roles with our community. But when it comes to key hires, finding the right fit in our niche space can be difficult. That's where Missing Link comes in. They're a talent team built for the web 3 era, helping projects across the ecosystem connect with the right candidates at the right time. They've worked with names you'll recognize. Ethereum Foundation, Matters, Lido, Mina, Web3 Foundation, and many more, filling critical roles that drive these teams forward. Whether you're an established project looking to fill a senior leadership role or a startup searching for specialized talent to refine your product market fit, Missing Link can help. They've done it for some of the biggest teams in the space,

helping them and now you find the right people fast. For more details, check out the show notes and visit their website at missing-link.io. So, thanks again, Missing Link. And now here's our [Music] episode. Today I'm here with my co-host GMO. Hey GMO. Hey. And we are here today with our guest Yelena Vukovich. Welcome to the show Yelena. Thanks for inviting me. Hi Anna. Hi GMO. So in this episode we're going to be going in a bit of a different direction than what we usually talk about. We do often talk about circuits on this show, but today we're going to not be talking about circuits in the way we normally do. I'm very

excited to get a chance to dig in. GMO, you introduced me to Yelena. How do you know each other? Maybe you can give us a little bit of backstory. So, Yelena was the first person I met at Stanford and also I guess the last person to see me out. She was my PhD adviser. Very nice. Okay. So, Yolena, I think it would be great to understand your path into this. I sometimes ask this to cryptographers, but like what was the spark that got you interested in the topic that you worked on? Do you remember kind of the moment that started you on this path? Well, uh, believe it or not, Quantum, uh, has been around for a very long time. I know it's very popular now, but I was in grad school at Caltech, uh, between 96 and end of 2001. And somewhere in the late

'9s at Caltech, we started the first project kind of experimental project focused on building quantum networks for secure long-distance quantum communication because that's a easier problem to tackle than quantum computing. But this idea of quantum computing and some theoretical work on quantum computing was ongoing there at that time. So that's John Prescll who is maybe you heard about him. he actually also is very important person these days in popularizing quantum computing. So I was in grad school there uh and working on photonics and uh then few faculty uh at Caltech decided to launch this project on building quantum networks and I think maybe there were only a couple

projects worldwide in the US probably all of them uh back then focused on that. So I started working on that because we were trying to combine this photonic platform that I was working on with atoms that we would put inside of photonics and then we would build quantum network. So that was the idea and I was just excited about this whole this whole you know concept you know building something that nobody has ever built and then you guarantee that you know if someone is eavesdropping on you and so on. Uh so that was there. Of course the project was kind of tiny years ahead of time.

So, so we quite didn't get to build exactly we even today there are only like three node quantum networks you know worldwide and that's what we were trying to build back then but uh certainly it launched a lot of um exciting efforts and uh um beyond that you know five six years later like many more projects were launching in that space and uh excitement was building up over the next 20 years right so I that's really where it started kind of late 90s which is now a little bit less than 30 years ago, right? Wow. Um, what about the optics and photonic side of things?

Like was that something that came later or was that also something you were looking at like right at the start? Well, I didn't really work on photonics when I was in undergrad. So I I'm originally from Serbia or what was former Yugoslavia back then and uh I graduated in electrical engineering and was interested in physics and math but uh was kind of uh looking around into different topics for uh graduate work. I did some work on error correcting codes at the University of Sydney in Australia after my undergrad. And when I came to Caltech um I thought that that's what I would be doing. That was um a time when there was a lot of uh excitement regarding read Solomon codes which are

used in I think uh CD and DVD recorders and then um and now by the way used in zero knowledge proof as we talk about in this show. I have I have papers on read Solomon codes maybe I should read. We'll try to dig the sub up added to the show notes. Hey. Yes. And you know there was also some excitement about turbo codes back then. So I came to Caltech thinking that's what I would be doing right and but then I met my uh PhD advisor Axel Sharer you know I we had some qualifying exam for PhD and I did really well in physics although I was planning to do applied math I mean I passed it I did well in information theory too but then you know he invited me to offer me job in his lab and I didn't know I mean I

knew about electromagnetism but I'd never done in my life uh any photonics or experimental you know photonics Mhm. And he told me about this periodic dialectric media photonic crystals that that they were also they were only theoretically proposed at that time but uh his group was the first group trying to make them and that's those are structures on chip where you can control flow of light by drilling some holes in two dimensions or three dimensions. GMA knows all about this. He took my classes, did research on this. Uh, and you know, that looked like uh, wow, you know, I mean, he told me one thing that, uh, maybe was was important in making my decision. He said, oh, you know, you can

go anywhere, you know, go back home and, you know, work on math, but, you know, here you have labs where you can make things. Why don't you try to do experiments, right? And I'm like, okay, you know, that's uh, really good advice. Maybe I'll just spend three months uh trying to do experiments seeing how this whole nanop fabrication and photonics I was about to ask. Had you ever fed anything before? That was the first time you fed stuff. I assume yeah of course no that was the first time and you know then at the end of three months I mean I guess 25 30 years later I'm still doing that. I uh it was just uh something that I thought oh it will keep me busy. It's interesting at least for PhD and then who knows what I'll be doing. But by the time when I was graduating, I mean at

that at the time when I started grad school, we were the only group trying to make photonic crystals and at the time when I was graduating, everyone wanted to do it and then it was a good time to get a job and there was much more excitement both about photonics and quantum excitement is was building up. So I I kept doing it. Yeah, I'm going to do a lot of this I think in this episode, but little like lang uh word checks. GMA, you just said you fabbed or you fab something. Yeah. What does that mean? Like fabricate. So it's like building something when you say that. It's like creating. Yes. Nanop fabrication. Yeah. Exactly. Like you know for electronics or for GPUs, Nvidia designs the chips but then they fabricate them or at TSMC in Taiwan or

you know Apple does the same thing. Uh so we do that. We use commercial foundaries like you know global foundaries and so on for some of our silicon photonic chips but we also make them ourselves. The ones for quantum computing we make ourselves. Do you make them in Stanford? Yeah, at Stanford. Stanford. Stanford has Yeah, Stanford has uh one of the best uh university nanop fabrication facilities. Actually, it's the first university nanop fabrication facility ever started. Oh, I didn't realize it was a first. I guess it makes sense given Silicon Valley and all, but yeah, it was in 1980s. It was started by uh professors of electrical engineering because of course a lot of people doing chip manufacturing. Y um around the world in Taiwan in the US

actually came got education in Stafford electrical engineering. So it started by all of these companies donating I think tools to electrical engineering. So electrical engineering set it up but now it's a much bigger facility with many more tools and staff and our students and not just students also people from companies nearby come here to uh when they want to do some proof of concept fabrication of electronic or photonic circuits that's what they do at Stanford. I will say one of my biggest regrets is still not uh you know having done fab something something in the in the fab. I still can. Yes, I'm like right here. You know, I should just drive down. Hey, I'll do it anytime actually. Yeah. Yeah. Yeah. Yeah. We

always need people who want to do fabrication. I'll do it. I mean, I'll probably totally screw it up as I normally do, but like that would be fun. That'd be part of the the experience. It's fine. It's fine. It's okay. It takes some time to ramp up, but yeah, you're welcome. Yeah. So Yelina when we first or just before we started this interview I was sort of using the term quantum computer a lot but you you sort of hinted actually maybe we should think more quantum technologies. Can you kind of make that distinction for us? Yes. Yes. Uh so quantum technologies is a broader term and I prefer to talk about quantum technologies because quantum computing is just a subset of quantum technologies and uh there are other interesting and more near-term quantum technologies such as quantum sensing,

quantum communication or networking, quantum simulation, right? And uh all of these technologies often share the same devices or components except that you configure them or connect them in different ways for different applications and and some of them are much u um easier to scale up or or build, right? they're more accessible such as quantum sensing where it's enough to have essentially what would be one cubit in a quantum computer to already do some some interesting sensing or quantum communication um which is we'll be talking about it later on but it doesn't really require millions of these quantum bits in order to do

something that's already useful but it's still a very very useful thing because you can know when someone is eavesdropping on you so there are a lot of shared elements some platforms are maybe a little bit more scaled in the context of quantum computing like superc conducting quantum processors like the ones that Google is building but they're not used for commu for communication because they use microwaves right and you can't send microwaves over long distances and you would have to cenically cool wires and send microwaves that's of course not something that anybody would be doing uh for scaling but actually there is something like that at ETH in Zurich they built 40 meter cryogenically cooled superconducting wires to to connect two

fridges. But you know going forward people will have to do photonics to connect different uh even superconducting quantum processors and that's called quantum transducers and quantum networks right so there are a lot of connections and even you know if a platform is not specifically photonic platform there is a lot of photonics there and I like to talk about quantum technologies again because of all of these shared elements and also because the something that maybe many people don't realize is that really The main uh secret sauce behind all of these quantum technologies is what we call quantum entanglement. So just producing independently millions of quantum bits is not enough to do anything useful,

right? Because they need to be connected in this special quantum mechanical way, which is what we call quantum entanglement in order to be truly useful for all of these applications. But integment you know can vary between two quantum bits or few quantum bits to millions of quantum bits uh depending on what application you're targeting right so the underlying mechanisms are are the same and technologies are often often shared right that leads me to a question that I've had which is and I was thinking more in the quantum computer context but I think I could ask it just for quantum technologies when you look at one of these items or one of these kind of systems Are they often generalized or are they specific? Like

are they meant to do a single thing or are they almost like programmable on some level? Uh well that's that's an excellent question, right? So if you're looking into quantum simulators, they're kind of what we we would call classically like uh an old computers, right? In classical computing space. So they're made to do one specific thing, right? uh in some case some some simulators can be reconfigured but generally they're really kind of uh special purpose computers that are simulating one specific other system right and they're not exactly reconfigurable basically uh you know you can build photonic system or that simulates uh another kind of

what they call condensed matter physics system what is happening inside some material right but it's simulates well particular process like maybe just super conductivity or some something like that, right? Like the way to think about it is like uh you have like a modeling problem. Yes, exactly. You're interested in right and then and you know normally the way we think about simulations is like you go and like write a program that simulates that thing. Here you actually create a device which itself simulates the behavior. So instead of constructing like this you know it's not a general program it's like this like particular device you've constructed just to perform the simulation of this particular system that you interested in. Exactly. For example, you can build hydraulic model of a electronic device

transistor, right? And you know it's a completely different mechanism. There is water there and some other elements but it's basically modeling the same physics, right? The same thing here. It's the uh physics of the processes that are modeled on different platforms are the same, but you basically can can uh see what is happening inside of your material by building it out of these elements that are more controllable. But it's then kind of configured to model one specific process which is what uh we refer to as quantum simulation. Quantum computing um is by definition more reconfigurable in a sense that you have all of these gates that you can connect

in certain ways and and build circuits that would be used to perform different types of algorithms. The way you're doing also in classical computers, right? Because you can write your program and then run different things on that computer, right? And again it's not kind of completely true that quantum simulators cannot be reconfigured because there are something called reconfigurable quantum programmable quantum simulators right but in general I this this is this is how how the systems are are distinguished right so it's not like gate based system so you've mentioned entanglement I think twice now and yes is that the big distinction between classical computing and like how we think of binary 01 one

energy on off like is that the part where it re it works very very differently like maybe we should talk a little bit about what entanglement is yeah so if you just look at the cubit level right cubit versus bit can be in states zero or one cubit can be in superposition states of zero and one and that's what people often emphasize oh there is this superposition of zero and one and then kind of I put everything in a superp position and that's why it runs faster that's not true like for quantum computer you know I'm just saying there is a superposition but that's not the main ingredient the fact that you can put millions of independent cubits in superp position is not going to bring you anywhere you need to also build up

this entanglement sometimes the algorithm itself produces some entanglement by interactions that's sort of what happens in shor's algorithm and some other algorithms sometimes you produce all of this entanglement and you use it for sensing or for networking or for some sort of simulation or computing. But that entanglement is the main ingredient. That's what distinguishes it completely from from classical systems. So what is entanglement? Entanglement is uh we sometimes refer it to as non-class correlations. You know what correlations are? Things can be correlated. You know, you have a process where uh for example, I always produce two coins. I I put them always heads up or heads down, right? So

then the I do that and then you know you know that outcomes are always correlated because I put them that way but quantum correlations or entanglement is different and that's also you may have heard about Einstein calling quantum mechanics spooky action uh at the distance uh there there were papers and uh arguments between Einstein Paulski Rosen what was I didn't hear what you said there though there what did Einstein say Einstein was referring to uh quantum entanglement as spooky action at a distance. Spooky action at a distance. Yes. Okay. He basically used quantum mechanical entanglement as the argument that something must be wrong with quantum mechanics theory. Right. It was it is so counterintuitive that it

was confusing even the best physicists of the time. Right. um and many physicists afterwards and of course later on you know it was of course proven that quantum entanglement is real and it's very different from uh classical correlations with Einstein was what Einstein was referring as some hidden variables and so on so so you know I haven't explained what it is I'll I'll tell you now right I'm just saying it is very counterintuitive because we don't have anything like that in classical uh systems but it is the main ingredient for all quantum technologies and okay this is why it's also very hard to scale them up because quantum entanglement generally is very fragile.

Uh when quantum mechanical objects interact with their environment they lose this you know entanglement quantum mechanical properties. Uh and that's why this is exactly hard to build and scale up right be uh for for making even more practical technologies. So quantum entanglement again these are correlations nonclassical correlations quantum correlations that only exist in quantum mechanical world and it's between what it's between like two different between objects between the objects between two or millions of different objects right and they guarantee that if you are measuring the state of one object you would know automatically what the state of the other object would be no matter where it

is. Interesting. And is it always the same or could it be like a different state? It could be different depending on the outcome of your measurement. You're setting up something that may be thousands of miles away from you in a particular state that you would know even without looking at it. And that's really, you know, kind of already gives you a hint of how this potentially could be useful for cryptography, right? or quantity distribution because if you have if you're sending distributing these entangled states between different parties and you know let's say between you and GMO and me and we're not at the same place but we all hold some objects that are entangled to each other and I

perform measurement and I measure zero for example I know that right away that you and GMO would also have zeros right okay even if you don't call me up and tell me that I know that you would have it right and if I measure one I know that you would have once as well. Of course you know someone may ease drop and mess that up and I would also you know from these algorithms know that someone has eavesdropped because there are some parts of these algorithm that assure me that no one has eaves dropped or someone has eavesdropped. So we can repeat the process until we are sure that eavesdropping is not happening. But again I think that's an example how integment can be useful for cryptography but then it similarly it could be useful for sensing and computing and so on.

Yeah, maybe here's an analog for the So, the the eavesdropping part is not does no classical analog, but Entanglement has a little bit of a very dumb classical analog and it's like imagine I just like pull two cards from a deck like the ace and the king, right? And I like shuffle them, right? And then I give one to Yelana and one to Yana. The second that you turn one of the cards around, you immediately know what the other person has because there's only two possible cards, right? Um so like in some sense this is like that's the like classical analog of it. What is crazy and what is like deeply annoying is actually um the quantum case also has distinguished by this fact that Yelina

was mentioning which is you can know that somebody is eavesdropping because the act of measuring itself will change the statistics that you get nurse cat yeah like exactly like that exactly like it's crazy right like you can literally find out someone's eavesdropping because it is impossible to measure the system without destroying it in some way. Wow. And that's the like real big change here too. Crazy. There's other kinds of correlations that we're not specifying here of course. So yeah. So we're you know distributing this entangled particles between three of us and you know we're doing our measurements and we're getting you know and we're not comparing anything but to find at the

end of the day let's say we distributed millions of these particles between us and usually that's done using photons because they can travel longer distances. Mhm. Then we call each other each other and we take some subset of those bits that we distributed and measured and you know well first we have to compare how we measure them and so on. But it's okay we're not revealing information there. I'm not going to go into details of that but just on the dropping we can kind of sacrifice some subset of them. Let's say you know hundred of these bits that we measure and we compare them right and assuming we measure them the same way because in all these protocols maybe kind of how you are measuring that that's part of the protocol but assuming we kind of set up

our measurement equipment in the same way quantum mechanics guarantees that we should be seeing the same bits right we should all be seeing like zeros or ones and so on if we're not seeing the same bits then we know you know that someone Eve droppy basically by the way extremely analogous to the way ZK proofs work. Okay, right like specifically think about it this way when somebody goes and measures like the let's say somebody is eavesdropping between the two of you. I'm eaves dropping between the two of you sorry um and when I do these measurements what like the state of these individual bits that have been transferred essentially you can think about it as they become uniformly

random. So when you pick a random subset of a 100 of these cubits and then compare them against each other, right, the probability that I get lucky and both of those are exactly identical is 2 to the 100 for the same reason that like the ZK proof is has the same thing where you do a test repeatedly until you're sure that the statistics like work out, right? Probability. Exactly the same idea. Okay. Exactly the same idea. Exactly. I didn't know. I mean, I'm learning here about ZK proofs, right? And I'm glad to have you as a translator here.

You speak both languages. It's so cool. Ah, I know. I'm like a I'm a secret like underground physicist. I'm not really a physicist. I was mostly I mostly did math. Uh, but I still like to pretend, so it's fun. Well, you were still exposed to enough physics. Yes, that's true. I have another question about a word that you've been using which I feel like maybe I should have asked earlier, but even like cubit. What is a cubit? Mhm. Yes. So cubit or short for quantum mechanical bit, right? So that's this kind of elementary object that we use in quantum technologies or quantum computing. So in classical computing, you know, there are streams of zeros and ones. So bit is zero or one, right? And switch can be placed in the state zero

or one. You know, transistors can be zeros or ones. That's how we do all the classical information processing. And even before that there were mechan you can do it with mechanical switches with anything that has two states right there were other platforms that people were using for computing like uh vacuum tubes switches and so on mechanical switches uh in quantum computing this um elementary um you know piece of any quantum technologies that we are using is a quantum bit and quantum bit can be in a superp position of zero and one right that can be mapped to different platforms like In classical computing, you can map bits to different platforms from mechanical switches to electrical

switches. We can map quantum bits to different platforms. Maybe easiest one to uh grasp is light, right? Because if you map uh zero to presence of one particle of light which we call photon in let's say one channel, right, which is some fiber, right? or or guide for light something like that and you map one to presence of a photon in this other guide then you can also put that same photon into a superp position of being in one or the other guide right and uh that's called quantum bit the other the other way of mapping it would

be uh photon what we call polarized you know we all know people wear polarization glasses right there may be different polarization of light which is essentially how electric field and and light is electromagnetic field oscillates. So you can put zero in what we call horizontal polarization. You can put one in vertical polarization. But then of course your photon can be polarized anywhere between zero like horizontal and vertical and that would be a superposition state right and those are all possible ways of mapping cubits to physical implementation. You know they could also be microwave current. It could be also state of some you know device electrical device. I mean there

are many different platforms that people are pursuing. My group for example is also pursuing states of spins right of electrons inside of some solids and uh you know particle can be in different spin spin states spin you know which we map to zeros and ones or superposition of spin states. So again these are just very different physical implementations right it could be presence of atom somewhere absence of atom somewhere many many different things and that's what you may hear about in the in the media as all of these different approaches to quantum technologies someone is talking about superc conducting circuits photonics spins ions atoms there are many many different flavors but those

are just different systems into which uh they can map right interesting and some of these some of these almost sound like platforms or techniques or something that's happening and some of these sound like objects like I know they're not exactly but like photon or atom like so I photons are objects okay yeah so I mean that's a again going deeper into quantum mechanics I know it's a little bit counterintuitive because we think about light as wave right but then photons are indivisible particles of light and the and the basic principle of quantum mechanics is vape particle duality that everything is at the same time both wave and a particle actually we think about atoms as objects

particles but atoms also since they are quantum mechanical I mean they're also waves and you know even bigger objects can be both waves and particles but of course the bigger they get uh the easier for them is to interfere with the environment so you kind of lose those uh lose the effects of maybe their wave properties right interesting Uh yes, but that maybe you know thinking about them as both waves and particles is very very important to grasp that superposition because probably for you it was not confusing when I was talking about polarization of a photon being in superp position of horizontal and vertical because that just means rotating polarizers in your glasses in a certain way. But then it's probably very

counterintuitive for you that an electron can be simultaneously in two different states, right? How can it be? Because it's an object. Can a tennis ball be simultaneously in two different states? Can a cat be dead and alive and so on, right? I mean, but when you think about objects as waves, then it's not counterintuitive anymore. Yeah. Exactly. I'll just shoot two things early on, right? There's like the cubit is like much the same way as a bit is an idealized model of like some let's think about it as like some like basic like object that we like to study. A bit isn't a real thing, right? Like like a bit has to be created by something. In

the case of you know RAM, it is like generated by like the existence of like a literal like atom in a cavity or a literal electron in a place and not right. In the case of you know NAND memory you like created via these latches but like a bit right is like like a like an idealized thing that we say like cool a bit like exists. A bit has to exist in the real world for us to be able to use it. Like we have to have like bits are in memory but we could create them in all sorts of ways is kind of but exactly but there's many physical manifestations of that. The same thing is true of a cubit here which is what Yalina is getting at right like yes uh not only that like there's also idealizations not just of like you know

systems but there's also idealizations of wires here too like how do you communicate different states across different bits right like in the standard sense you can think of like you can create a computer with water like you can actually there's like this cool sick Soviet machine at some point that was created in like the 60s that like somebody essentially created a computer with water that could add and subtract all sorts of stuff and in that case like the wire is a literal pipe carrying literal water right so that's the way we transfer bits from one place to the other and similarly you can think of and there's many ways of transferring kind of states or or information between two you know cubits right and setting up all sorts of like crazy things number one is you could have a fiber optic cable which like transfers light uh number two is

you could have a superconducting wire that transfers like certain microwaves so it's just like taking the it's it's thinking about two levels of abstraction ction which is the bit or the cubit as the like this is like a thing that like is completely an idealized model but is a very useful one because it can break down a lot of steps uh and then mapping it down to the reality which is like what yall does which is like how do you okay cool like yeah very cool like nice mathematical model you have here now make it real you know and so I think that's where really like where things get crazy exactly and then you know speaking a little bit uh uh beyond that you know you can kind of map these cub cubits to different platforms, photons or uh particles like spins of particles or currents and so on. Uh but then to

really get that main engine behind all of these quantum technologies which is quantum mechanical entanglement, it turns out you need to interact them somehow, right? And there sometimes processes that happen when you're generating them. You can produce entangled photons using certain processes and they're entangled right away. Sometimes you produce kind of individual photons first using some processes and then you can interact them somehow interfere them and build entanglement. But that part is really what is the hardest part you know producing millions of independent cubits is easy but then making it in millions of entangled cubits is hard right and that's that's really why the progress is

so slow. I'm so happy that listing of those techniques. I don't know what you call these but like or or forms or like I liked that list. What it made me think though is when you saw that like those announcements from Google like what are they doing like what what is the exact technique that they're playing with like what is the exact like object that they're using? Yes. So Google is primarily pursuing superconducting platform right. So these are superconducting systems. So as JMO was saying there are these superconducting wires and some superconducting elements uh gates uh that they are using that can be in different states right which are in some way kind of similar in physics

to the platforms that uh people are using with atoms or people I in my group are using with some you know uh impurities in semiconductors uh because these are objects that can be viewed as um quantum mechanical oscillators that can have different states, right? So, it's really um like that, but they're physically very different because they're made lithographically and they're made out of superconductors and the the initial platform was these Josephson junctions when you where basically you can produce some oscillations by by building superconducting systems, but then they've evolved into different types of superconducting cubits and that's really

their cubit, right? these superconducting elements. What is the material exactly? People are using different types of superconductors for this. That's why they I mean they're different kinds and different systems. I'm not 100% sure what Google is using. Maybe nobbium. I mean that's the one that many people are using but that's uh yeah so superconductors are materials that uh uh basically don't show resistance don't when you flow current inside of them. Uh but that generally happens at very low temperatures which is why when you see Google announcements you see these chandeliers which are open and chip right that chandelier is in inside of the dilution refrigerator. We

use delution refrigerators to cool systems to very low temperatures close to absolute zero. And depending on which material is used, the experiments have to be done at low temperature where the system is superconducting for example, right? Because at higher temperature it wouldn't be superconducting. Uh but you know all the other systems are also cooled in different ways, right? Even all photonic systems are cooled because they use superconducting detectors which also have to operate at very low temperature or systems based on impurities in diamond and silicon carbide that my group is using also have to be cooled not to millich Kelvin but few kelvin temperature so pretty close to to uh absolute zero. And atomic

physics systems are also cooled but they're not cooled in fridges. They're cooled with other lasers because you you have to cool them in order to suppress basically their interaction with the environment and preserve quantum mechanical properties that are very fragile. Yeah. Yeah. It's it's insane. Like yes, you can cool things with lasers. Like we've just like, you know, this is going to sound like magic to pretty much everybody on the show, which is very fun to me. But really quickly, I think it would be also very useful um to describe actually like just at a very high level like you know you say like oh we have these imperfections in like a material and we can use those as what do you mean like what does it mean to like like just from just a high level

perspective like what does it mean like how when you have to build this thing you know you go into the fab like I give you like uh you know a slab of some material so the the imperfections that I'm talking about that we are using um in my group to build these quantum system systems are what we call impurities uh some people call them defects they're intentional defects in diamond and in silicon carbide those are two materials that we use and these are leading platforms for quantum sensors like magnetometers there are also companies developing that now and also uh they've been used in state-of-the-art quantum networks and we didn't talk about this but you know quantum communication uses photons if you want

to do this secure quantum communication over distances beyond maybe 5000 kilometers where you start incurring losses in fibers then you have to build something called quantum repeater. You have to distribute quantum entanglement to larger distances using something that lasts longer than photons and that's where you use these impurities in these materials. So what is it right? Oh, like there is a think about it. Like you put on your bunny suit, you like open the door, you like come in with a thing. Like what do you do? Yeah. So, let me explain you what it is, right? So, what color is diamond, right? That's one of the materials that we use. Clear. Clear.

Exactly. But you can buy diamond jewelry in different colors, right? Like uh purple, red, and so on. Where is that coming from? It's still diamond. It comes from defects, impurities. Because people load, you know what is ruby? Ruby. Ruby is sapphire. Sapphire is clear, but ruby is sapphire which is filled up with chromium inside of it. And the color comes from these impurities. Okay. So we work with impurities in diamond. We use thin vacancy um in diamond and we also use silicon vacancy in silicon carbide which is basically another material used in in jewelry fake diamond jewelry because it's much cheaper than diamond it's also

used in power electronics electric cars and so on but when you see these materials like diamond you know you said Anna it's clear right but when you put impurities it can change color because these impurities can generate light at specific wavelengths Yeah, it has specific colors, but they also behave like atoms, individual atoms that are trapped inside of this material. So, we don't need to trap them in vacuum as people are doing in atomic physics and in some approaches to quantum computing. We can actually still control them with microwaves or with light and use them as cubits, right? Interesting. So when you look at our chips, they still look clear because there are cubits there, but

they're in much lower concentrations than concentrations of chromium in ruby or, you know, other impurities, chromium in um or nitrogen in like purple diamond, right? You can buy purple diamond which is filled up loaded with nitrogen while the environment is carbon. No, our chips still look clear but they have cubits. So these are intentionally inserted impurities at certain places and you can do that intentionally. You can control position. We collaborate with Sandia National Labs or putting them where we want and then we can also take these chips and then make uh structures inside of it that would guide light that would control

these cubids or guide microwaves. So we build, we go into the lab, we take the trip, we use exactly the same tail tools that people would use to make microprocessors, right? I mean the same machines doing some lithography, drawing the pattern, bombarding with some plasma ions to edge, to drill holes, putting into some chemicals to add more. You know when you see people in bunny suits and that kind of makes this um makes these structures that would actually guide light or microwaves which produce interactions between cubits.

Interesting. In doing research for this I saw a few graphics of this I believe. Um but maybe for the listener it would be useful to describe what that actually looks like. What does the graphic show? What are you doing? Looks like abstract painting. Yeah, probably right. So if you think about this so maybe GMA maybe the easiest uh uh way to start GMA was talking about uh micrfluidic models of u computers right people people can make make them right there is a micrfluidic transistor even model so if you're thinking about water right you can think about water right you know river and then it can branch into two different branches and half of water goes one way

half of water goes the other way they can recombine But you are using basically you you are guiding water down the river flow right and it can go in different branches split it out and so on with light or with electricity you know we can guide kind of electrons current using wires right if you open any device you need see a lot of wires those wires are guiding current right they're guiding it from uh plants you know whatever power plants you have all the way to your home to your device and so on and they're basically guiding current to all of the different devices that you have in all of your appliances in order to control some some of these switches and transistors that are acting

there as bits, right? Like in even inside of these computers, uh you can do the same thing for light, right? But you need to guide it. Yeah. So one thing that people are familiar with are optical fibers that are guiding light over long distances which are just very very long threads of glass, right? But if you would really like to do some processing on the chip, right, which is like piece of silicon or diamond or silicon carbide, then you need to produce these flows for lights that are bringing light in different parts of the chip and making it interact with these cubits and collecting light and processing it. So you can just think of it as kind of carving carving this

material in such a way that you can flow light, right? which you can kind of see it by analogy with water because it's waves. But are you are you activating the cubits with the light or are you like Yeah. Yeah. Exactly. We're activating we're controlling cubits exactly with light and electromagnetic waves. Right. Because for us cubits uh specifically in the platform that I'm using cubits are these kind of states of spins of some particles that are next to the impurities. I told you you you really want to build entanglement, right? Then you can build the entanglement only if you make them interact. So we make them interact by guiding light and and making them interact by kind of all being uh reached out by the same kind of flow of light.

Right. Information brought around I'm thinking not here but is is there any information that's moving with the photon or is it just like turning because photons? Yes. Yes. Exactly. Okay, information is moving with photons, right? And uh photons are some you may have heard that people are sometimes talking about stationary particular in the context of quantum communication, quantum networking, stationary and flying cubits. Photons themselves are also cubits, right? Because I was telling you how you can put them in superposition of different states. Uh and photons are are wonderful because they can travel very long distances. The problem with photons is they don't interact strongly, right? I mean, lightsaber, right? You know, we

know like when you hit photons from two lightsabers, they're kind of looks like they're interacting, right? But the in reality, if you just take two laser pointers, they would just pass through each other, right? Lightsabers don't don't exist. I mean, we know photons interact, but they interact very weakly. They don't unfortunately, they don't interact as strongly as they do in lightsabers. I wish they do, right? And that means that uh first they don't interact very strongly and second they kind of easily leak out into the environment. They're not very long lived. So we still use them as these flying cubits to make other things interact to transmit information between different processors like superconducting processors or uh different uh spin cubits like what we're

using or different atoms. But we use something more long lived as a cubit itself like memory so that we have more time to do all the different operations. Is that making sense a little bit? So like you sort of talked about the different I want to say like types the spins the microwave but do those in do those different states talk to each other or are you sometimes using all of those at the same time or are you usually focused on one and they're only talking to those in the similar context? Yes. So sometimes you are using multiple platforms for multiple things. I mean there are again different approaches there also approaches that are based only on photons but considering that photons don't kind of preserve quantum

mechanical states for very long. Generally those approaches are based on the fact that you try to produce some larger entangle state and then you repeat that many times right it's like happening decohering very fast but then maybe you try to do everything very fast and then you can kind of map it to certain certain types of algorithms I mean it hasn't been scaled that that well yet but that's kind of the underlying principle there and you need to suppress all the inefficiencies in your system as as well as you can platforms that I'm working on I mean most of the time is just this impurities that are interconnected with photons and then microwaves and that's what is there but then Google works on superconducting

processors but in order to connect different superconducting processors they have to somehow convert their microwaves and that's a very different part of electromagnetic spectrum that's what you have what your phone uses or what you have in your microwave oven very different from light But again electromagnetic wave that has to map efficiently to optical waves that you can transfer over long distances and that's what we call quantum transducer and to produce quantum transducer you would use some of the photonic technologies that we use and then connect two of their fridges efficiently. So that system would have superconducting platform plus photonics plus some of these impurities in diamond

or silicon carbide or something else that can also act as a transducer. Right. So so there are a lot of things that are put together at the same time. Interesting. In the work that you do, are you primarily modeling this and looking at what it would do or are you actually physically building this thing? Like you mentioned the fab. So it sounds like you're actually building it, but like building building because this has sort of been my question about like the like quantum generally like the field like how much of it is people using materials and and like doing the thing that they say they're doing and how much of it is like a theoretical like this is what would happen if Oh yeah yeah yeah definitely both are happening and both

are super important because uh there is a lot of effort in trying to find uh other you know quantum algorithms that would often o also offer exponential advantage like shor's algorithm which is very fast factoring algorithm would offer right so that's really one of the burning questions in the field how do you map this efficiently to other kind of broadly useful problems not just modeling physical systems not just doing I mean physical discovery in physics is important uh but I don't know model modeling black hole may not be so important for broad community you know maybe if you can do some very fast optimization of markets or something

that's much more more important and clearly shor's algorithm is important right because if someone can break all modern cryptography then that with a quantum computer of course we care about it right and that's that's an important problem so there is a lot of work on quantum information theory on the theoretical basis of this and it's a very important work there is also a lot of important work on verification of quantum computers because you know if you build a machine that is doing something that no classical machine can can do how do you even know that it's doing something correctly and that's a very important question there is a big community working on the verification problems and you know we're already even in the systems that my group is working

with where we have about 10 cubits in a resonator uh that are simulating something we're already running into problems that you know we can scale this up it's already very hard to classically simulate So we collaborate with a group in Munich that is doing this on supercomputers simulating what this tiny tiny thing is doing in the lab. Right. Interesting. So those are very very uh important theoretical questions. Um I think I just want to say we did an interview I think two years ago with a researcher called Orsat and I think this is I think he's working on quantum information theory because a lot of what he was talking about was like theoretically what would happen and it was cryptography as well. Um, so I'm going to just add this in the show notes. It is a slightly relevant episode

to listen to as well. Um, but yeah, sorry, keep going. So your particular So my work is more on the practical side of things. How do you build this? And especially, you know, I'm interested in the technologies that would enable building this in a very scalable way in the future. You know, you can think about it like computers initially, you know, starting with vacuum tubes and you know, and then eventually switching into this integrated first like discrete transistors getting discovery from Bellabs and then going into integrated chip scale platform. So how do we build something that is truly you know scalable but still offers all of these quantum mechanical um properties entanglement and so on and something

that we can build on a cheap scale and use light and electricity microwaves to control quantum entanglement and do quantum computing potentially and quantum simulation in addition to quantum sensing and networking where these platforms are already used. They are state-of-the-art, right? So people already use these exactly these impurities in diamond as leading platforms for sensing and and networking. So that's that's what I'm interested in and uh we do make things and we measure them in the lab. Uh design is very important part of this because I said there is a lot of photonics but this is a different kind of photonics. Photonics that we can use in optical communications can be lossy.

It's okay. Even if you lose 50% of photons, you just crack up your laser and you send your pulses and you can still stream your Netflix videos at home, it's fine. But, you know, in quantum applications, we can't afford so much losses. So, we have to design things differently and we sometimes need new types of components, right? That would simply just produce this better interaction for for cubits. And uh this inverse design that we were mentioning earlier on was born out of that need to do things in a much better way than we are normally doing it in photonics. And of course now people are very interested in it also in photonics. But again the the need the pressing need to do designs

in a better way uh really came out of the quantum research that we've been doing. When you say better like what are you optimizing for with this? We're opi often optimizing for footprint for efficiency. I mentioned you know in traditional optical uh communication systems. If you go to a foundry that makes chips for photonics uh they would have something called coupler that would use to couple your fiber optical fiber that can bring optical signals long distance onto your optical chip that would do some sort of processing there.

uh maybe communicating to information between different electronic pieces on on your chip, right? And you know in classical communication systems it's okay that coupling element loses 50% of the photons, right? Uh but when we want to design things, we want that to basically not lose photons because information is encoded even in state of individual photons sometimes, right? So we are trying to improve efficiency bring it as close to 100% as possible for couplers. Uh we don't want to lose any photons when we are doing some sort of you know routing of photons down the chip between different cubits because again information about some some

stationary cubit is encoding in state of that photon and if you lose it then we have to repeat everything then we can't scale scale things up or do whatever processing we're doing. Are you also trying to make the like in terms of the actual hardware like the chip design, are you trying to make that something accessible? I'm just sort of curious like how close to industry is this like are you trying to make something cheap and usable or is it still like kind of in the realm of we're you know we're trying to make this work. Yeah, usable. Definitely cheap. I don't think that anybody worries about that. We're not Maybe we're not there yet then. Maybe we're not on the cheap side. Nobody is there yet. Cheap. This is an expensive hobby

that but everyone it requires a small nan fabrication facility at Stanford. Yeah. Everyone every approach is expensive right and there is no cheap quantum processing platform. Everything is super expensive. Everyone uses fridges. So, so we're working on things again, you know, all the platforms are expensive, require very expensive equipment lasers dilution refrigerators, fabrication, you know, that's all very expensive. But we are trying to put things onto cheap scale which I I feel eventually in the longer term would be uh more scalable and in some way more accessible because we can pack a lot of cubits into the footprint

of uh of the chip and we can also benefit from decades of development in semiconductor integration and processing although we're using a little bit different materials. Is there any point where you're trying to make things work over wider spaces or with more nodes? Like is there ever definitely for quantum networking that's what exactly what you want to do because you need to build nodes every let's say 50 miles from west to east coast if you're building quantum repeater. And then in each of these nodes you would have some of these entangled objects that you would distribute so that you can eventually uh do uh secure quantum communication between east and west

coast. You can't just send entangled photons over 2,000 miles because um it won't work because attenuation lens of fibers are maybe 50 miles and you will lose them, right? And that's why uh quantum cryptography systems are generally short distance. But then you can maybe build and in the context of quantum computing this is also important because maybe on a single chip for example what the Google is doing they can have 100 cubits that are doing something but they're also running into problems of bringing in control wires into one dilution refrigerator. So so they want to shift into the idea of distributed quantum computing. We already have that in classical computing. But then how do you connect these multiple bridges and and build

quantum network of quantum processors because you also have to kind of distribute entanglement in that case as you do in a in a quantum network. And that's also the approach the goal towards which we're going although you know at the moment we don't have limitations in terms of how many cubits we can pack on a single chip. I mean eventually maybe we will have but we're not there yet. Uh so there are limits coming from different uh for different reasons in different platforms that are not there yet for the platform where we're going because we're talking about kind of putting a single atomic impurity at some position on a chip and if chip is four or six inches you can pack billions and billions of them there

right I want to kind of go back to the pictures that we painted a little while ago like the chips and the and the diamond with like light photons basically moving through it in a specific pattern activating the different cubits. I just sort of want to understand like is this an kind of closed quantum system where you're the the things you're communicating between are the these different cubits within this or does it communicate with something with like a duplication of itself somewhere else. It could be a closed quantum system that you initialize and it's doing some sort of simulation or running some quantum algorithm and at the end you just read out the state of that system and you put obtain some measurement result which

provides information about whatever you are doing with the system. But you know of course you want it you want it isolated from the environment as much as possible so that it doesn't lose that those quantum mechanical properties. But it could also be uh a network of system that are communicating together because they're sharing this entanglement as they would just in a simple quantum communication network. But it's basically network of quantum processors where you can achieve additional capability by basically distributing some quantum algorithm over multiple nodes, right, that you are able to to implement. And there are also other ideas and people are also thinking about distributed networks of quantum sensors.

In some cases you can obtain advantage there relative to just localized sensors. So it could be localized to a chip. It could be an system of chips that are connected within one lab. It could also be a system of chips that are connected like over longer distance especially if you would like to do communication. Yeah. Interesting. So, one of the things I've been hearing about lately, especially in the side of quantum technologies, but not so much in the side of quantum computing, has been this notion of quantum sensors. And yeah, I've heard this idea of quantum GPS, which I have no idea what it is, by the way. No clue uh how familiar you are with it, but uh but also similarly like things like quantum hall effect sensing and all this stuff. And it seems that

this is very much in the realm of things that are coming in the near future. But like what are the current like sensing technologies enabled by like technologies now that that like are not so obvious, you know? Yeah. First I mean important thing to mention is a lot of technologies that have existed for decades are quantum mechanical in nature including GPS which is based on atomic physics systems. Uh also time keeping because of the clocks right if I recall. Exactly. Exactly. You know, all of the timekeeping, precision, you know, clocks are also based on atomic physics systems and atomic transitions that are quantum mechanical in nature. But there are new things that are emerging, right? For example, uh magnetometers, right? So a

single one of these cubits that I was talking about one single impur one single atom of nitrogen in a block of diamond can be used to measure with very high sensitivity magnetic fields and with also very high resolution you know in space because it's just a single atom which is you know sub atomic resolution subnanometer resolution. So those are uh interesting ideas that are emerging and again here you don't need to build entanglement of millions of these you just you can still use one of them for something useful and you read out its state optically. So, so people are thinking of using them for um various measurements in medicine, right? I mean

EKG or you know um whatever you produce some sort of magnetic fields that could be sensed with with this interesting. There's a word that's sometimes used and I think after what we've talked about this actually makes a lot more sense but it's teleportation and I'm wondering when we kind of going back to the cryptography side of things like that intersection is it the teleportation that is the sort of system where you can start to send secrets or is that something else? So teleportation is a very interesting concept. So kind of going back to Star Trek, my fave teleportation we all you know grew up thinking well you can teleport things

and fast travels faster than speed of light you know and you can teleport humans teleportation well there there are few caveats here you know we can't teleport humans not organic matter yet I know maybe eventually not organic second it's not faster than speed of light because when you look at teleportation protocols you can teleport the state of an quantum mechanical object from one position to another position, right? You can kind of replicate it at another position, but you're destroying the state at the original position. So if something wrong goes wrong in the teleportation process, you're losing the object at the original position and you are not recreating it

at the the final position because there is something called no cloning theorem in quantum mechanics which says you can prove that you can't clone quantum mechanical objects. So you can replicate their state at another position while kind of affecting their original state destroying it. But the resource again that you use to perform teleportation is entanglement. Right? So without entanglement you can't do teleportation. You f in order to teleport object from state A to state B you first have to distribute entanglement between these two points. And not just that in the protocol you also need to use classical line like a phone line to oh transmit some information in order to perform

teleportation u accurately. Which means that the process is limited to speed of light and it's limited to existing technologies then if or are you using a separate type of quantum thing to make that we're using quantum entanglement so whatever you would like to teleport right if you would like to teleport state of one cubit from one position to another I mean you need a resource to do that which is you need some entangled cubits between two states right so let's say I have something that I would like to teleport right like state of some photon Mhm. In order to do that, I would also need entangled photon pair as a resource between my position and let's say your position. And then we go through that some some protocol, you

know, we're doing some measurements, we're doing something, we're calling each other, and boom, at the end of this, you have that photon in exactly the same state as my original photon, right? Because we shared the pair. So your half of that entangled pair would be exactly in that same state, right, as my original photon, but my original photon would not be in that state anymore, right? Okay. So you need entanglement again as a resource. And we started this discussion by saying, well, all of these, you know, beautiful things that you can do in quantum technologies all require the same ingredient, quantum mechanical entanglement. And that's true also for quantum teleportation.

Teleportation also needs entanglement. But and here's kind of going back to what I was saying with like a encrypted message or something sending something like that sort of like using this for cryptography. Yeah. Does it use teleportation or is teleportation? So it's just a state teleportation. So you would never be able to send info. Yeah. Yeah. Just a state teleportation. I mean you don't get any info unless you are when you're performing teleportation protocol. you know that the replicated state of something that I had quantum mechanical state of something but to to transmit information there are different types of protocols there are many different protocols I mean they're pretty old uh they're 30 well 20 40 years old right 1984 is BB-84 BBM92

that's those are short for Bennett Brasser Bennett Brasser Merine some of them occurred in IBM and so on then you know they they were proposed a long time ago also demonstrated long time ago, you know, 30 years ago for the first time, but over short distances. The new thing is long distance entanglement distribution that you would need as a resource in order to uh scale this quantum communication to longer distances. M so while you know I think there are a lot of similar steps in teleportation protocols and all these quantum communication algorithms or protocols they're different protocols consisting of the same operations that you're performing along the way I see right because you have to in order to

transmit a message securely between us right we need to extract some information from whatever we're transmitting if you just have one object that you know is the same as the object that I had I I mean you are not really extracting any information from that right? Yeah. But what if you had something each of you had something and it was the teleportation that unlocked those things? Does did they ever explore that where it's like I think K I think you had sort of mentioned or this maybe before we started recording where it was like a type of randomness or something where it matched a similar type of randomness like is there something where you can almost like have the same message on either side and it like

unlocks something? So the no cloning theorem actually is very annoying. I mean it's very cool because it's like the reason why all this works but like very annoying right because this is the no cloning theorem is the reason why like when we when you two you know did this thing like whatever 45 minutes ago where you like spread randomness is like why I can't go in and like copy oh yes the cubits and like get your randomness. Okay. Right. So, but that but the fundamental like construction there is that like the the data is random like you cannot get around that like the data for for you two to share the same secret right that is copied in two different places you cannot transmit information

using that data except for the fact that you both know you have some random string that is the same but it it has to be a random string you cannot avoid that course yeah but with that can you do something on either side of course then yeah so shared randomness lets If you do all sorts of crazy nonsense, for example, cryptography. Yeah. Right. If you both share a secret, even if it's random, in fact, ideally, if it's random, then of course you can use it as your private key, which you both now share and you know, nobody has snooped into because you perform this cool check, right? The difference here with um teleportation is that you're not performing any cloning because by construction you cannot do it like you can't like take the cubit and like make another cubit just like it. I see. Like

you have to destroy like in some sense in order to construct the second cubit you have to destroy the first cubit. Got it. Got it. Like weird philosophical I understand. I think I'm mixing it up. I think I thought the teleportation was that sort of randomness being existing in two places. Those are quantumation protocols. That's quantum communication. Got it. one communication protocols where you're producing some stream random stream of bits and you know at the end of the protocol that it's the same between two parties and then you can use it as a secure secure key one time pad or secure key right that that's so cool yeah okay I I now I I think I fully understand also my my confusion I had thought the teleportation was that it's not it's this quantum communication

teleportation is still super useful right for for example connecting different quantum processors you go through some quantum teleportation protocol. So it's not like it's a just a concept from Star Trek that's not useful. It's super useful, right? It's just that and it's the same sequence of operations you use in all of these but just protocols are all a bit different because eventually in communication you want to produce some stream of bits that you know is identical between you and GMO, right? And then you can use it for for your cryptography. Yeah, we just went over that case of the randomness, but is there any other ways that quantum communication, quantum systems are being used to like share secrets to do something that's sort of more edging

onto the cryptography side of things? On the cryptography side of things, I mean there is uh that's a relatively simple thing and there are already commercial systems for this random number generators right? people if they because in classical computers it's pseudon random number that we're generating. So if you uh would truly like to generate random numbers then you can use a very simple quantum mechanical system consisting of a single cubit right because when you perform measurement it's truly in a 50% probability of being heads up or heads down right if it's a quantum mechanical object. So it's not uh biased in any way and if people want to have uh really a truly random number generator then they

can use quantum mechanical system right and it's a simple quantum mechanical system. That's one additional example right of a of a simple quantum mechanical system that could be used for something useful. Cool. With what you've built today like are there ways for people outside of the lab to actually engage with your work? like are there things that people can like are there libraries are there things that people can play with? Yeah, we have open-source softwares. I mean for we were talking about photonics inverse design and uh if people go to GitHub uh my lab's website um Stanford nanoscale quantum photonics group NQP or GitHub spins photonics spin is short for Stanford photonics inver

design software but now there's my former student started started a company focusing on that but we have open-source softwares there and you know GMA was contributing to some of this when he was at Stanford on the bounce way back in the day. back in the day. But if if you would like to play with designing interesting photonic devices, I mean actually you are not doing much. You don't even need to know photonics or electromagnetics to do that. you just define the problem, you know, and say you I want to my light from the input to kind of change in a certain way at the output and this is a black box and basically um then uh this software searches the whole parameter space and designs these structures that would

really do that uh for you. And you know if someone wants to make them in a foundry then they can take this pattern and send to a foundry or go make things themselves that would basically do this processing on light. Yeah. And these are open source. People can download use them on Google Collab or you know their uh GPUs gaming GPUs or if they have you know H100s they can do use them. That be cool. Very nice. Those are hard to get nowadays you know. Yeah we have them now at Stanford actually. Maybe it's time for you to get back into Yeah, maybe I should go back. What am I even doing here? Yeah. Yeah, thank you so much for coming on the show and sharing with us all of this information about quantum engineering, quantum technologies,

photonic chips, the creation of these chips, the modeling. It's been wonderful to get a chance to ask you all of these questions. I've had lots of questions about this field. So, yeah, thank you so much. Thank you. My pleasure. Thanks for inviting me. Thank you so much. It's always always lovely to see you again. You know, it has been a little bit too long. Yeah. Yeah. Stop by sometime. We should grab coffee or something. Chat. Yes. If if you're not careful, I will take you up on the night. You're going to have to deal with me. Yeah. You're always welcome. Yeah. Of course. You're always welcome. Yeah. Thank you, Guo, for in introducing us to Yelena. Thank you so much for that. And for joining us, our co-host today. Um, I want to say thank you to the podcast team, Rachel, Henrik, and Tanya, and to our listeners.

Thanks for listening. [Music]