Building time crystals with quantum computers

Welcome to season 2 of the Coherence Times, where we bring coherence to the entangled world of quantum computing. I'm your host, Ryan Mandelbound. Every other week, I'll bring you stories about how scientists, developers, and businesses are making quantum computing a reality. We'll explore the latest research and development, highlight the latest advances in the field, and see how users are trying to extend quantum to realworld use cases. And today [music] I'm going to talk to you about time crystals. [music] So, we're all eager to see quantum help solve these hard business challenges, but something that I'm really excited about is how we're using quantum computers to better understand the underlying behavior of matter itself. Now, time crystals might sound like a science fiction weapon, uh, but they're actually just a system of interacting particles with unique behaviors, uh, that only emerge when they receive a regular pulse of energy. So, quantum computers happen to be really good tools to study these kinds of systems [music] because they give us unprecedented access to the mathematical rules that govern these behaviors. [music] Recently, a team of researchers across the United States, Spain, and Ireland used a quantum computer to create and study the properties of a two-dimensional time crystal. And here to unpack that, I have two of the authors of the paper. First up is Dr. Eric Schwitzer, a physicist at the National Institute of Standards and Technology. Dr. Schwitzer is especially interested in studying the fundamental behavior of phenomena like time crystals with the ultimate goal of creating atomic scale electronics and devices. Also joining me is Dr. Dr. Nicolas Lorente, a scientific researcher with the Spanish National Research Council, who's interested in directly observing the behavior of atoms. [music] So, let's just get the first question out of the way. Nicholas, as simply as you can, what is a time crystal? >> A time crystal, well, it's a crystal. So, you know, a crystal in the normal life, let's put it that way, it's a group of atoms that are ordered. In a time crystal, what you do is you don't order the atoms spatially, but what you do is you order them in time. What does that mean? It means that if you have some evolution, like the atoms are somehow moving around, what you're going to do is that this motion is ordered. Okay? Now the interesting thing about the um a crystal in general is that the interactions don't tell you how the ordering is going to be. It's more when you put them together the properties of the atoms will make it be order in a certain fashion. In the tank crystal the same thing happens. You can start pumping energy in the system to make it move in a certain way. By the time crystal is going to fall back to where it wants to be in its order phase order phase in time. And basically that gives you an idea about one the ordering of the tank crystal and second the robustness the rigidity of the crystal. >> So I always think about this people always use the word lattice whenever they talk about crystals right and I always think like a lattice fence right so that you have like at every x you have these diamonds and let's say every diamond is an atom in this lattice fence and that's a crystal in space. So I guess what you're saying is like if we could take that lattice fence and now turn it sort of in time right? So now if you imagine the same fence, but instead from left to right, you're moving forward in time. That's kind of like what a time crystal is. Rather than a two-dimensional crystal in space, think of like a one-dimensional crystal where it's here and then, you know, up and then down and then up and down and up and down over and over again. >> Yeah. At the end of the day is a motion that repeats itself, right? That's why you have an ordering or it could be something more complicated, right? Um, so it's a time evolution. >> Just a basic idea. Somebody's like, "Well, okay, I can just create a time crystal by flipping a light switch up and down then, right? Because just it's on, it's off, it's on and off. Biff just made a time crystal, right? I've done a great job and I've solved all the physics, right?" >> Well, I mean, you need more things, right? You would have to um uh have the many body aspect of it, right? A crystal is made of many atoms. So, you need to have many many many uh lights blinking on and off. And then you could make a time crystal with that. uh but you need that the periodicity of blinking is proper to the time crystal. So it's not the one that you are switching on and off. It's how the system resides decides to react. And that is one of the keys why it's called a crystal because it's not what you want it to be but what it decides it wants to be. And the second thing is is rigidity. Even if you are switching crazily, it will go back to switch as it wants to. So it's rigid. So Eric, what we're saying here is that it's not like essentially it is this sort of light switch flipping up and down, right? A time crystal actually does require some sort of input of pulsing energy, right? It's not just like a thing that naturally occurs in nature. >> That's right. There's no free lunch in us trying to create a time crystal, for example. There has to be some sort of energy that both kicks this thing into that repeated pattern, but also helps stabilize that. And the source of that energy can be a variety of different sources. It's like time crystals. It comes from two sides, right? On the one hand, it has this name that's like very science fiction. And on the other hand, it has the sort of uh trivial definition of it doesn't sound as exciting as the name is, but it's actually kind of exactly in the middle where it's this system that has this innate property of orderedness in time. This sort of periodic flipping in time that while you're sort of kicking it in time, it does something different than what you're doing. Like you're kicking this thing over and over again. It's different than what you are doing to the kick. It's like if you kick the soccer ball, but in, you know, or, you know, swung a pendulum back and forth and you're hitting it, you know, over and over again, but the pendulum's swinging twice every time you hit it once for some reason. >> That's right. It's kind of that weird sort of interaction in which we have what's called a subharmonic response and where we kick it and then we actually have to go another kind of waiting period before we get back that original state. And that seems um kind of counterintuitive. I Nico, do you have more info on that? >> Yeah, you're totally right. I mean, instead of uh kicking it producing two swings, it produces half a swing. >> Yeah. >> It's like you're expecting it to with your if it's just moving to your kick. It shouldn't just be moving once per kick, but it's doing more than that. It's this system where all these atoms innately are responding twice to every one kick, even though you'd expect it to do it only once. >> I thought it's the other way around, right? >> Oh, other way. responding half the time even though you expect it to only do it once. >> Right. Right. But that is a consequence of this many body interaction which is something really difficult and it's one of the big topics of physics to understand how many particles interact with each other and behave. >> I see. So this is the actual sort of profoundness of it is that you yes you've created an interesting system but like how on earth has this strange property emerged from these atoms when you're expecting to act one way but because it's all these atoms acting together it's now acting a different way. >> Understood? Now if I take a step back this isn't just about studying like a cool system. This is actually and this is where it starts to get almost a little science fiction. This is exploring like an entirely new kind of phase of matter, right? That we're trying to basically understand what matter looks like when it's in this sort of place where it's basically being receiving these constant kicks. And we're essentially saying that matter itself is beginning to act differently and under goes these phase changes when it receives these regular pulses. Am I right? >> Yeah, totally. I mean, what you're doing is you're exploring a dynamical response that you wouldn't have it if you wouldn't have this driving, right? this thing of ejecting energy as Eric said. So that allows you to explore totally different properties and create matter a dynamical matter at the end of the day. >> I mean this is kind of cool if you think about it. I mean like if we do all of our physics studying based on these sort of these textbook systems where nothing is happening. It's just in a vacuum. You imagine how these systems might interact. But in the real world actually like matter is constantly receiving inputs of energy and you know we live in a world with gravity and electromagnetism and so what's actually happening might even be a little closer to these time crystals in a lot of ways right that actually there's this unexplored weird world of of these sort of driven matter with energy being inputed that makes it sort of different than just like what it says in our textbooks. >> Exactly. Exactly. And that is one of the lines of research nowadays. We are more and more moving into these dynamical phases. Yeah. >> Can you now now we call these systems like out of equilibrium systems? Is that's what it's called? >> Yeah. The you know an equilibrian is what you just described. It's these boring phases that allow you to understand andize everything. But at the end equilibrian is something like death, right? You cannot do anything. So but when you get out of equilibrium is where life happens. And not only life, all the devices, every device that you're using, everything is happening in out of equilibrium because you're driving charges across. You are making a change in time. And this is exactly what we are trying to do now. We are trying to understand what happens to these systems when they are dynamical. >> And then um so Eric why why would it be interesting to study this system like what can we do with this thing? So that's kind of a a complicated question because when we look at the physics of it, um what we can do with it is a bit more subtle. So for example, it as you've kind of alluded to here, a time crystal isn't a device that can let us travel through time or store unlimited energy or build like a magical clock. What it does is it gives us a new kind of organized behavior in matter. So their use is primarily scientific right now. So the way that I see it is that we're using these as test beds for studying these quantum systems far from equilibrium. Um looking at that disturbed rather than sitting quietly in the lowest energy state like Nico was saying where with the crystals that you would see inside of a science museum and from some of the quantum technologies that we're envisioning and thinking about this is exactly the kind of like messy driven environment that we want to study and the time crystals give us that ability to do so. >> I see. So it's really like a platform. we have this interesting system with this really interesting property that is getting us giving us basically access to a world that is closer to our own world than the textbook world. Um so even though it sounds very science fictiony, it's actually like this thing that's helping us study a field that we just haven't really had access to studying before. >> That's right. That's right. >> I mean and also I don't know anything but you have this sort of driven system that maintains like an exact Right. It basically is a metronome that's resistant to you kicking it, right? Because this is something we didn't really touch on that if you change the kick a little bit. Like you're kicking it, you know, boom, boom, boom, boom, and you sort of mess up a little bit. Boom, boom, boom, boom. Then your metronome is not responding differently. It's keeping that exact halfep order, right? So that's actually kind of cool. Yeah, that was actually something that was kind of surprising when I first looked at this research and I was reading about it is that you can kind of get away with these imperfect kicks or you can get away with a level disorder, but that's really kind of the hallmark of a crystal, right? That you can have these external forces that are acting on it that are somewhat chaotic or a little bit random or maybe imperfect, but the crystal stays a crystal. And so in the same way, the time crystal should also in some way stay a time crystal even with those imperfections, >> right? Like a diamond is the perfect example of this, right? like diamonds are the are the ideal crystal. They actually look basically like a threedimensional version of this lattice fence model that I just thought of. And then but if you like they're very hard, right? They're resistant to to scratches because they are really good at being crystal. They're like the the number, you know, the number one crystal like >> Yeah. The bond strength that that you see in diamonds obviously is very very strong. >> I prefer sodium chloride instead of diamond. But yeah, why not? >> Sure. And I'm I'm going to go with you on that, Mo. I love sodium chloride, especially on stage. >> What's good about it? It's square instead of sort of >> square and you said that yeah diamond is super complicated. Yeah. >> Cool. So let's go a bit. You know I don't know if our listeners have heard like kind of gotten some of the backstory of time crystals. So Nico like has there been a sort of endeavor to actually get us to this point? Like how where did we start and how did we end up here with these with time crystal research? Oh, so I think the first um articles that people where people um start talking about this uh first um uh they they it was a very theoretical idea. The in the same way that in the interactions that you have in matter in a normal crystal you have this symmetry breaking and then everything becomes ordered instead of just being chaotic or homogeneous or something. Uh then um uh Vilsec had the idea that maybe what is in time you could do the same thing that you could just instead of having something that is boring and stop or moving in a chaotic way you could have an order phase and that's how it it became and very quickly they realized that this couldn't happen in equilibrium uh because in equilibrium it would basically mean that this thing is always moving it will be this um um perpetual mole right and um however uh when you are out of equilibrium, when you're injecting energy in the system, you can make it move and then it will react as it wants to, not as you want to because it's a tank crystal. >> So this is this is Nobel Prize winner Frank Wilchek we're talking about, right? >> He's cool. He answers my emails when I was a journalist. >> Yeah, I always appreciate him for that. Um and uh you know what? So I know that the first kind of realizations of time crystals kind of are not that far away. I mean it was only about 10 years ago that and I guess it was the folks like Norman Yao's group and folks like that who sort of started realizing these first representations. Um how did you all feel like Eric when you saw the first time Crystal paper come out? I think it was either it was probably in nature like what what what was it like reading that paper? >> It was actually kind of funny because I was wasn't working in physics at all. I was actually working at a health insurance company working on leadership development and I saw this little news article saying time crystals and when I first saw it I was incredibly skeptical. I was like, "What the heck is a time crystal? Why can they get away calling it this?" But the more that you look into it, the kind of like the beauty of symmetry breaking and how that can lead to this like phase of matter really got me interested. And in fact, I think it was in 2017 when I saw um two results. one from professor Chris Monroe who's at the joint quantum institute and another from professor Mika Lucan over at Harvard where they demonstrated this inside of uh one-dimensional spin chain in the uh the group of uh professor Monroe um and then there was another one uh this nitrogen vacancy centers in diamond so to see these time crystals showing up in these like more complicated systems was like okay this might be real this might be something that is worth studying and and uh pursuing further >> I see so it's basically like you know, not only of course you can, not of course, but you see these things being done in trapped atom systems and you're like, okay, well, that's a pretty simple system. It's cool that we see it there, but then actually starting to see it emerge in these complex systems. We're talking about these uh nitrogen vacancy centers, right? Those are diamonds where we've replaced some of the carbon atoms with with nitrogen. Is that correct? >> That's correct. Yep. >> Right. And so you're actually seeing this emerge in like real interesting sort of macroscopic systems. It's like, oh, this thing is real. >> That's right. and and it's kind of like uh took that 1D nature thing that was showing up inside of all these papers and it really took it to that next level of hey maybe we need to be looking at more complex time crystals things that emulate uh the stuff that is happening in nature because obviously there are these experiments that are showing that something is happening and maybe we can harness that complicated nature for uh to create even more complicated time crystals. >> Cool. Well, and Nico, you too. I mean, you know, you're interested in study sort of observing like atomic basically actually seeing what's going on at the atomic scale. So, you were probably excited even from the first demonstration, right? Seeing these 1D systems beginning to experience these behaviors. >> Yeah. Yeah. Definitely, definitely. It's super interesting. And actually, one of the things that we are doing in in in our group here in in San Sebastian in Spain is to study uh atoms on a surface. So you can put atoms one by one using the scanning tiny microscope and then you could actually uh make one of these tank crystals by ensembling one-dimensional atoms. So it was definitely from the very beginning one of the lines of research that we wanted to pursue. >> I think that's like the perfect segue into what we're talking about here which is most of these time crystals like kind of these first demonstrations in the past couple you know past 10 years have been on just single straight lines essentially of atoms. So, you know, you can imagine like red, blue, red, blue. But what you know, you've actually taken this into the next dimension. Not only are we looking into the time dimension, but we actually have a sort of like a time cube going on like a you know, like a checkerboard that's kind of moving through time and maintaining this checkerboard back and forth order, right? Um, so what is the importance like like tell tell me what you actually are studying and what and what you found? Why why move into the second dimension? >> With onedimensional systems, they're nice. uh they're somewhat kind of straightforward to study, but when you get to two dimensions, things become very complicated because of their connectivity. So the example I like to think is like a conga line, right? You're interacting with the neighbor ahead of you and behind you in 1D, but once you get to 2D, now you have a situation where you can interact with a lot of these neighbors. So that fundamental jump into 2D was already kind of a difficult problem to study. But also um we looked at some prior research that had looked at something like 2D and they were looking at these very simplistic interactions. But it turns out that if you add this ingredient called spin flip uh essentially the spin flip strength it becomes so complicated that there are certain parameters with different measurement outcomes that are so large that you could not do this on a classical computer. It was something that was very difficult to to get in a reasonable amount of time. >> And guess what? we have a quantum computer. Um but we should take a step back on this one because I think this is the part that um is hard to understand. It's all a little hard to understand but this part is is a you know a bit of a challenge right. So we've implemented these time crystals on a quantum computer because a quantum computer ultimately is a one-dimensional it sorry it is a quantum system where we can control these this quantum property. every cubit can be like a single point or node in this crystal and then we can use the quantum computer's properties to basically create a real life quantum real life time crystal right I mean Nico quantum computer is a time crystal am I correct >> well yeah I mean a quantum computer is something much more versatile and useful than just a time crystal right but what you can do you could think at it as a um an experiment the quantum computer is like a an experimental platform where you can program the experiment you want to perform form right? So in this case what we are doing is we are having cubits that will replace our atoms say in our crystal and then we will put interaction between the cubits and that's something that the quantum computer naturally gives you and then you can design the kind of driving that you're going to do how these things are going to move up and down. So then you can design you can program two things the interaction and the driving how you are going to pump energy in the system and then you can just exactly map a um tank crystal and you are actually making a tank crystal because that's what you're doing putting cubits and making them interact and making them evolve in time >> right you have this many body system you have you're welcome to induce the kick drive through your quantum gates and then you're welcome to measure it as you do your experiment and observe this order. How has this sort of opened up time crystal research? I mean, Eric, prior to this, people were doing some pretty hands-on experiments with lasers and and and you know, single atoms and these crystals, but now you just have access to like a computer that is a time crystal, right? How has that changed the game? So, when we get now to these uh 2D experiments, we now have a little bit more complicated interactions, but still the fact that we have like hands-on access to the machine essentially to be able to create these things is something that I don't think that we had before. It's kind of logical when I mean if you look at the paper there's this beautiful diagram. I mean there's a lot of physics words but then if you look at the diagram it becomes really clear. You have just a map of the cubits which if you haven't seen it looks like a basically a bricks you know and at each sort of place where there's a T in the brick wall there's a cubit at the sort of juncture. Um and if you imagine that it's just at each of the bricks there's a black and white you know it's alternating black and white. It's essentially looks like a brick pattern where there's blacks and white circles on it and then, you know, colored lines connecting the black and white circles to represent the gates between them. But it just looks like a checkerboard. I mean, it really is, you know, profound. It's like, oh, there you go. It's a time crystal. The cubits, you know, they're in one state when they're black, they're in the other state when they're white. It switchs black, you know, back and forth. And uh, you know, that's the time crystal right there implemented on the quantum computer, right? >> That's correct. Yeah. And those colors that you're referring to on that uh that figure, that was the initial state that we gave to the system. But yes, we saw that every time that we had this uh every other kick, we would see that that pattern reemerge back to itself. And we used some complicated math to kind of describe, you know, how good that pattern came back. But nonetheless, we got to see that thing flip-flop back and forth. >> And um you know, maybe Nico, when you first saw the results, what was it like? Do they look good or were you how did you feel when you first saw the sort of measurements come back from your first run of this? >> The first three months were pretty tough because the results didn't make much sense. [laughter] But then uh IBM made one of these miracles and suddenly the quantum computer was really working perfectly or very well let's put it that way and uh then we could get a signal that was there. I mean we could actually see the tank crystal happening in real time basically and that was super cool. Yeah. And it was fairly early. I mean we think about it 3 months it research is nothing right. So yeah, it was very nice. >> The fact that in only 3 months, you basically took this entire area of research that was only being done in one dimension and then we're like, "All right, let's just see what it's like in two dimensions, but I I want to go from this one dimension to two dimension thing." So, how long did it take to sort of go from this sort of one-dimensional system up into this sort of checkerboard system? >> Yeah, it actually didn't take that long. I think the first uh experiment that I did right after we did this 1D um it maybe took about a week because there was a bunch of lessons learned uh what we had to do. It was just a matter of trying to map this to the device properly and also the added complexity. Um if we mention back again this figure that we have we have three different colors for the gates. So the fact that we have more interactions that are occurring all at once, three different interactions actually per means that we have to kind of make a more complicated circuit diagram. So that part took a little bit of time as well. But at the end of the day, I would say it would be less than a month to really get the the good results. >> Wow. I mean, right, the onedimensional model is kind of like you have to draw a snake through the bricks and then once you get to the, you know, that's your 1D is just cubit one connects to cubit 2 and then it connects to cubit 3, etc. While this two dimensional model from this diagram I've seen essentially like every cubid is connected to all of its neighbors in this sort of checkerboard. >> That's right. There are all those what looks like in the diagram kind of like a rectangles or squares that are connected in a 3x uh five or 3x six pattern. >> Got it. Um now um how did it differ? Did it look I mean so surely stuff emerged from your new sort of added dimension that you didn't sort of see before. Right. >> Actually I think it was kind of the opposite. I think we assumed we had had a hypothesis that as soon as we went to 2D there was more connections there was more ways for information to be scrambled inside the system. So therefore the time crystal maybe doesn't exist in so many different areas and it turns out actually uh to our surprise that it does. It exists in very similar regions that happens in 1D. Uh Nico do you have anything to add to that? >> I agree totally with you. Yeah. Mhm. So basically the surprise was that it worked like it did in one dimension. >> Yeah. I mean the parameter space is a bit different, right? There are some properties that um two dimensions are a bit different and that's something that we are actually working on right now, right? >> Yeah. >> Can you get into that? So I mean I'm interested totally in the research here and how it's differed and what's uh what's new. >> Well um now we are actually studying the diffusion of information inside the time crystal. So uh like when you if you say instead of just giving energy to all these spins all the cubits that are flipping around changing their state between zero and one we just give it only one then I want to know when a cubit that it is say n cubits away notice that this one flipped right and how this information diffuses and we are looking at this and then when you do it in two dimensions this is totally different because it has many more paths to reach for reaching a certain point right and uh that is a very complicated problem and um we are working on it. >> So when you say that information diffuses um can you dig that into me a little bit more just because um you know obviously my understanding the time crystal is just sort of like everything is kind of flipping on schedule alongside everything else. So what kind of information would be diffusing and and and for what reason? information means that um the the the um say the state of one of the cubits is going to affect the state of another cubit and how this happens you could see it as information that this one is learning about the state of this one. So that would be the information right? But another way of saying it in physics is that you have an interaction and this interaction is propagating in time and actually we talk about the light cone because we can imagine that in relativistic physics there is a certain speed where you can actually send your interaction your information the information propagates at maximum the the speed of light. Now in this system is not a relativistic system. So the velocity is something totally different. the velocity is actually proportional to the interaction that you are setting between cubits and then that is going to give you the higher velocity where you can transmit your uh velocity but that will depend on what kind of regime you are because then because it's a diffusive system you can be in different regimes where the speed is higher or even the ly cone starts bending over or etc. It's a very interesting problem, >> right? This light cone concept was hard for me when I first started learning quantum, but I think it actually is quite intuitive, which is basically like if you're egg number one on the egg crate, like you can't sort of get to egg number 12 on the egg crate instantly. You know, you have to sort of hop over to each egg as you're getting from 1 to 12. And so that's happening on all at all eggs at the same time. And that has that takes some amount of time. So if you make a graph of it, there's sort of this cone shape as you move forward in time. >> Yeah. But instead of X, we have cubits. Yeah, it's more or less the same thing. >> We have totally buried the lead here. It is crazy that it is so intuitive to do this on a quantum computer because it's just like you can just study time crystals on a quantum computer. Just set the cubits, put the gates. I'm oversimplifying, but it's it's actually makes a lot of sense. But this is like so freaking hard to do on a classical computer. This is like this is actually really challenging, right? >> Yeah. Yeah. Yeah. Very quickly, you're limited. >> Can you can tell me about how hard like how hard is this really on a classical computer and why is it so hard? >> Right. So there was a a a point inside of our phase diagram that we were looking at that it turns out that it took uh our partners over at IBM almost a month to try to calculate on their classical computer whereas it took several minutes on the quantum computer. And it's because actually of this growth of entanglement that occurs inside of our system. It happens at a rate that in some cases can be so fast that the classical computer needs a lot of memory and a lot of different uh uh processing power in order to be able to compute that. But we just don't have that right now. And so therefore there was no way that we were going to be able to simulate this in a reasonable amount of time. And I think this happened several times uh while we were trying to explore this phase diagram. >> Definitely uh you can just think about it like in a classical computer what you do is you're solving equations. So you have to store all this information and you have to do a very complicated mathematical algorithm. In a quantum computer, what you're basically doing is what you said before. You're doing an experiment. You have your experiment, you put the right parameters and voila, you have the answer, right? So um it's a totally different thing. >> I mean this is like we we talk a lot about quantum advantage and when we're going to hit quantum advantage. We don't have to say that this is a quantum advantage. But what I will say is that it's like you have this system that is the time crystal like it is it just you're making a time crystal with the quantum computer because that's the that's what it is. It's the properties that you can use. You can exactly recreate the exact thing and study the exact physical laws of time crystals on this quantum computer. Or you can like hamfistedly create these like big matrix tensor sort of things and do everything really, you know, by hand on a sheet of paper and then you could make your study a time crystal with the laws of physics on a classical computer. It's like why wouldn't you just do it on the thing that looks like a time crystal? >> I mean at the end of the day is like comparing a digital computer with an analog computer, right? In an analog computer you are playing with the laws of physics. So here's a little bit the same thing. you are playing with the quantum laws of physics and uh that gives you enormous advantage >> but I don't want to totally discount the classical computer right because one of the reasons that this worked was because of this sort of interplay between classical and quantum computing right like can you tell me a bit more about that >> go ahead Eric [laughter] >> Eric tell me okay uh so yeah the uh there was this algorithm that was developed actually by our IBM friends over at IBM Ireland and IBM quantum and They essentially had us do an additional circuit that we knew could be simulated on the classical computer and then using some tensor network methods, we were able to kind of extract the impact of noise and errors on this kind of flipping thing that we were seeing. And because they were able to do that on the classical computer, we could augment the signal and then we could see the time crystal a lot better. And if we didn't have that uh that ability, then we wouldn't have been able to claim that we saw uh the time crystal in behavior that we did. We should shout out Neil Robertson who everybody, you know, I don't think he could be there, but he definitely, uh, he was very, I know that he kind of helped you all a lot out with this, right? >> Yeah, he did. >> Now, um, you know, what's even more cool than this is, you know, we talk about these tensor network methods. Now, obviously, anytime I hear the word tensor, the first thing I think is, oh, you can do that on a GPU, right? >> Sort of. >> Sort of. I mean isn't there I mean what is aren't there plans to sort of augment this with potentially not just sort of classical processing but also you know maybe like GPU processing as well >> I think so yeah I mean uh the capability of doing massively parallel calculations in a GPU can be used in terms of networks and I think they are using it now one of the uh projects that we have actually here in San Sebastian um in the company multiverse is to use this in um together with GPUs for optimization problems and particularly for um AI and things like this because this gives you massive uh parallel um optimization and tens of networks is doing this >> in terms of optimization problems. Are you saying that like the time crystal research is going to help with that or sort of beyond that we are also working with tensor network methods for the optimization problems? >> I think it's beyond that. It's more like you know tensor necros allows you to like work with enormous amounts of numbers. So, can you tell me a bit about like some of the challenges that kind of arose in attempting to do this? I mean, it's I mean, I'm think we're making it sound pretty easy, but I don't actually think it was easy at all, right? I mean, what what how do you actually go about the sort of process of implementing a two-dimensional time crystal on a quantum computer? >> Yeah, it was actually kind of uh difficult to begin with. I think this was alluded to before um with noise and then how do we design the circuits in such a way that minimizes those noise and if we just were to use out of the box solutions um it turns out that the depth of the circuit was just much too high. So we had to get creative. We had to look at some research that have been done by others that showed that you can actually compress the circuit as much as possible down to its uh smallest amount essentially per uh per flow k per time step. Uh we also saw problems of chords with noise. I think that's unavoidable especially with these noisy quantum devices that we're using. But again this is where that tensor network approach really came in and made things a lot better for us to see the signal. And then I want to go back just to the um some of the kind of interesting results that emerged from this. Now um when I sort of was looking through the paper, I saw actually like these phase diagrams that we're actually observing sort of this changing phase between three different phases. Can you one of you take me through that? >> Simplifying things pretty much. Uh basically we have three phases. One is the many body localized one which actually means that um you disorder the system. You assume that your system the interactions are disordered. When you do that what you create is like pockets of states such that they don't communicate properly with each other because they are localized as the many body localization. You have interactions many body but because there are there is disorder they cannot communicate very well between pockets. So then uh that allows you to have um um giving energy to the system as much as you want to without having an infinite temperature because you could think about it right uh like if you have a system you start giving it energy and you have no way of dissipating your system will end up blowing up uh so you will reach infinite temperature at some point. Now many body localization prevents that and when you have that um and you are applying the right kick to your system then you can create the time crystal and that would be the second phase that is the dynamical part and in between you what you could happen is that you are not in the condition of creating localization. Why? because you are connecting very strongly uh the uh fluctuations of one spin with another one and then you create entanglement. In the moment that you create entanglement very quickly information spreads out all over your system and then you reach infinite temperature and then you blow up the system basically and that is the erodic phase. The erodic phase is like highly calic and yeah you can imagine infinite temperature. Can you take me into like what this might look like if we mapped it onto sort of a real world system? You know, I mean, obviously in my head I'm thinking about things like glasses or real world crystals or things like that. I mean, how do these sort of what would it look like for these sort of phases to arise in, you know, a diamond I was throwing down the steps? >> Yeah, excellent question. Actually, the many localized phase would be like a a a um glass. Actually, we call it spin glass many times, right? This would be the spin glass phase, right? Where everything is like frozen in a disordered pattern. Now you can start driving it and because it's sort of frozen, you drive it locally, but they don't connect to each other. Uh and then you don't make the full system blow up. And uh in the moment that this thing starts percolating, then you go into this highly entangled phase and then you have the er godic phase. And yeah, it's it could be like melting your um your your glass, >> right? And this is also exactly why you can't just do it on a classical computer. Like surely you can represent bits as zero and one and then say okay spin up is zero spin one is down spin whatever spin down is zero whatever zero and one up and down and you map it to a computer but then you don't have the ability to study the interactions. You only have logic gates to study. So this is actually where quantum gives you that ability is you have quantum interactions that you can just natively model between the in the cubits. you actually have like an experiment you can make while in the classical computer what you do is you solve equations right >> now what direction do we go next I mean we've gone from one dimension to two dimensions we can go to three dimensions if we want I mean what what what sort of questions are have have have we left unlock unasked >> I I've got a bunch of questions Nico I don't know about you about uh where we're going with this [laughter] um so you actually mentioned it different types of 2D systems going up to 3D systems but I'm also interested in how the type of geometry in 2D matters. Like for example, does a square lice or a kagome lattice give us different behavior or not? Um and again going back to Nico's qu uh line of research here, how does entanglement spread in these systems uh beyond in 2D and maybe even if we go to 3D um and are there more complicated interactions uh that we see in other devices that we could try to model that may also show collective non-equilibrium behavior? I I don't have a good answer to that and I know that our community is looking at this. Um but the question is can we even apply the noisy quantum devices that we have today to help try to solve those problems. >> So if you're interested in square lises then you're probably excited to implement this on Nighthawk. Uh >> uh yes >> yes what can we do with so Nighthawk you know we have our existing suite of chips which are these heavy hexagons these sort of brick-like pattern arrangements of cubits and of course soon you I think already we have access to these he you know these square lises where every cubit is connected to four other cubits. Now what is what do you get you know for this experiment when you have access to these more connected cubits versus these less connected cubits. So we get more connectivity which means that the spread of information and the spread of disorder for example is going to be different. I think at least I hypothesize that we should see a phase diagram that looks a little bit different than what we saw for the heavy hex. Um I know that we have experiments underway right now to try to tackle that question. >> And then the last thing I have just a question from the paper which I couldn't help but notice the word cat scar. Um and I just don't know what that is but it sounds bad. So tell me about it and I just want to make sure my cat will be okay. cat scars aren't bad. They're good. So, uh these are many body localized scars. That's the ones that it looked like uh that we found. So, essentially when we look at this phase diagram, we fully expect in some regions that because of this spin flip strength that the system is going to thermalize um regardless of the initial state that we put in there. But it turns >> thermalize meaning like just fall into randomness essentially. >> That's right. But there were certain initial states uh for example one in which all these spins were all aligned at the very beginning. It turns out that that was a very special initial state and that as we ramped up the spin flip strength instead of kind of disappearing to the chaotic mess it actually stayed there. So we call this scar because it's leaving this visible trace of those dynamics that occur in the system that would be otherwise jumbled and chaotic. So this this is like every time I hear this sort of the sort of long-term memory of order in a quantum thing it is exciting right I mean this is a you know what is sort of the implication of having access to sort of this I want almost say memory that the system has um for these systems >> I think it shows us that there is some limits to um what initial states can kind of survive with the disorder that's around it I think that kind of gives us the bounds to that I'm not sure if there's uh anything more that we can pull from that. >> So then um you know let's just talk about I mean we asked a little bit about what's next but um you know maybe uh Nico thinking about the future of your research you know where do you want to end up taking this like where what would you do if you had access to sort of an all powerful quantum computer for something like this? >> Well I think that the very exciting thing is to be able to um explore um the behavior of matter under different conditions. So for example, one thing that we have done in this study was to map these three uh phases that we described a moment ago. But we can think about going to other regions that we have not explored like for example um spin liquids where the quantum fluctuations are so strong that you don't have the freezing of the uh spin glass. And then uh this is a super exciting uh different topic that people have been working on it for many years and having a quantum computer that can give you direct access to this phase is is super interesting. That's one thing. The other thing is again this scrambling of information how you propagate signals how you can create eventually devices. we can think about actually creating a device uh functioning with this kind of properties and yeah we we have to explore all that. >> Cool. And Eric, you're interested in I know you're interested in atomic machines. Tell me about the kind of device that we would be uh sort of thinking about with access to a much larger you know higher coherence time quantum computer. So some of the systems that I'm looking at have very strong correlations between atomic scale sites, but the particles are also very localized and that makes it very difficult to study with a classical computer. And some of these systems that we're looking at are also in 2D. So you can imagine that a quantum computer that's large enough has low enough uh error rates will be able to take that more complicated model and watch the dynamics of this thing see what does the ground state look like what does the phase diagram look like >> and what kind of machine would that end up being like a pulley a lever Xbox >> I'm thinking something like uh like devices in the siliconbased devices uh that we look at for example I know at uh the division that I work at the nanocale device character erization division. We look in the atom scale device group in these uh defects inside of silicon based devices and trying to harness those defects at the nanocale. Um we've even seen how some of these uh defects and defected systems can be used as analog quantum simulators. So maybe we use a digital quantum simulator like the uh quantum computer to simulate that device and maybe create an even better one of those devices. So it's almost like u you know feedback loop where you can basically because you can simulate quantum really well with the quantum computer. You can use these sort of earlier quantum computers to simulate better and better quantum devices. Especially as we kind of envision this big world of quantum technology of connected quantum mainframes and data centers and sensors, you know, and like uh having access to a quantum computer that can create really interesting quantum systems can make even more interesting quantum systems as we work to realize this uh quantum future that we all want to see. >> Definitely. That's right. The feedback loop of simulate the simulator. >> Awesome. Um, and my final question for you, um, let me think. What do you want a world with quantum computing to look like? Eric, you first. >> Okay. Uh, a world of of quantum computing. Okay. I [laughter] so I do envision kind of a a a world in which we have this quantum simulation that's being done by all these different type of simulators um for very large complicating systems but that can be studied not just by physicists like Nico and I but can be done by other type of scientists in all different type of fields cuz I bet you that there's a lot of different applications that we haven't explored yet and we just need to uh get these machines bigger and better and then maybe they can do that. So like unlocking these quantum properties if I'm in a field where I'm working with quantum stuff but might not be innately quantum smart person myself like medicine or you know energy things like that. >> Right. Right. Trying to make it as usable as possible for the the wider community I think is a good way to go. >> Awesome. And then Nico what is your ideal world with quantum computing look like? Well, at the end of the day is sort of what Eric is saying, but uh I would phrase it in a different way. Like what we want is to have a computer that is really a computer, not just a well, not just this is super interesting what we have now, right? A place where we can program the experiments. But we don't want to do experiments. What we want is an agnostic system that can do any kind of calculation. And the access to quantum will mean that we have this exponential growth and we will be able to solve super complicated equations but we will be able to do enormous amounts of transactions or we will be able to use it in any um um uh subject as uh Eric said but it will be just a computer right something that we don't know how it's working inside but it is having this tremendous quantum advantage. >> Awesome. All right folks well that was really fun. I really appreciate you giving me all this time to talk about time crystals [laughter] >> and uh yeah, thank you again. >> That's it for this episode of the Coherence Times. If you enjoyed the conversation, please be sure to subscribe wherever you get your podcasts. [music] Comment in the comments section and share it with somebody who is curious about quantum. You can find us on Spotify, Apple Podcasts, and YouTube via the research channel. And for more episodes, resources, and deep dives, please visit us at ibm.com/think/mpodcasts. I'm Ryan Mandelbound. Thanks for tuning in. And remember, the quantum future isn't just coming. We're building it right now.