23082 >>: Today we have Roman Lutchyn here from -- he actually works at Microsoft Research at Station Q in Santa Barbara. And he's visiting today. He graduated from the University of Minnesota and then spent three years at University of Maryland before coming to Station Q where he has been for a year. Just about a year. So he is going to talk to us about topological quantum computing with Majorana fermions. Roman. Roman Lutchyn: Thank you very much for the introduction and for the invitation to come here and give a talk. It's always a pleasure to visit Redmond. So this is Microsoft Station Q. We are -- you know, that is view from our building. We are located in a very nice campus. UCSB campus. And our group consists of mathematicians, physicists and computer scientists. And what we're trying do, we're trying to explore the -- this interface between three different disciplines to build a quantum computer. And the particular approach that we pursue is topological quantum computing. And I'll explain today what this is about and why this approach is very promising. So most of the people in our group are theoreticians or -- but we also coordinate with many experimentalists around the world, and we just ask them to do some experiments to test our ideas. So this is the outline for my talk. I will start with some introduction of quantum computing and I will explain why it is advantageous and then I'll explain what is topological quantum computing and present some realistic systems where we think we can implement this approach. So as you know from every day world that most of the physics -- most of our every day life can be explained in terms of classical laws. Newton's law, Maxwell equations of electro magnetism, et cetera. But if you scratch the surface, we can quickly realize that the world is actually quantum mechanical. And these are devices that we use in everyday life, and some of the things in those devices is actually based on quantum physics and transistors tunnelling. Tunnelling is a particular quantum phenomena. In lasers, the fact that you have a discrete two-level system, so you have quantitization of levels is also a quantum mechanical phenomenon. In MRI, the fact that atoms have nuclear spins, nuclear spin is also quantum mechanical object. So I would say we use these properties in everyday life, but we use them not in -- not to the fullest extent. And in order to do that, we have to go back to the principle. In order to explore quantum mechanics in depth, we have to go back and use the most fundamental principle of quantum mechanics, the principle of superposition. So if quantum mechanics postulates that if you have a state Psi, then the state can be a superposition of two other states, zero and one. And principle of superposition can be illustrated in this very nice experiment, double slid experiment which we all know and love. So if the light was -- I mean, the light is a wave, so principle superposition is just the fact that we add the amplitudes for the light and because we add amplitudes, not probabilities, we can have this interference pattern on the screen. So the idea is you send the light here and here, and then the light can go either through this pass or this pass. And then depending on the position of the screen, you'll see a sequence of these dark and bright spots. And so that's the principle of superposition because you have to add the amplitudes for electromagnetic field and then square the amplitudes to get the intensity on the screen. So a long time ago, in 1982, Richard Feynman, very famous American physicist, asked the following question: Can we explore the God-given resource, the source of quantum mechanics for something useful, for quantum computation, for example. So if you do that, then you immediately realize that because each of the -- each state of the system, if it's a two-level system, can be in superposition than if you have N such two-level systems, then the information that is encoded in these two -- in such a system scales is two to the n. So it scales exponentially with the size -- with the size of the system. And that's actually something we are trying to explore in quantum computing. We're trying to explore the fact that we can store much more information in quantum realm than in classical realm where you can just have zero and once. So what is the potential of quantum information processing? Actually, there are many. And, for example, if you want solve a system and you want to -- you would like to understand, say, properties of particular system, like a chemical reaction that involves some quantum mechanics, then we can simulate that on a -- on this quantum computer so we can basically solve any quantum system by just programming it into our quantum computer and running the simulation. Another application is factorization of large integer numbers. And that basically triggered this interesting quantum computing because this particular operation lead to -- this particular algorithm leads to exponential speedup. So our security will be in danger if we have a quantum computer, or maybe we can invent something more complicated. Quantum encryption. Another thing which is fairly useful is quantum search algorithm. And so basically a search of an entry in a database, but here we have algebraic speedup rather than exponential, but it's also very useful. And finally quantum cryptography. So quantum physics allows you to send information securely and the security is guaranteed by the physics. And so these are just a few potential applications. People are working and discovering more and more algorithms, but what I would like to talk today is actually how to implement, how to build a quantum computer. What are the systems that are suitable to do this task. And Krista here is doing research in quantum algorithms, so if you are interested, you might want to talk to her and learn some more details about this. So in classical computer, which I have a bit which is a basic logical element. It has zero and one. So we need to find something similar in quantum computer. And it's called qubit or quantum bit. Essentially qubit is a quantum two-level system, so these are the two levels. And by controlling the couplings, so very often two-level system as we present it by spin, so you can have essential [indiscernible] algebra. And if you control all the couplings, if you control all the couples with a spin, then essentially you can program the state you want to get. So a qubit is a controllable two-level system. You can drag this vector by applying some gate, by applying some gate voltages and signals and changing these coefficients, you can drag this vector and put anywhere on a bloch here you want. So basically, you can change all the coefficients here and here. So that's an idea. So you would like to have a controllable two-level system. Well, these are the potential realizations of these controllable quantum two-level systems, and I would say they come in two flavors. So the first one is based on microscopic degrees of freedom. So like a nuclear spin in MRI, this is a microscopic degree of freedom. Or position of an atom in particular location, that's also a microscopic degree of freedom. And another type of qubit are -- another type of qubit is placed on collective degrees of freedom. And these are superconducting qubits and actually also topological qubits that I'll talk about today. So the idea is there is that a lot of electrons, they behave in a coherent way so that it's like an orchestra. You listen to orchestra and you see that everybody is following same, you know, pattern. So the same thing happens here to the extent that you can describe ten to the nine electrons by a single variable. So that's what happens in superconductors, and in topological qubits, we would like to, you know, we would like to have that but even go beyond that paradigm. So one of the problems in quantum computing, I mean, these approaches fairly developed now and people have created these two-level systems. They manipulate them. They couple them and try to play with them. But there's a fundamental problem. And the fundamental problem is called decoherence. So if you have a microscopic system and you want to couple to this microscopic system and control its position on a bloch sphere, it turns out there are other couplings which are unavoidable. And we cannot just filter out useful couplings and, say, from unuseful couplings. So these unavoidable, say, harmful couplings lead to errors in our quantum computation. And this process is called decoherence. So that's why we talk to experimentalists, because quantum phenomena occur in the lab, and in the lab, you really have to make your hands dirty, and you know. So the process of quantum decoherence, something that people know from the lab experiments and they measure it, so they start in some state and then they see that this state will decay to a ground state, for example, if you start in the excited state with time. So your information can be preserved for some time. And we would like to actually -- we would like to try to correct these errors, but it turns out that, unlike in classical computers where error correction is very developed and fairly successful, in quantum computers, it is very hard to do. And the reason it is very hard to do, because if you want correct more and more, you need to introduce more and more resources. So there are some approaches where people established what should be the error rate, and this error rate is ten to the minus five. So it's fairly -- fairly small. And so far, we are trying to achieve this error rate but it's very hard. >>: [Indiscernible]? Roman Lutchyn: Yeah. So you have a decay time, and you have a speed of particular operation. So if you divide these two, that will be your error rate. >>: [Indiscernible]. Roman Lutchyn: Excuse me? >>: [Indiscernible] can people do today? Roman Lutchyn: It depends on particular system. So I think we are reaching ten to the minus four in certain qubits, but the thing is, there is something you think explored here. This is error rate for some models. And if we scale the system, this actually decay rate might also scale with the sizeable system, so this area needs to be explored more, especially when you have more and more qubits together. Because there is also a collective effect. But what we are trying to do at Station Q in Santa Barbara, we are trying to solve this problem in a different way. We're trying to find a system which will be immune to these errors at the hardware level. So rather than having to correct for these errors, we would like to build a hardware which is fault tolerance. And the idea for this approach actually goes back to somebody who spent a lot of time here in Redmond, Alexi Kitaev, who is now professor at Caltech, and he said the following thing. He said, well, let's try to encode our information in varying nonlocal degrees of freedom. And the errors that we usually have, they come from some local fluctuations, local changes in the environment. But if you have -- if information is stored locally, then it will be insensitive to these local perturbation. So you can think about this in using a topologist's view. And topologists, they say, well, these three objects are actually identical. They all have just a single hole. So as long as we don't do anything drastic to the system, as long as we don't destroy this whole, the information which is encoded in this geometry will be protected. You can bite a piece of this donut and the information is still going to be preserved. And so that's the idea. You'd like to find a system which will be insensitive to these local perturbations because we encode the information in a completely different way. So when -- so usually we are talking about topological phases of matter because phases of matter were experimentally observable quantities can be insensitive to local perturbations. And one of these -- one of the well-known examples in condensed matter physics is quantum whole state. Quantum whole state is a state of a semiconductor in a strong magnetic field. So these two discoveries actually were awarded Nobel Prizes in '85 and '98 and there's some difference between them, but I don't want to go into details as far as -- this is concerned -- this is the topological phase of matter. So if you take an integer quantum whole state, which is a state, say, in lower magnetic field, and you apply electric field to the system, then there will be a current flowing in the perpendicular direction. And then if you just measure this current and find whole conductivity of the system, this is essentially the coefficient proportional to the electric field applied, what you'll find, that this coefficient will be proportional to fundamental constants, E squared over H times the integer. And this integer is accurate to ten to the minus nine. So you take a sample, different geometry, the sample can have different disorder (inaudible), different parameters, but still this quantity is protected to ten to the minus nine. In fact, this whole conductivity is a standard of resistance now. >>: And again, you're talking about the lifetime of this integer, not about -- I mean, the integer is accurate to one, not to ten minus nine. Roman Lutchyn: No, no. You measure something and something is close to one, but how close is it to one? Right? So I changed parameters in the system, and it still stays in one, whereas in some other system, nontopological, you change something, this coefficient will change completely. It's going to be zero. Or it's going to be ten. So the point is that this inter, this number close to one [indiscernible] ten to the minus nine. And it doesn't matter what I do to the system. >>: Another way to say it is if it was a billion, it wouldn't be a billion and one or ->>: I see. It's always [indiscernible]. >>: It's an integer, but it's -- yeah. >>: Yeah. It's accurate [indiscernible]. Roman Lutchyn: Yes. >>: So I'll ask a question [indiscernible]. There are the software approaches to where correction [indiscernible] quantum computing. What do you propose [indiscernible] hardware approach for that [indiscernible] uses replications, right? And since this is what happened, the state is not captured by a single particle. Instead, it is captured by a lot of repetitions. So what is the fundamental difference here? Why not [indiscernible] in software and just have so many repetitions? Roman Lutchyn: Well, I mean, if you refer to collective nature of the degrees of freedom which we are encoding information as repetition, then in this sense, yeah, there's some similarity. But you have to realize that you know, there is just a single parameter that characterizes a system which consists of, you know, ten to the ten electrons. And we don't have to control each of the electrons. It just happens that the nature does it in a particular way. Now, if I have a large system, then the number of electrons can be ten to the 23 or, you know, even larger. And I don't have to control each of them. But I can characterize the state just by a single parameter. So it's different from error correction where you control different elements separately, right? So there are similarities, and there are differences. Okay. Did I answer your question? >>: Probably need to study more physics. >>: Mm-hmm. Roman Lutchyn: So that's an example of a topological phase of matter. It's a phase of matter which the state of which is not very sensitive to local perturbations. And what you want to try to find is a state which is similar but has additional properties. It's even more sophisticated state of matter, and it's called non-Abelian topological -- this phase is called non-Abelian topological phase. So once you find such phase, and I'll give you some examples where we predict these phases can occur, we also have to be able to manipulate certain excitations in these systems. They're called quasiparticles or anyons. So these are the candidate systems. And in today's talk, I will spend some time on superconductors. And I will show you that superconductors are very interesting systems which can actually be engineered. So in some sense we are trying to engineer the state by combining various properties of different materials. So I mentioned the world non-Abelian, and this word non-Abelian is at the heart of the topological quantum computing proposal. So let me explain why this is so. When, as a freshman in the graduate school, I took quantum mechanics, I was taught that there are bosons and fermions. There are two particles: bosons and fermions. But that's true only in three-dimensional world. And the reason it's true in three dimensions comes from this argument. So we know that if you take a [indiscernible] function of the system and try to exchange two particles, then depending whether they have bosonic or fermionic statistics, the wave function will pick up a plus or minus sign. And the reason there are just these two possibilities comes to the, you know, essential to the reality of a braid group in three dimensions. So if I exchange these two particles, and then I will exchange these two particles twice, so essential I return to the same point. So I did some operation on the system. But I can deform this loop. I can essentially take advantage of this degree of freedom outside of the board, and essentially deform it to a zero, to a point. Which means I have to return to essentially the same thing, which means that there operation of exchanging T, T squared, should be actual to one. And if T squared is equal to one, the only possibility is that T is equal to plus minus one. And that's why you have only bosons and fermions. But if you take two dimensions, in two dimensions, we have more possibilities because we don't have this degree of freedom. We don't have -- if you can imagine -- if you constrain our particles to the plane, we don't have this degree of freedom. And so these exchanges are actually nontrivial. So you cannot deform it to -- this argument doesn't hold. And that was the prediction made by two Norweigian scientists, Leinaas and Myrheim, and also Frank Wilczek from the 1980s. And it has been realized later that such particles exist in these two dimensional system, quantum hole systems, and there are -- there are situations there where when you exchange, you pick up the phase which can be anything. That's why these particles in two dimensions are called anyons, which means any phase. >>: [Indiscernible] non-Abelian statistics on this [indiscernible]. Roman Lutchyn: Should be non-Abelian statistics. [Laughter]. Non-Abelian statistics is -- happens when you have the generalcy in the many body ground state. So let's imagine we have many states which have the same energy, and then such exchanges that I'm talking about here can induce certain transformations between these many body ground states. So then, I cannot just describe this operation as just a phase. It can be described by matrix. That's what -- something that we are trying to exploit here. We are trying to exploit the fact that when you have an non-Abelian statistics, then exchanging different particles between each other can be explained in terms of different matrices. And matrices do not commute, so we can make various combinations and encode information in these states. >>: [Indiscernible]. Roman Lutchyn: Yeah. So we have a -- >>: [Indiscernible]? Roman Lutchyn: Yeah. You have a state which has many particles. You know, X, N, so X one, X2 N. I'm writing [indiscernible] function which explains a state. And then I can take two particles from this state and try to exchange them together. So I can exchange X one and X three, and that will be described by a matrix M, or I can exchange X2 and X three. And that's going to be explained by matrix M. >>: [Indiscernible]. Roman Lutchyn: Because in three dimensions, as I explained, the exchange is very trivial. The braid group is trivial. So two exchanges give you just one. And so they -- the square root of one can be either plus or minus one. >>: So there is an implicit [indiscernible] similarity behind all this. You were referring to additional degree of freedom where it improved to a point. But if you puncture the plane, you cannot prove [indiscernible]. Roman Lutchyn: >>: What is the similarity that -- what do you puncture in both cases? Roman Lutchyn: somewhere. >>: Well, you have a particle, physical particle sitting Right. Roman Lutchyn: >>: Yes. And if you can go around this particle -- I see. Roman Lutchyn: -- then you can avoid the similarity. But if you don't have a degree of freedom, it's like having a plane with defect which has a line, and you cannot just go around the line. So that's the difference. That's the way to think about it. And the reason we are trying to avoid -- we can avoid the coherence here because these manipulations never involve excited state of the system. So in the previous proposal where we had quantum bits, but say conventional quantum bits, they always involved excited and ground state. So in some superposition between zero and one. And we know that, well, excited states, they decay. So they relax to a ground state. So lifetime is limit. >>: [Indiscernible]. excited state -- I can [indiscernible]. Doesn't have to be Roman Lutchyn: Well, that's not igan (phonetic) state of the still. The superposition is not an igan state of the system. >>: [Indiscernible]. Two position doesn't have to be [indiscernible]. Roman Lutchyn: Well, zero and one refers to ground state and excited state, right? And you're trying -- and you're saying it's superposition? >>: [Indiscernible]. Roman Lutchyn: Well -- >>: I mean [indiscernible]. that [indiscernible]. I'm just confused as to why you're saying Roman Lutchyn: Right. So okay, maybe I'll just draw something here. So this is a two-level system. Two-level system has a ground state and excited state. >>: [Indiscernible] what was all that? side by side too. Roman Lutchyn: >>: Yeah, right. I mean, I can write zero, one, Then it's a degenerative two-level system? If I have a superposition. Roman Lutchyn: So if you have a degenerative two-level system, it's kind of similar what I'm saying, but [indiscernible]. So the reason it doesn't work if you have a degenerate two-level system which is kind of local, because you add some small perturbation, and that's the -- this degeneracy is going to be splitted. So this degeneracy is accidental, and you change something [indiscernible] and the degeneracy is gone. >>: For example, very small magnetic field from somewhere. Roman Lutchyn: >>: Exactly. Spin up and spin down. Roman Lutchyn: So this is not the robust degeneracy, but you actually getting to the point they'll explain in the next slide. So you would like to create a degenerate system which is robust. Which is robust to these small perturbations. And you're trying to use these degeneracies, the robust degeneracies to encode the information. Okay. So that's the basic idea. That's why if you have a state which is non-Abelian state, then you do manipulations in the ground state, doesn't invoke any excited state and the information is going to be robust. And this robustness will be guaranteed by the quantum mechanics and by physics. So it turns out that the particles that the state that satisfies these properties is the state which carries Majorana fermions. So that's why my topic today was quantum computing was Majorana fermions. If you find Majorana fermions, then you'll find the state which will satisfy these properties. So what is Majorana fermion anyway? We know about fermions. Now what is Majorana fermion? What's so special about this Majorana fermion? It's an interesting object which was predicted a long time ago, in 1937 by [indiscernible] Majorana, who asked the question in the context of high energy physics whether the -- whether Dirac equation which describes essentially spin one-half -- corresponds to spin one-half particles, has to be necessarily complex. And then he showed that by choosing a particular representation, you can make equation real and therefore there should be a real solution of this equation. And if there is a real solution of the equation, then essentially the particle will be the same as its own antiparticle. And since then, there has been a lot of surge in particle physics, high energy physics, super [indiscernible] which is also related to high energy physics and also solid state systems. But as of now, we don't have a convincing evidence for the existence, and I would like to present one road to overcome this problem. And I'll present complete proposal where we can find these Majoranas? >>: I have a quick question. So take a Dirac equation and split it [indiscernible]. So it must be something more [indiscernible] on that. The first question. The second question is why [indiscernible]? Roman Lutchyn: Well, the [indiscernible] emulations, which are formulaic. And now, if you really split Dirac equation two, two, I mean, you can always split complex number into real parts. But the point is that in these Majorana presentations, there is no complex part. There's just real part. So if you count degrees of freedom ->>: [Indiscernible]. Roman Lutchyn: It's a real equation. A real equation has a real solution. So if you have real solution, then it means that the complex part is gone. But I will also get to this in the next slide. >>: [Indiscernible]. Actually in this slide. >>: Yes [indiscernible]. Roman Lutchyn: No, no, you can always -- they can have complex solutions, but if you have a real equation, you can always find a basis where there's a real solution. >>: Okay. Roman Lutchyn: If you have a complex equation, there is no basis where you find a real solution. So really, this Majorana fermion is a real -- you can think about it as a real part of Dirac fermion. And if I want -- if I want to construct a Dirac fermion, I have to take this to [indiscernible]. So our intuition actually comes from having Dirac fermions. You know, in [indiscernible], we describe everything by Dirac fermion. And this Dirac fermion that I create from two Majoranas is very special because it's not a composite object, and these two Majoranas as depicted in this picture can be far away from each other. So essentially, we created -- if you have Majoranas, you'll have a fermion will be, you know, [indiscernible] and local. And this locality helps us to make a protected systems against decoherence. Yes? >>: I have a very basic question because I'm not a physicist. state here is represented by the relative position of the two particles? Is that -- So the Roman Lutchyn: The state is represented here by the fact whether this -- it's occupied or unoccupied by a particular Dirac fermion. So I have -- I created artificially this Dirac fermion. Okay. And now I want to enumerate my Hilbert space, my states. And I can enumerate by either saying, well, this mold is unoccupied, then it's zero. Or it's occupied, which is this -- which is denoted by this, then it's kind of one. And so my logical qubit is a qubit which is either occupied or unoccupied by this Dirac fermion. And because of that, if I have many of these particles, then from 2n Majoranas, I can roughly create N qubits. And so that's the idea. You have these Majoranas on the plane, and you know, you have a Hilbert space that they correspond to, and this Hilbert space will be degenerate because these Majoranas, they are sitting at zero energy. So there's a degeneracy involved. >>: So can you explain in which sense this is [indiscernible]? Roman Lutchyn: Well it's a complex, right, object. So each of these two are real, but now if I conjugate this, I'll pick up a minus sign. So definition of Dirac fermion is that C and C conjugated are not equal to each other. And Majorana fermion, they're equal to each other. So it's essentially a half a degree of premium. So Majoranas, one Majorana is a half of Dirac fermion. It's only real part of Dirac fermion. So now I want to illustrate why exchanging these particles will actually lead to something nontrivial and why this happens in the system. So suppose now I make -- I make the situation a little bit more complicated. I add another Dirac fermion, which is localized between A three and A four. And then I can say, well my Hilbert state, my Hilbert space can be encoded in either this mode occupied or unoccupied or this mode occupied and unoccupied. And there are four states which essential enumerate this Hilbert space. And now I can do something very nontrivial because I can take a half of my fermion, which is encoded here, and exchange with another half of the fermion, which is encoded here. So normally, in other systems, you would take one electron and exchange with another electron. But here, because this electron is very nonlocal, it's actually a composite object, you can take a half of a electron in exchange for another half of electron. And that's why some -- you know, we are doing something nontrivial. And if you just check the math, you will find that you know, the state you will end up will be different. And that's basically a simple quantum operation -quantum gate operation. >>: [Indiscernible]? Roman Lutchyn: >>: Red and blue? [Indiscernible]. Roman Lutchyn: Oh, oh, there is no significance. know, just labeling which ones we are braiding. >>: It's just, you So [indiscernible]? Roman Lutchyn: Yeah. So this is the -- this is the excitement in the community because of these ideas. And the reason there is so much excitement, because we found how to borrow a trick from superconductor and use this trick to find a systems where Majorana fermions can be detected. And the idea is the following: So I'm going to give us some sort of introduction for superconductors, about you if you guys are familiar with superconductors, I can just skip it. So I'll -- should I -- >>: Do it. Roman Lutchyn: Okay. [Laughter]. So superconductors are actually interesting -- is very -- superconductivity is very interesting state of matter. It's a state of the matter where at low temperatures, once you lower the temperature, the state doesn't have any resistance. So if I measure -- if I take mercury and I lower the temperature and measure the resistance of mercury as a function of temperature, then at some point, resistance is essentially vanishing. And that was discovered by Kamerlingh Onnes in 1911. And soon after discovery, two years after discovery, he was awarded Nobel Prize. And the theory of superconductivity remained unsolved until 1957 where these three theorists, Bardeen, Cooper, and Schrieffer, came up with the explanation why this happens, why you have zero resistant state at low temperature. And the basic idea is that electrons in the metal, they kind of glue to each other and they form Cooper pairs. From two electrons, you can create the boson, quote, which is called Cooper pair. And bosons and fermions have different -- very different properties. At low temperatures, bosons want to occupy zero momentum state. They want to condense and sit in the same level, whereas fermions, they don't want to do that. And so -- and that actually is an explanation for the zero resistance. Bosons can have zero resistance because there is coherence between them. But there's another property which is important in the theory of BCS. The theory of BCS has the state -- the Hamiltonian, the effect of Hamiltonian that is described by the BCS theory mixes states with different particle number. So the particle number is not conserve. And that's actually key to having Majoranas because if the particle number is not conserved in Hamiltonian -- and I wanted to say that this theory of the superconductivity is so developed that we can even explain this cool stuff. We have a magnet on top of superconductors, this magnet can be levitated because of the Meissner effect. So we know superconductivity pretty well now. [indiscernible] superconductivity. At least conventional But I was going to say that this fact that superintendent conducting Hamiltonian doesn't conserve the particle number is actually key to the following of the ration. In order to diagonalize Hamiltonian which doesn't conserve particle number, I have to have a superposition between sort of particles and holes or electron creation and [indiscernible] operators. And the superposition has some amplitudes, UNV. And if I want to create a Majorana fermion, all I immediate to do in the superconductor is to find the state where these two amplitudes are actually the same. Well, up to conjugation. Then you can just check because [indiscernible] that you know, if U is equal to the star, then you will get the particle will be self-conjugated. So that is a Majorana fermion. So where do you find these states [indiscernible]? Well, where do we find these states with such properties? Well, we find them whenever we have an equal superposition of a particle in a hole. And where do we find such superposition? Essentially somewhere between -- somewhere between the empty and occupied band. So somewhere here. So we have to look at zero energy. And if you find the state of zero energy, such state actually, by construction, will have these properties. And the reason that it has by construction these properties actually is something fundamental in the theory of superconductivity. There is certain symmetry in the serial superconductivity which is called particle-hole symmetry. And the symmetry tells you that if you have a solution which corresponds to energy E, then there should be another solution which corresponds to energy of minus E. Now, if I have a solution with zero energy, you can just check that this symmetry guarantees that U should be equal to the star, and therefore, the state that is sitting at zero energy will be self-conjugated. Yes? >>: [Indiscernible]. one-half. I thought the rule is Fermions [indiscernible] Roman Lutchyn: Yes, that's right. This is a model for now, but I'll give you a physical realization later. I'll show you how one can enslave the spin. So basically, we are guaranteed to have the zero energy state if, as you correctly pointed out, we have spinless Fermions. Now, the zero energy state is not going to be accidental like in the other example. It's going to be protected by the symmetry. Because see, the symmetry tells us that we should have states that come in pairs, E and minus E. Now if I tried to -- if I add some I try to lift the state from zero, partner. And since it's not going violate particle hole symmetries. allowed. perturbation to the Hamiltonian and then it's not going to have a to have a partner, that should And so this -- these changes are not So unless we do something drastic to the system and we completely destroy it, weak perturbations do not live to degeneracy. So it is robust. Okay. Now, I'll give you some examples of spinless for now, but then I will show how one can engineer the spinless Hamiltonians from spinful Fermions. So why do we have these zero energy states annual where do we have these zero energy states? So there is a simple model which one can write, and it's a superconductivity model where the older parameter is proportional to momentum. So this coefficient depends on momentum and is proportional to P. And within this simple model, what I wanted to illustrate here, without any fine-tuning, you have these zero energy states. So let's imagine that we have a situation where there's chemical potential, sort of varying in space, like this. Then whenever chemical potential is positive, we have a metal. That one becomes negative, we have an insulator, because there is no fermion surface, so the Fermions cannot populate. There is no -- the dense state is zero. So if you assume this configuration, actually this configuration models one dimensional system [indiscernible] these points. Now I'll do another simplification, and I'll say that, well, this chemical potential changes very slowly in space. Okay. Then I can drop essentially higher order derivatives and what I end up with is something well known in high energy physics. It's called one plus one Dirac equation, which we are solving for zero energy. One plus one Dirac equation is very well studied and in particular, it is well known that whenever the mass of this Dirac equation which here plays the role of chemical potential, changes a sine, you'll have these states. You have some bound states. In other words, you don't need to know high energy physics. You can just say I have this equation. I want to find the normalizable solution of this differential equation, and what you will find that there are normalizable solutions at the ends. So bottom line, if I have this spinless [indiscernible] superconductor at the ends of one dimensional system, I'll have zero energy solution. And these zero energy states will be robust against small perturbations. So I don't need to fine tune any parameters like delta or new or anything. As long as I have a [indiscernible] spinless [indiscernible] superconductor, I will have these states. So that's idea. We need to somehow engineer a spinless superconductor. How can we do that when all electrons have spin? So there are two ingredients that we will use, and first ingredient is spin-or bit interaction. So spin-or bit interaction is an interaction which couples spin to momentum. And if you can somehow enslave spin to the momentum, we can get rid of this half of a degree of freedom that is redundant, and we can make the system effectively spinless. And the second thing is superconducting proximity effect. So we want to induce -- rather than looking for these materials in nature, you want to have the superconducting state on demand. So then you'll use some other superconductor and make -- and create some interfaces, and those interfaces, we will have the necessary states. So I'm be talking today about superconductor, semiconductor heterostructures. And these heterostructures will have essentially the required properties. So first, let me just spend a few minutes on spin-or bit coupling. Actually, spin-or bit coupling in condensed [indiscernible] systems appears quite generically. It appears because at the interface between different materials, you might have an electric field. A very strong electric field, in fact. And then, if the motion of electrons is, say, constrained to this plane, then if I make a, you know, a [indiscernible] transformation and go to a arrest frame of this electron, this electric field will induce a magnetic field in this rest frame will be pointing up. And if the electron has a spin, obviously it interacts with magnetic fields, so therefore, I have a coupling between the spin and electric field. And this effect is called -- I mean, this mechanism is called Rashba mechanism, and this mechanism is -- can be written in terms of -- in this form. So I will use this in the rest of my talk. And what is important for this Rashba spin-or bit coupling that it actually does the job because now you can see that the spins are enslaved to the momentum which live on different fermion surfaces. So the spins, they precess around the fermion surfaces either in clockwise or counterclockwise fashion. The second ingredient that I will need is the proximity effect. what is proximity effect? So Well, you have a semiconductor, and you put it very close to superconductor. So electrons in the semiconductor will tunnel into superconductor. And for some short time, they will feel these correlations, superconducting correlations. So they will inherit these properties from the superconductor. And effectively, it's a way to create a Hamiltonian which doesn't conserve particle number because what I can do, I can send in an electron and then measure the -- and this electron will be reflected as a hole, and as a result of this, the Cooper pair, the two electrons will be emitted in this superconductor. So if I want to describe this kind of special scattering that appears at this interface, I ever to add a term in the Hamiltonian which describes this appearance of two electrons from the semiconductor. So that's effectively a term which allows us to break particle number conservation. Okay. So that's the proposal. So suppose we first take this one-dimensional system which has the Rashba spin-or bit coupling. And we also have some magnetic field along the longitudinal direction. And so what happens in the geometry that first the -- this Rashba coupling separates the spectrum into two probable lows, but then this additional term essentially opens up a gap in the spectrum at zero momentum. So at the end of the day, if the chemical potential is somewhere here, you will have a single fermion surface. So it's going to be -- if you have a single fermion surface, which is kind of tricky because I started from a spinful system, but in the basis that diagonalizes this Hamiltonian, I will have spinless particles. I'll have single specie of particles that live on this single fermion surface. And so that's how we get around the fermion doubling. You spin degeneracy. The nanowires actually that are useful for our purposes are indium arsenide and indium antimonide nanowires. And for technical reasons, they are very favorable. They have very large couplings, very large couplings to magnetic field, to Rashba also -- Rashba [indiscernible] coupling, and they have very good contacts with metals. So now this is a rather technical slide, but I want to present the schematics, why this state, this complicated Hamiltonian is related to this spinless [indiscernible] superconductor. So you have -- so now let's imagine that we have a wire that is lying on top of the superconductor and then you'll have these additional terms to use the proximity effect. Then let's do the following trick. Let's first diagonalize this part and then we write the other part in the basis that is diagonal where the single particle part is diagonal. So what happens when the chemical potential is here, that essentially this band is unoccupied. And I can project the system. I can just keep the lowest band. And if I do that, all these terms that are here, I will throw away side plus component. So they are gone. And now what I'm left with is essentially single component for single species. Now index minus is kind of a spin index. And I have a P wave superconductor. Okay. So this construction allows us to map this spinful model to a spinless P-wave superconductor, and I showed you that spinless P-wave superconductors have zero molds at the ends. So therefore, this model should have zero molds at the ends as well. And you can see that these zero -- that these zero molds can be only if the chemical potential is somewhere in the gap. If the chemical potential is somewhere here, then this construction that I showed you doesn't work. So what we filed, we find that this system can generate Majorana fermions on demand. Okay. So I changed some parameter in the system, and I can go from a state which has these molds to a state which doesn't have these molds. Okay. And you can keep all the details, keep the full Hamiltonian and check these ideas numerically. So that's what you find. So this is the plot of the density of states is a function of the length of this wire. So this is a length. And the function of energy. And it's just numerical [indiscernible] of the Hamiltonian and then computing density of states. And you can see that you have zero energy states at the end where the magnetic field is in the right regime. And in the opposite regime, you just don't have anything. So they disappear. >>: Technical question. Roman Lutchyn: >>: What do you use [indiscernible] Hamiltonian? How we did the -- [Indiscernible]. Roman Lutchyn: Oh, it's mathematical. Mathematic [indiscernible]. You can do many things with mathematics these days? >>: Oh, yeah. Painful, but yes. Roman Lutchyn: Okay. All right. So that's basically the system that we'd like to use to build our topological quantum computer. Well, one can also say, well, what if you're theory is missing some important ingredients and how do you test your theory? Well, there are simple experiments that one can do to test this theory. I mean, the fact that you have zero energy states in one phase and you don't have it in the other phase should be seen in the experiments. And the simplest experiment is to try some kind of a tunnelling or canning tunnelling microscopy and try to tunnel at the end of this wire. If you have a state of zero energy, then the tunnelling should be allowed. If you don't have it, then the tunnelling should be forbidden. And what you should find -- actually there's a concrete prediction that if the state is in topological phase, then this tunnelling conductor, the IDV, should be equal to something very universal. Two E squared over H. And in the other limit, it should be exactly zero. So there are other experiments which also have interesting signatures of these Majorana fermions in the system, but I just wanted to maybe talk more about quantum computation in this audience. I would be happy to explain more about experiments if somebody is interested. So suppose we have this system. we do with it? Suppose we have this phase. What can Well, by the way, this is how the current experimental situation -this is where the current experimental situation. Actually, we have these nanowires and we are doing the experiments as we speak. And you know, we'll keep you posted? >>: I think they're [indiscernible]. Roman Lutchyn: >>: Yeah. Right. [Indiscernible]. Roman Lutchyn: >>: Okay. [Laughter]. Right. [Indiscernible]. Roman Lutchyn: So these are the gates that you can apply to change chemical potential in the system. And this is the wire lying on top. And beneath, you have a superconductor. >>: Just a regular S-wave. Roman Lutchyn: >>: Regular S-wave superconductor? [Indiscernible]. Roman Lutchyn: So how can we use these systems for quantum computation. So there isn't -- there is a very interesting effect. So I mentioned that if I change the chemical potential somewhere, I can go through the phase transition, right? If I change the chemical potential I can go from a state which doesn't support Majorana to a state that supports them. Right. So suppose I change the chemical potential only in half of the system. [Indiscernible] instead of essentially I move this Majorana particle somewhere here. And essential, that allows me to exchange -- to move these particles and exchange them around because if I create now a T junction, so I can move this particle here, then move this other particle here then move it back, and so on, I can do the exchange that -- I can do the braiding operation that I mentioned before. So these wires essentially allow you to do braiding operations. A network of these wires allow you to exchange the particles. Because it's essentially, you know, a two dimensional network of one dimensional wires. And one can also measure the state of the system by coupling these wires. So this is the wire and this is a set of gates. By coupling the system to a squid, so this is the squid. And you have C Joseph and junctions. And it turns out that because the state is either occupied by an electron or unoccupied, these [indiscernible] junctions, these squids are sensitive to -- and they can distinguish different states with different fermion number. So I just measure some characteristic of the squid like a splitting energy and I will be able to distinguish the difference between the state occupied and unoccupied. So I can do the redial of the state. >>: So what are those tunnel barriers? surfaces [indiscernible]? Are they just a fictional Roman Lutchyn: No. You can get them for free actually. So if you have aluminum and expose the aluminum in, you know, to the air, it will create a natural layer of oxide. And so this oxide is a barrier. And so if you basically take them out from the vacuum and then stick them together after that, and you have a junction. >>: Usually don't put them in the air [indiscernible] some chamber or [indiscernible] basically. Roman Lutchyn: >>: The basic idea is that they use oxides. [Indiscernible]. They could have done it here. Roman Lutchyn: So another thing, what you can do, you can actually create a device which is called topological quantum bus. So I mentioned that the information that is encoded in these topological degrees of freedom is very different than informations that encoded in conventional qubits where the user freedom are microscopic. So if you want to couple, say, conventional quantum architecture and topological quantum [indiscernible], we need to find a bus which can do that. And these are the ideas that allow you do that. Again, it helps that these are superconducting systems and here's, for example, a proposal to couple a semiconductor double-dot qubit where essentially the state is encoded by the presence of electron or absence on this dot. So that's one proposal. Another proposal is to couple this wire network with so-called transmon qubit, which is just a generalization of a charge qubit in superconductors. But what is interesting, that one can then couple these superconductors to photons. And photons, one can teleport, they can live for over large distances. So here's the example. You have a strip line resonator which it's basically a cavity with very good mirrors. And you can control the number of photons in this cavity by lowering the temperature. And essentially this photon couples to this transmon qubit, which is basically the qubit which consists of these two superconducting kind of islands, large islands. And then transmon qubit couples to topological qubit. So bottom line, we also can use advantage of these conventional qubits to do the logical operations which are very fast in those qubits and then teleport information into topological space and store the information there. So I'll leave with you this final slide where you envision some kind of topological quantum computer. It's basically a computer which is in the [indiscernible] but it's also connected to some classical machine and you can just dial a particular task that you want the quantum computer to do and it will do the task and spit out the result and communicate it to a classical computer. So with this, I will end. [Applause]. And thank you very much for your attention. Yes? >>: You mentioned speed. Is that -- so the topological computer tends to be slower than the [indiscernible] ways of doing -Roman Lutchyn: Well, you have to exchange the particles which tends to be slower. Also ->>: Are there other disadvantages that you're aware of? Roman Lutchyn: Also, one gate that we need to teleport from somewhere from conventional computers, we know how to make many gates with topological, but there is one gate that is missing, and we would like to teleport in gate from the conventional quantum computers. >>: Yes, conventional quantum computers as opposed to -- Roman Lutchyn: Yeah, yeah. So it's -- I understand. Quantum computing is already unconventional, but you know, for the people in the field, it's kind of conventional and topological is unconventional. >>: Sort of as a follow-up, is there a limit that one-dimensionality of it? Does that limit what you can do with it? Or can you make a grid of those wires? Roman Lutchyn: You can make a grid. There's no problem. You can make actually various networks and study the properties of these ->>: There's no problem with them over laying each other? Roman Lutchyn: >>: You can -- No. -- [indiscernible] interaction between them? >>: Yeah, the key is [indiscernible]s and surfaces and so on. And if you can manufacturer those in quantity, then you have got a lot of opportunity. That's all. Roman Lutchyn: >>: How long before you expect to get results on this? Roman Lutchyn: >>: There was another question? Well, it's not the right question to me. Well, there's an amplitude, right? Roman Lutchyn: Yeah. [Laughter]. We hope soon. I mean, they start fairly recently. They actually started maybe in the spring of this year. So it's hard to predict, but they already have some good results. I mean, they've tested the parameters that we used for example for some estimates of these nanowires and they find that they're in the same ballpark. So now they have to couple them to superconductors and characterize that part. >>: And then to adjust the -- Roman Lutchyn: Chemical potential, tune into the right regime. >>: Is this one of the experiments that it's been two years analyzing the data afterwards? Roman Lutchyn: No, no. They will -- this is not a high energy experiment. There you just measure something that comes out and analyze it next day. >>: [Indiscernible] three, four, five, or six times to be sure. Roman Lutchyn: >>: Okay. [Indiscernible]. Yeah. >>: What is the device you are using to try and observe the emergence of the [indiscernible]? Roman Lutchyn: So [indiscernible] experiment, for example? So you just have some normal metal tip which is in tunnel contact with the end of the wire. And you will tunnel and see where the electron actually has some density of states to go into or not. >>: It's hard to see [indiscernible]. Roman Lutchyn: No, no, but it's very easy to see the IDV because you're just measuring electric signal. And this electric signal will be either zero or something finite. In fact, not just finite, but something that is two E squared over H. Two times fundamental constants. So these are very concrete prediction. This is though zero temperature results. So if you have low temperatures sufficiently low temperature, which is doable. >>: This process has such high application in cryptology. Could it be, say, that NSA already have that and they just won't tell anyone? [Laughter]. Roman Lutchyn: Well, if NAS has it, then I don't know of it. Okay? But the rest companies that are very interested in these devices, like Lockheed Martin, for example. So maybe they might have it. I don't know. Yes? >>: So there's some double [indiscernible] experiments that they're looking for [indiscernible]. Roman Lutchyn: Yes. In high energy physics, people are looking for Majorana fermion in double (inaudible) decay. >>: So if they find that, that confirms that [indiscernible] can exist but it doesn't necessarily mean that [indiscernible]. Roman Lutchyn: Exactly. So here, in condensed matter, what we are predicting is actually more than just Majorana. We are predicting a state which has [indiscernible] statistics and has degeneracy. It just happens to be that these particles at zero energy, they satisfy Majorana condition. So once you find the Majorana in solid state, then it automatically should have these properties. But in high energy, it doesn't have to be the case. >>: Majoranas don't have a lot of measurement nicety for the [indiscernible] crowd any more than [indiscernible] do as a matter of fact. I mean, they just don't have a lot of properties that you can measure in a [indiscernible]. Roman Lutchyn: But this double barrel decay, I think that's the experiment that they tried to do and as far as I know, the data is inconclusive because there's not enough statistics. And they're trying to ->>: [Indiscernible]. I mean it doesn't happen that often. Roman Lutchyn: Right. They're trying to reproduce this experiment, but that costs a lot of money. Costs millions of dollars, if not more? >>: Here it's kind of interesting because the Majorana's trapped in the end of that wire. We know where it is, if it is. Roman Lutchyn: Right. >>: And so the tunnelling those where to look, I guess, is the way to say it, right? Roman Lutchyn: Yeah. >>: Get to decide whether there's a Majorana there or just the end of these pairs. You know. Roman Lutchyn: Yeah. Exactly. >>: What does that mean physically when you said the changed the chemical energy to try and kind of change the endpoint of the wire? that -Roman Lutchyn: Yeah. >>: Chemical potential. I don't know why it's called chemical potential, but it's . . . >>: What are you doing to the wire? Roman Lutchyn: So see, I can have some gate on top. change the chemical potential in the ->>: Are you adding electrons? Roman Lutchyn: >>: And I can What are you doing -- what is -- It's a capacity coupling. Okay. >>: Yeah, what you're doing is you're changing the level where the states are occupied. Roman Lutchyn: Exactly. Is >>: Right. So [indiscernible] or not, you're trying to get -- Roman Lutchyn: Trying to tune that level. >>: You're trying to get this thing at zero energy by jacking that thing up so that it's a permitted thing. It gets protected once you're there because of this [indiscernible] ->>: [Indiscernible]. This is an electrical change in the material, not like a physical ->>: Yeah, but we are talking yesterday about population changes, and I asked Roman, I said, so what are all the ways we can change the chemical potential? Roman Lutchyn: Well, let's don't go there. >>: No, but I mean, you know. >>: There's lots of ways. Roman Lutchyn: [Laughter]. Yes. >>: [Indiscernible] many suspect [indiscernible]. to do it. But the NSA -Maryland, [indiscernible] people [indiscernible] quantum mechanics. I Now, the Russians and Chinese might be able if a hundred [indiscernible] people go be noticed. Roman Lutchyn: Yeah. They actually created the institute, joint quantum institute, so they already have a lot of people there. >>: Okay. >>: And some of the people in Maryland know where [indiscernible] is too. Roman Lutchyn: >>: Okay. Let's thank Roman again. [Applause.]