#498 – Quantum Computing with Andrea Morello

Andrea Morello: This is The Amp Hour Podcast. Released June 28th, 2020. Episode 498, sponsored by Mauser Electronics. Quantum Computing with Andrea Morello.

Andrea Morello: Hi, I'm here with Andre Morello, who's a quantum computing... Good morning. Sorry, Professor Andre Morello.

Andrea Morello: Come here, Andre, it's okay.

Andrea Morello: From the University of New South Wales Quantum Computing Department.

Andrea Morello: Well, Electrical Engineering Department, really. Technically, electrical engineering. Yes, yes, yes, that's the department I'm on. And we're creating, you know, the quantum engineering of the future. So it's all blended together.

Andrea Morello: So for the benefit of my electrical engineering, electronics engineering audience, how would you explain quantum computing to electrical engineers?

Andrea Morello: All right, so electrical engineers will know that a classical computer that we use every day and that maybe some of your audience has helped developing, the microelectronics engineers in particular, are built with transistors. And when they are used for logic, they act essentially as switches that have two states, you know, a low voltage state and a high voltage state. So that's your zeros and ones in digital logic. And then you build a processor where you have, you know, a large interconnected array of nowadays billions of those transistors. And those are the chips that you use today to do classical computations. So information is encoded in the electrical state of a nanoscale transistor in silicon. It's encoded in a binary mode. Zeros and ones corresponds to lower high voltages. And then you do logic operations by having essentially the state of a transistor switching or not depending on the state of another transistor. A quantum computer is something that retains the binary logic. So it's still based upon zeros and ones. But those zeros and ones are not the high or low voltage state of a transistor, but they are one of the two quantum states of a suitable quantum mechanical object. So the simplest example one can give is that of an electron that can jump between two atoms. So in my particular research, I work with dopant atoms in silicon. Again, hopefully an electrical engineer will have done in their second year electronics some, you know, introduction to what a semiconductor device is and how it works. First you take a crystal of silicon and introduce dopants, which can be phosphorus or arsenic or antimony.

Andrea Morello: You're primarily using phosphorus though, aren't you?

Andrea Morello: I am, but also antimony for other reasons that I can go into if you're curious. Sure. So they're N-type dopants. Okay. So normally that dopant will donate, it's a donor, it will donate an electron to the conduction band of silicon. And now imagine you set up your electronic device in such a way that you have two dopants close to each other and just one electron. Right? And you could say, okay, I'm going to encode a bit of information here. I call a zero, the electron on the left, and a one, the electron on the right. Okay? It's a system that can have two options. Another possibility, which is the one that I actually work on, is to use the spin of the electron. An electron not only as a charge, but also as a spin. The spin is the fundamental microscopic magnetic dipole of elementary particles like electrons, protons, and neutrons. And so if I place this electron in a magnetic field, the spin will have two basis quantum mechanical states pointing up or pointing down. So I can call spin down the zero and spin up the one, for example. So I could make digital logic that way. But an electron is not just like a transistor. It is a genuine quantum object. So, again, think of the two atoms and one electron shared between them. That electron doesn't need to be choosing one atom or the other. It can be in a quantum superposition of being on both. Which, again, when you say it that way, people go all crazy. Oh, this counterintuitive will work quantum. This is actually completely logical. Right? It is. If you have two identical atoms and one electron and the system is completely symmetric, which atom will the electron choose?

Andrea Morello: It's going to choose either.

Andrea Morello: Both.

Andrea Morello: Both. Both. Yeah. Both.

Andrea Morello: The logical, natural answer is that it spreads out across both. Yeah. So I never let anyone get away with saying that quantum mechanics is counterintuitive. You know what I mean? It's actually completely logical. You choose both when you have equal opportunities and equal choices. So that means that you can make a quantum bit that is in the zero and one state at the same time.

Andrea Morello: Yeah.

Andrea Morello: So that's the second step. Okay.

Andrea Morello: Now this is...

Andrea Morello: That's entanglement? No. No, that's not entanglement. That's superposition.

Andrea Morello: Superposition.

Andrea Morello: Superposition. That is the next step. Yes. And that's where it gets really interesting. Again, for the benefit of our electrical engineering friends, I quite often get the question from electrical engineer and say, okay, so you have this quantum bit that can be in an arbitrary superposition of being between zero and one. Isn't that the same as an analog circuit? Right? So if I take an analog amplifier that can have an output voltage between zero and five volts, I can have any range of voltages between zero and five volts. So does that mean I've made a quantum computer? No. And to see why that is, you need to take it to the next step, which is the entanglement.

Andrea Morello: Right.

Andrea Morello: So the entanglement is a little bit more complicated, but again, it's... You have to think of the naturality of it. Okay? So now let's say that... Let me do the example. What do you think is best, the spin or the charge? Spin. Spin.

Andrea Morello: Okay, let's do the spin. I'm more familiar with spin. Good.

Andrea Morello: I think everyone else would be more familiar with spin. Fantastic. Let's do spin, which is my baby. Okay. So now let's say you have two of these electrons close to each other, right? So they have a spin that, you know, in its simple state can be up or down, but it can also be in a superposition. Okay? Um, let's say that this spin is pointing up. Right? And really, you can take the classical image that you've seen in all your little geography books when you were a kid, you know, in the middle of the magnetic field produced by the Earth that makes these lines of magnetic field like these that come out of the North Pole and wind around and get into the South Pole. Okay? So if you have a spin pointing up this way, it makes a magnetic field that goes up and then winds back down on the side. Right?

Andrea Morello: So what scale are we talking about there? Nanometers. It's nanometers. Nanometers.

Andrea Morello: Yeah. Well, I mean, the field spreads out to infinity, but it becomes infinitely small as you go away. So, you know, to have a significant effect, you need to be nanometers close. Okay. So I got a spin pointing this way up. And then I have another spin here, right? So this spin will be subjected to the magnetic field produced by the first spin. So on the side, the magnetic field is pointing down. So this spin will prefer to point this way. Because that's the lower energy state. That's the lower energy state of the two magnetically coupled spins. Right? So this is the preferred orientation for these two. But what if I turn them this way? Then it's equally enjoyable as before, right? Because this is now making a film. They're both happy. Yep. So now, what is the natural quantum state of these two spins that coupled through this magnetic interaction? Is it this one or is it that one?

Andrea Morello: They don't care. They don't care. They don't care.

Andrea Morello: They don't care. It's either. It's both. But things are now a little bit more cheeky. Because now, if I ask you, in that state where they are at the same time, like this and like that, which direction is this spin pointing?

Andrea Morello: It's going to be always opposite to the other one.

Andrea Morello: Correct.

Andrea Morello: So if you know one, you know the other. Yes. Hence why entanglement works. Yes.

Andrea Morello: Is that correct? Correct. But the point is this spin doesn't have a direction of its own anymore. So if you ask me which way is this spin pointing, the correct answer is nowhere.

Andrea Morello: Nowhere.

Andrea Morello: Nowhere. And if you actually do the calculation, it's really a simple calculation that I teach in third year to electrical engineers. You can calculate very simply what is the expected value of the spin orientation. And it's zero in every direction.

Andrea Morello: Right.

Andrea Morello: The spin has essentially evaporated.

Andrea Morello: So the number pops out as zero for all directions.

Andrea Morello: Yeah, for all directions. Right.

Andrea Morello: Right. That's why you can't know.

Andrea Morello: That's why you have something that a classical system cannot reproduce. Got it. Right. Yes. So if you now take two analog circuits and you couple them together, you will always have some voltage you can measure at the output of that circuit. Yes. Whereas here you can't.

Andrea Morello: You can't.

Andrea Morello: Right. So once you get to entanglement, that's where you really see the difference between classical, you know, continuous variables, and quantum, quantum systems. Gotcha. Now, this quantum state here where they are in the up-down and down-up state at the same time, constitutes a completely legitimate digital code for a quantum computer. Okay. So in a quantum computer with two quantum bits, I can encode four different combinations that are completely legitimate. So I can have the down-down, the up-up, the combination of down-up and up-down where they are opposite to each other. And then there's another one where they are parallel to each other, but they point nowhere in the equatorial plane.

Andrea Morello: Got it. And that's the extra over basic, like digital binary. Exactly.

Andrea Morello: So these entangled codes, and now if you want to tell me which of the four combinations is that set of two quantum bits taking, you need to give me the coefficient of each one of those four combinations. So to completely describe those two quantum bits, you need to give me four numbers. Right. You need to give me the coefficient of the down-down, the up-up, this one, and this one.

Andrea Morello: And a number is a piece of information. Yes. So you need four pieces of information to describe two quantum bits.

Andrea Morello: If I had three, you need eight. If you have four, you had 16. And so you see that the density of information contained in a set of n quantum bits is two to the power of n. Two to the power of n. Versus n in a classical computer.

Andrea Morello: Exactly. So this is why you only need, say, 300 odd bits or something as an example. 300 odd qubits we're talking about now. Is a qubit one? A qubit is one bit. Right. Is there other words, other terms for like two bits and three?

Andrea Morello: No, but people use the word q-dit for a D-dimensional system. So you were asking me before, you normally use phosphorus as the dopant where you encode the information. Actually, I also use antimony because antimony from an electrical point of view is equivalent to phosphorus. It's on the same column of the periodic table, right? But the nuclear spin of antimony has a spin 7.5, which means it has eight possible orientation of the spin of the nucleus. So that becomes an eight-dimensional quantum system. Right. So you have eight possibilities instead of two. So that's a q-dit with D equal eight, dimensions equal eight. Got it.

Andrea Morello: Is there any other advantage to that apart from information density?

Andrea Morello: Well, for quantum computing, you would... I don't know if I can call it an advantage. It's a difference. One important aspect of quantum computing is how resilient they are to noise. Right. Right. So quantum states are very fragile. We'll get into that. Why they're cryogenic. We'll get into that. Why they're... Yeah, yeah, yeah. So if you imagine having... Let's say, with an eight-dimensional spin, it's the equivalent of having three qubits, right? Because two to the three is eight. Right. So one atom of one nucleus of antimony is equivalent to three nuclei of phosphorus. Yeah. Equivalent, but it's different because the way they will be subjected to noise will be different. Okay. Imagine you have magnetic field noise, okay? So there is a fluctuating magnetic field in the environment. If you have three phosphorus atoms side by side, the magnetic field might be slightly different on each one of them. So you have noise which may be uncorrelated. Right. Whereas in that single nucleus of antimony, the noise is by definition correlated. You know, all the levels see because it's one atom, they all see the same noise. Same noise, yes. So this can be a bad thing in certain encodings. It can be a good thing in other encodings, depending on how you run it. So this is in the subtleties I probably don't want to go into, but it's... I wouldn't call it better or worse. It's different.

Andrea Morello: Okay. But for this discussion, we'll stick with the phosphorus. Let's stick with the qubits, which is a simple thing. Yes, which is a simple thing. Okay. Right. So can we think of a qubit as a storage register? Would that be an accurate... Is it a storage element?

Andrea Morello: It is with one caveat that you cannot clone the information. You cannot make a copy.

Andrea Morello: Can't make a copy. Okay.

Andrea Morello: This is a fundamental theorem of quantum mechanics called the no cloning theorem. You can transfer. So for example, I can encode a bit of quantum information on one phosphorus atom here. And if I have another phosphorus atom next to it, I can transfer the information from here to there. But once I've done that, this one is erased. There is nothing left on this. Got it.

Andrea Morello: So it's non-volatile as long as you don't touch it.

Andrea Morello: Yeah.

Andrea Morello: As long as you don't measure it.

Andrea Morello: Yeah, but you can't duplicate it.

Andrea Morello: You can't duplicate it. Yeah. Right.

Andrea Morello: Got it. You can transfer it, but you lose the original. Got it. You have one copy. Excellent. You can copy it, but you lose the original. Yeah.

Andrea Morello: So most people think, okay, a qubit is where we store the information that we're going to process in our quantum computer.

Andrea Morello: No, but you also process it on the qubit.

Andrea Morello: Yes. The processing. This is what we need to get into. Okay. Before we get into how the processing works, the actual computation, how does the quantum measurement works? Because you affect the state of it by measuring it.

Andrea Morello: Is that correct? Yes. So this is hopefully a nice example for our electrical engineering audience. So here is where we use actual transistors. Right. For the measurement. So the technology that I use is based upon using the Doe-Plant atoms as the qubits, the spin of the atom. So the readout device is actually essentially a modified MOSFET. Right. It's a small transistor that we fabricate in our clean room. It's about 50 by 100 nanometers in size. So it's small. It's not even as small as the ones you have in the chip in your camera probably, but you know, that's what we can do. Now that transistor is designed in a way that we can make it very nonlinear in its response. So it's not, it's not acting like a linear amplifier. It's, it's a switch that switches from the change in position of even a single electron in its vicinity. This is actually not as hard as it sounds. Okay. Moving one electron in the vicinity of a, you know, 50 nanometer size transistor actually has a significant effect on the bias point of that transistor. It's, it's equivalent to moving. Let me think. It's equivalent to applying, you know, some bottom millivolt on the transistor. Because you're looking at nanometer distances. It's really just an electron charge, but an electron charge at that distance has, you know, matters. And then this whole system is cooled down to near absolute zero temperatures. So it's, you know, the system is extremely sensitive. And so this transition transistor can switch from off to on by simply displacing one electron in its vicinity.

Andrea Morello: Right.

Andrea Morello: Okay. And then what we do is something that's called spin to charge conversion. Essentially, we make the displacement of the electron dependent on the orientation of the spin. So the idea is this, and that probably already answers the question that you may have had for later on, which is why do you need to go to near absolute zero temperature? And why do you need to do the things you do? So if you came to my lab, you will see some giant refrigerator that cools to 0.01 degrees above absolute zero. And you will see a rack of electronics that is full of, you know, high frequency, you know, microwave generators and, you know, very sensitive amplifiers and super quiet voltage sources. And one of the things I like the most when I explain it to the students is that you can look at this whole rack of instruments and refrigerators and everything is there, in there, is the result of ratios of constants of nature. Right. It's the Bohr magneton, Planck constant and Boltzmann constant.

Andrea Morello: Got it.

Andrea Morello: Given those numbers, you can understand why you need that rack of instruments, why you need that refrigerator. Right. Okay? So the idea is this. If you take the spin of an electron and you place it in a magnetic field of one Tesla, one Tesla is a fairly strong magnetic field. It's large, yeah. Okay? So it's, you know, so the Earth magnetic field in Sydney is about 60 microtesla, I think. I think that's about right. Something like that. So you put a one Tesla magnetic field, which we do either with a superconducting magnet or with, we're using nowadays, some small arrays of permanent magnets. If you take a strong neodymium magnet, it's actually 1.3 Tesla. Oh, nice. So it's actually about right. Okay. And so we make some little arrays and we bolt them to the coldest point in this refrigerator. So an electron spin in a one Tesla magnetic field has an energy difference between the spin down and the spin up state that is equivalent to 1.3 Kelvin.

Andrea Morello: That is why you can't have anything above 1.3 Kelvin because then the thermal noise would be higher. Because you won't see the difference. You won't see the difference. Yeah.

Andrea Morello: You won't see the difference.

Andrea Morello: Now, is this a fundamental, will this always be a fundamental limitation of quantum computers or is there some, do you guys have some grand vision to overcome this at like, will room temperature quantum computers ever happen?

Andrea Morello: The question is a little more subtle than this. So maybe let me just finish explaining how I use the transistor and then I'll tell you how you can do some other things. Okay. So you have this pin when it's down, it's in the lowest energy state. When it's up, it's 1.3 Kelvin above the lowest energy state in energy. And in frequency units. So now you divide the energy by the Planck constant that corresponds to 28 gigahertz. Right. So if you come to my lab, you will see a 40 gigahertz microwave generator because that's what we need. That's given by the Planck constant.

Andrea Morello: So you have to excite it at the frequency derived by the Planck constant.

Andrea Morello: At the frequency derived by the Planck constant. Right. Given the magnetic field.

Andrea Morello: How tight does that have to be? What tolerance on that?

Andrea Morello: Very tight. Very tight. Because these spins are extremely coherent, meaning the resolution we have on what is the frequency at which they respond is about 1 kilohertz. Oh, okay.

Andrea Morello: Right.

Andrea Morello: This is very, very sharp. And that's exactly what we want. Because the uncertainty on that frequency corresponds to an uncertainty on the quantum state as it evolves in time. It's like a clock. Yep. So you want to keep track of all the clocks you have in your system. And if the clocks start to go slow or fast, then you lost the relation between the phase of the clocks.

Andrea Morello: Right.

Andrea Morello: So then we have this, we have this pin that can be, you know, down or up. If it's up, it's 1.3 Kelvin, which is 120 microvolts for electrical engineers. Yes. 120 microvolts above the lowest energy state. And this electron is in the proximity of the transistor. And when the electron is in the high energy state, it has just enough energy to escape the atom and be sunk into the drain of the transistor.

Andrea Morello: Oh, I thought it got into the gate. No, no, no, no.

Andrea Morello: It got into the drain. The gate is isolated. Okay. The gate is isolated. Yep. So think of, you know, second year electronics transistor. You got a source and a drain. You got a silicon oxide insulator and the gate is on the top. Yep. The gate controls the potential, but is electrically insulated. Cool. And then in the body of the silicon, you got the source and the drain. This particular transistor is a little different. It's called a single electron transistor. It's got a little island of electrons between the source and the drain. That's what makes it so non-linear. But, you know, for the purpose of this discussion, we can kind of forget about it. Just imagine the electron bound to the atom, if it's in the high energy state, can escape into the drain of the transistor and just fly away. So now you have a positive charge in the vicinity of the transistor. That positive charge will shift the bias point of the transistor and make it conduct. And when it conducts, it will give us about a nano-amp of current that we can measure with a sensitivity. You can measure a nano-amp? Yeah, sure. We can measure it in real time. So you can watch in real time with your eyes the quantum state of a single spin by watching a step in the, in fact, a blip in the current through a transistor.

Andrea Morello: So you can watch it on your oscilloscope as a digital wave form, essentially. Yeah.

Andrea Morello: It's just a blip on the oscilloscope that digitizes the output of a current amplifier. Fantastic.

Andrea Morello: So you can switch one nano-amp of current, essentially, which you can measure based on the spin orientation of the electron. Yeah. So that's just a single phosphorus. That's a single phosphorus atom.

Andrea Morello: That's a single phosphorus atom.

Andrea Morello: Yeah. How does it, but you talked about pairs. Yes.

Andrea Morello: So now let's say you have two of those phosphorus atoms side by side. Okay. And they're close enough that they interact with each other. They interact with this kind of interaction. I gave you the example before of the magnetic field, the magnetic dipole field, but it's actually not how we do it. We use what's called the exchange interaction, which is what happens when the wave functions. So the probability functions of the electrons actually overlap.

Andrea Morello: Okay.

Andrea Morello: Got it. So they actually mingle with each other, but then it works the same way. You have the preferred state is the one where they are opposite. Okay. So now you can do something where the resonance frequency of one electron depends on the state of the other. Right. So if this one is pointing down, this will have a certain frequency. If this one is pointing up, it will have another frequency. So now I can do the following thing. I can put this one in a superposition state, which I do with a burst of microwaves at say 28 GHz. And then I flip this one conditional, for example, on this being spin up. Okay. So if I actually have flipped it all the way up, I flip this, and that will be the equivalent of a classical XOR gate. So they flip if they started opposite. Yeah. But what happens if this is a superposition of being up and not being up? Then this is a superposition of flipping and not flipping.

Andrea Morello: Right.

Andrea Morello: So that naturally creates that entangled state where they are just existing in relation to each other. So this is the quantum version of the classical XOR gate that's called a control knot. Yes. So it's a knot. It's an inversion of the bit conditional on the state of the other bit.

Andrea Morello: So it's like feeding the one input back in, is that...?

Andrea Morello: No, it's flipping one bit conditional on the state of the other. But because the other can be made in a superposition, then the flipping is also in a superposition of happening and not happening. Okay. And that's how you create entanglement.

Andrea Morello: You would have inherent error in such a system, wouldn't you? How do you deal with...?

Andrea Morello: Yeah, well, so the error comes from a number of things. Yeah. It comes from calibration of the classical control fields. So to make, let's say, a perfect flipping of this bit, quantum bit, you need a burst of microwaves that needs to be exactly the right frequency, the right amplitude for the right duration. Yeah? And so just calibrating that is, you know, is a task. I wouldn't call it a challenge, but it's a task. And you will calibrate it to some precision, which in our case, we have demonstrated 99.94% precision in calibrating this burst. Okay, that's very good. That's very good. That's very good. Then you have errors that come from the free evolution. So after you've done this operation, which consists of applying some bursts of microwaves, there will be some idle times during which these spins, if they've been put in a superposition, essentially precess, like gyroscopes. Right? And when you continue to do the next operation, the outcome of that operation will depend on where this spin is pointing in the equatorial plane. Okay. So if for whatever reason there is something that has slowed down or accelerated this precession, then the spin will be pointing in an incorrect direction in the equatorial plane. And that counts as an error. Got it. And so for that, we do all sorts of research and development in materials. For example, we use not the silicone that's in your computer and mobile phone, but we use a highly isotopically purified silicone. So silicone in its normal form comes with three isotopes. So silicone 28, which is the most abundant, it's about 92%, and it has zero nuclear spin. So it's a completely non-magnetic atom. Right. Then there is a 4.7% of silicone 29, which has an extra neutron in the nucleus, and that neutron has a spin. So it is slightly magnetic. Right. And that spin fluctuates in time. So it creates a random magnetic field that randomizes the precession frequency of my qubit. So it randomizes the speed of my clock, in a sense.

Andrea Morello: Why would you want to randomize this? I don't want to. Oh, okay. I don't want to. I want to avoid it.

Andrea Morello: I was going to say, right. I want to avoid it. I get from colleagues in Japan, in fact, a special, we call it an epi layer. It's a micron thick extra layer of silicone grown on top of a silicone, normal silicone wafer, that has been grown using isotopically purified silicone 28. So there is almost no silicone 29 nuclear spin in the vicinity. Got it. So that gets rid of the magnetic noise.

Andrea Morello: Don't they use that for the kilogram that you are, the sphere, the silicone sphere? Yeah, yeah, yeah.

Andrea Morello: Exactly.

Andrea Morello: Same stuff. Same stuff, yeah. Oh, okay. Yeah. Right. How do they build that up, that sphere up, if you're saying it's coming in one micron layers? How do they build that?

Andrea Morello: Oh, that's not done that way. Oh, okay. So that's grown in a different way. Oh, okay. That's grown in big lumps, lumps this big, and then they chop it, and then they kind of shave it off. Oh, okay. So most of the experiments I've done are from material that doesn't come from that avogadro sphere. Right. But I actually have a little piece. Oh, excellent, excellent. That comes from the sphere. You've got the prototype, yeah. Okay, so you just take the chips off it, and then you go, yeah. Yeah, yeah. But there's not a lot of it. Right, okay. It's a rare thing.

Andrea Morello: Yes. Okay. Right, so you need to use that, otherwise your condom computer's just going to be a mess.

Andrea Morello: It's going to be much more noisy. Right. The point is, you can tolerate some noise. Okay, so this is probably the most important realization in the history of quantum information, is that you have the ability to do quantum error correction. Okay, so if you read the papers on quantum computing in the early 1990s, there were a lot of very eminent luminaries, who were really kind of laughing at the idea of a quantum computer, because they said, oh, come on, this thing has no protection from noise. And it doesn't latch. Ah, okay. It doesn't latch. Yeah. I never met him in person, but there is a really eminent scientist called Rolf Landauer. He was working at IBM in the US, and he's one of the fathers of both classical and quantum information science. And there is this legend that he once called in his office, a colleague who was working on quantum computing, called him in his office, and he slammed the door closed, and he said, you see, this is why quantum computers will never work, because they don't latch like a door does.

Andrea Morello: A door does. Right.

Andrea Morello: Except he was wrong. Right. It doesn't matter. It doesn't matter. It doesn't matter. Right. Because you can do quantum error correction. You can actually correct some fraction of errors if it's below a certain threshold. There are ways to encode quantum information where you use multiple qubits to encode one, what's called a logical qubit. Aha. This is... And that gives you some tolerance to errors. Got it. So when you ask how accurate is your qubit, how protected from noise is it, I mean, we are all working as hard as we can to reduce these errors as low as we can, but we don't need to get to zero. We have some slack.

Andrea Morello: Some slack. Right. This will lead into a question later. Hopefully I don't forget about why we're not near to decoding, using quantum computers to decode cryptography. Yeah. So, yeah, hopefully we'll get on it. We won't sidetrack that now. Mm-hmm. So, does that error correction happen at the computational level or does it happen at the qubit?

Andrea Morello: Hardware level. Both.

Andrea Morello: Okay.

Andrea Morello: Both. So this is a very interesting question you asked. There is... So the error correction is built into the quantum hardware. So the simplest quantum error correcting code you could make uses simply redundance. Mm-hmm. So you could encode a zero into three qubits that are down, down, down, and then one into three qubits that are up, up, up. Okay? And then you say, okay, you're gonna look at these qubits and if there is up, up, down, you'll say, well, there was probably an up, up, up, and then this qubit flipped by mistake. Okay?

Andrea Morello: Got it. So that's a CR set, that's like a built-in hardware quantum error correction through redundancy, kind of?

Andrea Morello: Yes, but of course the clever listener will say, but wait a minute, you can't just measure the qubits because it will destroy the quantum information. Yeah. That's encoded in them. So you need to be a little clever in the way you detect the errors. So you don't actually measure the actual state of each of the spins, but you do often something that's called a parity measurement. So you can find out if you have an even or an odd number of qubits up.

Andrea Morello: Everyone's familiar with parity, error correction. Yep.

Andrea Morello: Same thing. So you can find out actually destroying the quantum information that's on it. And you do this repeatedly as you run your computation. And this will generate a lot of data that then a classical computer somewhere up at room temperature will have to keep track of and manipulate and then understand in order to correct the errors. So quantum error correction schemes involve both a quantum hardware encoding and a classical computing decoding and correction. So in fact, it's challenging both the quantum and the classical. If you look at the computational requirements on a classical computer to keep track of the error correction necessary to run a quantum computer, it's really, really large. So it's, you know, we need to, you know, we'll never, we'll always need to be good friends with our classical computing colleagues because we need to do this all together.

Andrea Morello: That's probably a question for the end is quantum computers are not a replacement for classical computers. No. They never will be? No. Is that a claim you can? Right. Okay. I don't think they ever will be.

Andrea Morello: So the qubits can store information. We can store information in qubits. We can store more information than what's in the universe, potentially in a handful of qubits. So how do we construct, how do you physically construct the CNOT gates? What is the architecture? What's the computing architecture? How do you construct the computational aspect of it? Is it like a FPGA matrix? Do you have like the individual quantum bits are storing the information, then you have the computational fabric around it? What is the architecture?

Andrea Morello: So there are many, this is still essentially work in progress, right? So I don't think there is a universal agreement on what is the way to do it. But the most popular one involves a two dimensional array of qubits, some of which are essentially data qubits and some others are parity measurement qubits. So you have to imagine this two dimensional array where you have interspersed data qubits and measurement qubits. And so you will do the operations by applying, you know, pulses of microwaves and with certain timings and also by controlling which pairs of qubits interact with each other. So this, depending on the physical implementation you are using to encode the qubits, will be done by, for example, controlling some voltages on some gates that you may have in an electronic circuit, or it may be done by changing the flux through a little loop of superconducting material with two tunnel junction that couples to superconducting qubits. So, you know, depending on the details, there will be different ways to turn the coupling on and off. And then there will be ways to, you know, flip the spins from zero to one or to some superposition. And then there will be measurements on the measurement qubits. But broadly speaking, you have to imagine some two dimensional grid of qubits with data and measurement qubits interspersed where you can switch interactions and you can do operations and measurements in a clocked way.

Andrea Morello: Okay. In a clocked way, do you mean that they're all parallel? They're all clocked in parallel, essentially? This is one of the benefits of quantum computing, obviously, is to do processing in parallel. That's right. I didn't expect there to be one best way to do the computation. The architecture is probably still being... So the physics of the quantum computer and spin and all that is pretty much nailed?

Andrea Morello: Yeah, the quantum gates are fairly well understood now. Of course, people are still working on sort of out of field options. We have these topological qubits that have a built-in error correction. They basically have... Well, they're supposed to have... They're expected to have some intrinsic resilience from noise, which means you need to do less steps of error correction to keep the computation going. One thing that's worth noting is that... So depending on how you lay out the architecture and the operations, you will have a different threshold for quantum error correction. So the most famous architecture is called the surface code, which is the one I was telling you about, this two-dimensional arrays of sort of interlaced measurement and data qubits. It gives you a 1%, about 1% tolerance on errors, which is a lot, right? Okay. It's quite a comforting number. But at 1%, it means that you have essentially an infinite number of physical qubits encoding one error-corrected logical qubit. So you want to go below the threshold. And the further below you go, the less overhead you have to implement the quantum error correction. So if you are a factor 10 below the threshold, you will probably need hundreds of physical qubits to encode one error-protected qubit. But if you go a factor 100 or 1,000 below the threshold, then you only need a handful. So you're more resource efficient. So it's always a good idea to have as low errors as possible.

Andrea Morello: Here in the Amp Hour, we look for sponsors that help our listeners learn. Today we're talking with Paul Galata, a senior technology specialist from Mauser Electronics. Paul has been creating content around fog computing. I asked him to disambiguate terms like edge, AI, and fog.

Chris Gammell: The edge acts as kind of a gateway for IoT devices for these sensors for these connections to be able to translate information that they have and be able to send them somewhere. AI includes things such as inferencing. Inferencing is the ability to step through a program or a routine and then make a decision and adjust that decision. FOG is really the edge plus things like AI, kind of decision making and computational type of resources. And when they go up, they go into this FOG and combine and collect and collate and correspond and then make some decisions, some inferences, and get it right back out where it's supposed to be.

Andrea Morello: So what is an example of a system that is using edge computing versus FOG computing?

Chris Gammell: So it's very common to have an edge at a factory. So you can imagine a manufacturing process. Let's just pretend I got a water bottle in front of me. If something went awry, if caps were not in position to, you know, be sealed and ultimately they weren't going to be able to put the caps on, right? A decision could be made by the machines at the edge of what to do, where they would send a recovery signal or an error signal or some type of information localized to that premise, to whatever's going on, to take an appropriate action for whatever the situation might be.

Andrea Morello: Okay. So then what is the difference between that and FOG?

Chris Gammell: The FOG is one step away from the edges or the devices where the AI, the inferencing, the decision making, the processing is happening at a level. So right at the water bottle capping station, it might not know exactly what to do. It just knows I have an error. I have a situation. I have something going on. But it steps a step up away from the edge from whatever's going on locally. And it goes into the control room where a bunch of engineers or production people are operating. And it says, because of this, this, this, I have this many caps left. I have this much water. I have to make this many more shipments. This is what I should do. And again, because these decisions emulate human decisions because of the AI, it uses this process of inferencing, of making adjustable decisions. It emulates humans by adjusting itself towards the goal.

Andrea Morello: I knew that online distributors like Mauser hosted application content, but I was interested that there was higher level content like this. So I asked Paul, why are they working on these high level topics in the first place?

Chris Gammell: We are trying to bring together great ideas about these products and technologies and what's going on into one place and give people the opportunity then to move more specifically. We recognize that we might have 10,000 or 100,000 or a million parts that support this. It's very hard to do people there before having a conversation about these are the important things. Now you can go, you know, north, south, east or west in your search for whatever to find what, what works for you.

Andrea Morello: For more information about fog computing and to see the range of other high level resources that are available for Mauser Electronics, check out the amp hour dot com slash fog dash computing. And now back to the show.

Andrea Morello: There's probably a handful of applications for quantum computers. They fix they're not a replacement for general purposes, like a dozen.

Andrea Morello: There's a handful of application we know of, you know of, yes. And I want to make this point really clear. It's hard to imagine applications for something that doesn't exist. Yes.

Andrea Morello: Right.

Andrea Morello: I remember years ago, I was fortunate enough to meet Charlie Towns, who is the guy who got a Nobel Prize for inventing the laser. He was in his 90s at the time, but he still came to the conferences. And one day I saw him there at the lunch buffet and I sat next to him. Oh, yeah. And, you know, very nice guy, very friendly old man. And I asked him, so when you, you know, invented the laser, what did you imagine that it would be used, you know, to cut frozen chicken, to read data from a DVD and play a movie or to, you know, correct eyesight? I said, well, of course not. It was just a curiosity. How could he? Yeah. Right. So I want to make this absolutely clear before we talk about applications.

Andrea Morello: Yeah.

Andrea Morello: So the main, and also about sort of short term quantum computer prototype, the main role of near term quantum computers is to give us a playground to understand what to do with real quantum computers. Right. Right. Got it. There is no quantum computer in the world right now that does a useful computation that a classical computer cannot do. Got it. Okay. There has been a big result from Google last year when they showed what they call quantum supremacy, which means the execution of an algorithm that would be really intractable by even the most powerful classical supercomputer. And that's a genuine result. It's a real breakthrough. But that calculation that was executed is not a useful calculation as such. It was just to prove.

Andrea Morello: It was just to prove. Quantum supremacy exists. Exist.

Andrea Morello: Yeah. But the important point, and the Google people are very explicit and honest about it, is that having a machine like that in your hands is what you need for quantum software developers to learn what you can do with a quantum computer once you have one. It's just really hard to write code for a computer that you don't have.

Andrea Morello: Exactly.

Andrea Morello: So it's actually, I find it remarkable and almost a miracle of human intellect that we do have quantum algorithms that people have cooked up in their head without actually having a computer to run it on.

Andrea Morello: Yeah, yeah. Well, that's a, well, there's a whole history of that in computing. People writing simulating. Yeah. So can we quickly talk about, because that brings up D-Wave, for example. A lot of people say that's not really a quantum computer.

Andrea Morello: It's a quantum annealer, but again, for D-Wave, I will say in their defense, a lot of quantum algorithms have been invented and developed just by the sheer existence of the D-Wave machine. Right. Now, the results that those calculation yield are not results that you couldn't have achieved using a classical computer.

Andrea Morello: So not quantum supremacy.

Andrea Morello: But just by having the machine there, a lot of clever people have had the opportunity to develop quantum algorithms that own a more powerful machine will then actually be useful.

Andrea Morello: Got it. Okay. So it's more of a development platform really than any, producing any useful...

Andrea Morello: At this point it is, yeah. Okay. I mean, they all are. Right, yes.

Andrea Morello: They all are. Yeah, we don't really have a quantum computer that's doing useful work, really.

Andrea Morello: No.

Andrea Morello: No. Okay. So we've got quantum, quantum supremacy has been proven. What about classical supremacy? Are there, for want of a better term, is that a term? I don't know. Where there are things that can be done on a classical computer that will never be able to be performed on a quantum computer? Well, okay.

Andrea Morello: So first of all, there are theorems that show that any classically computable function can be computed on a quantum computer. Oh, okay. It's just that you would never do it. It's just... It's like taking a Boeing 747 to go and buy a loaf of bread at the shop up there, you know? Okay. You could, but would you? Yeah. Right. There's one really interesting thing that's happening in the last couple of years. They call it de-quantizing quantum algorithms. So there's some... Okay. So there's some clever people who invent quantum algorithms that are at face value superior to the known classical algorithms that are known to exist. And then the classical computer scientists de-quantize it, meaning they find the better classical algorithm and they often take inspiration from the quantum one. Ah, interesting. So the ideas and the insights of the quantum software developers inspire classical software developers to come up with a new algorithm that runs on a classical computer that they probably otherwise would not have come up with.

Andrea Morello: And inspire them to think outside their box. Yeah. So, ah, that's interesting. Okay. So, I think you covered it there. There is no such thing as classical supremacy where you said a quantum computer could in theory do anything a classical computer could do.

Andrea Morello: Yeah, but it would normally do it much slower. Right. So, for example, one thing to keep in mind is clock speed. Okay? That's something we haven't talked about. You know, any classical microchip runs at a gigahertz or two or three nowadays, you buy them for a few dollars from the shop. Quantum computers have a clock speed that depends on the physical details of the hardware that is being chosen. But the clock speed rarely exceeds a few tens of megahertz. Okay. So even the fastest ones. Fastest ones. Yeah. Okay. Even the fastest ones. Fastest ones rarely go past some tens, maybe a hundred megahertz.

Andrea Morello: Can you see a future where that scales higher or are there sort of fundamental? Uh, at the moment anyway, you can't say in a hundred years we're not going to do it.

Andrea Morello: You can't say, but at the moment there aren't. The thing is that if you try and go faster, the errors go up. Okay.

Andrea Morello: So what is the speed affected by is the computational clock speed affected by the, you talked about the logical elements. So you could have logical. Sorry. A logical qubit can be made up of, you know, a hundred physical qubits. Does that affect the speed?

Andrea Morello: Well, so the logical, the way in which you build the logical qubit will eat up hardware resources, but the clock speed is the clock speed of the basic elements. To answer your question, in a sense, if you think of a fundamental speed limit, it usually boils down to the actual energy difference of the logical quantum state of the qubit. So pretty much most of the useful qubits that we know of, with some exceptions, most of them work in the gigahertz range of precession frequency. And then when you think about how you operate them, you want an interaction between them that is a small fraction of their energy difference. Otherwise, it's not qubits anymore. They become like a blob. So if you run this thing as 10 gigahertz, you want it to couple to the other system by 100 megahertz, maybe a gigahertz maximum. And that is fundamentally the speed at which you do things.

Andrea Morello: Okay. So there appears to be a fund for all intents and purposes. There's a limit there.

Andrea Morello: Yeah. There are other systems like there are atomic clocks, there are optical clocks. They work at, of course, you know, tens of terahertz.

Andrea Morello: Yeah, terahertz. Yeah.

Andrea Morello: But then making them interact with each other is not easy. So it's, yeah. But, you know, this is part of the beauty of quantum technologies. We have some, you know, leading platforms that are well developed, they're having great results. But we haven't, you know, we don't have the equivalent of the CMOS transistor in quantum computing yet, right? There is still a lot of platforms, each one with its pros and cons.

Andrea Morello: Is that kind of the aim that you want? You want this element that you can, because the way they produce CPUs these days is they have these elements, they can just drop them in and they just work.

Andrea Morello: Yeah. But then again, I'll respond to this by saying that this is the way it was until maybe 10 years ago. Now you're getting a lot more of those application specific hardware, right? So the classic CPU that you always use for everything is not really the way it's done anymore. I mean, it's still there, but if you really want to push, now you get the GPU for this, you get some mathematical.

Andrea Morello: Asics everywhere. They're very expensive, but they can, they're more energy efficient, they're faster, they can do everything better.

Andrea Morello: So we're kind of going out into the specialization there as well. And for quantum computing, for all you know, it might always stay that way. There might be certain types of hardware that are most suited for certain kinds of simulation. For example, the things that D-Wave makes, the quantum annealers, they might remain the preferred platform for certain optimization problem. That's what they naturally do. Whereas for some other kinds of calculations, you just need a different system, which might be built in a completely different hardware.

Andrea Morello: Will there, will quantum computers be like general purpose computers or will they be like application specific? As you said, are they better to design?

Andrea Morello: It's hard to tell. So people like myself and many colleagues around the world work on what we call a universal digital quantum computer. So we, our goal is to make the equivalent of your PC, you know, you just, you can program it and it does whatever you want. This, if it ever happens, will be a very long term goal. I think for the next couple of decades, we will have ASICs. Right. That's the session specific.

Andrea Morello: Okay. Right. They're application specific. So there's other teams, they think, oh no, look, ASIC is where we're just going to focus on the ASIC type stuff. Yeah. Right. So, so you've got to essentially program, you've got to build that silly. So if it's out of silicon, you have to build that silicon for that specific task to solve prime number crunching to solve, uh, some specific task you're doing, you're modeling how molecules work and things like that. Put it this way.

Andrea Morello: The reason I'm working on silicon is because it seems to me and to a lot of people, the, one of the most plausible platforms for universal quantum computers. Okay. The ones that will get way down the track, the ones that will be really general purpose and really have a broad, deep impact. If I wanted to make a medium term application specific device, I may or may not have chosen silicon. Okay. Right. Silicon has, well, okay. Silicon from a physics point of view has some interesting advantages. such as the time it can hold the quantum information in it because of this purification. Yes. As a purification.

Andrea Morello: How long are we talking about?

Andrea Morello: Uh, for the electron, we're coming near one second. For the nucleus, we have shown 35 seconds a couple of years ago in my group.

Andrea Morello: Okay. Yeah. I expected much longer. No, no, no. Okay. So it's got to do the computation in that time on that data.

Andrea Morello: Yeah. Yeah. I mean, we, okay. So the electron runs at a hundred nanosecond, a few hundred nanosecond clock. And it's got a one second lifetime. Right.

Andrea Morello: So yeah. Okay.

Andrea Morello: Right. So you've got to.

Andrea Morello: The nucleus runs a bit slower. So you've got to program it with the information and then process that information within essentially. Yeah.

Andrea Morello: You don't need to get to the end of the computation in one second. Oh, okay. No, because as you run it, you can correct for little errors. Right. So you've got this clock at, let's say a micro, let's say a megahertz clock. And so in a second, you can do a thousand clock cycles. And among those clock cycles, there will be cycles that do error detection and correction. So as long as you're processing.

Andrea Morello: Yeah.

Andrea Morello: As long as you're doing it, it just keeps your life.

Andrea Morello: Okay. So is it the accumulation of errors that would cause, if you do nothing with it, if you're programming in the information.

Andrea Morello: If you do nothing, then you've got one second to go. It just totally dissipates.

Andrea Morello: Yeah.

Andrea Morello: But if you run the quantum error correction, then it keeps it up.

Andrea Morello: It's like props. So why does it essentially, why does the information in there essentially, why does it dissipate?

Andrea Morello: Well, so that, it's actually not, it's not exactly dissipation. Dissipation. I was looking for a better term. Yeah. Dissipation is the word you use when it's dissipated in energy loss. Energy loss. And actually our qubits have excellent energy loss. Actually, non-loss. Non-loss.

Andrea Morello: Okay.

Andrea Morello: Right. So the electron actually is about five, ten seconds. The nucleus is literally the age of the universe. So the energy loss of the nucleus is unmeasurable. If you ask me how much it is, I don't know. I never had the patience to measure it. Okay. But what it's called, it's de-phasing. It's like clock desynchronization. Okay. So imagine these qubits as little clocks. And the quantum information is capping something that you may think resembles the, the relation between clocks. You know, like when you go to those old, well they don't do it anymore. You know, those airports or those offices in the multinationals. Well, yeah, they have all the clocks with all the time in all their main offices around the world. You know, imagine those clocks don't all run at the same speed.

Andrea Morello: They're all drifting.

Andrea Morello: At some point you just don't know what time it is anywhere.

Andrea Morello: Right? And you can't correct it. At some point you lose the ability to correct it.

Andrea Morello: Yeah. Okay. But if you correct and check often enough, there are some theorems that show you that if you do it well enough and often enough and the desynchronization is slow enough, you can actually keep track of it.

Andrea Morello: Fascinating. Okay. I assume that they stayed there for, that's interesting. No, no, no.

Andrea Morello: They don't need to. That's the beauty. They don't need to. This is for our engineer friends. You know, there is, it is a genuine engineering problem. There are tolerances like in any engineering design. The tolerance is not zero. It's finite.

Andrea Morello: That's on regular silicon CPUs as well. The tolerance is there, you know, you get cosmic ray impacts, you get other, you get electron migration and you get all sorts of other issues involved.

Andrea Morello: Yeah. Okay. And you design the whole, both the hardware and the software that runs it in a way that is capable of, detecting and correcting these problems. Now going back to silicon. So I just wanted to conclude this thought. Silicon has some advantages from the point of view of, you know, basic research of someone like me who works in a university. So, um, it does pose some challenges because, um, you know, silicon nanoelectronics is expensive. It's complex. But keep in mind that if you look at the, the devices that I make or my students make more precisely in the, in the university labs, they are essentially the artisan version of silicon MOSFETs with individual dopants next to it. Okay. If you put that side by side to the chip that's in your camera over there, I mean, that chip is so much more sophisticated because it's been made in a billion dollar foundry. And I don't have, but it's there. Exactly. It's there. And it doesn't need to be reinvented. Right. Right. That technology exists. So what we are looking for is my job is to do the basic science and a little bit of the basic engineering that will then allow me to translate this, uh, you know, design, this quantum design rules that I've developed in this work into manufacturing design rules that can be put into place. Right. Using facilities that are not too dissimilar or, you know, in my dreams, identical to the ones that are used to make the classical chips.

Andrea Morello: In, in practice though, I don't think you've reached that. They would have to modify that. There would be some modification. There would be some special processes.

Andrea Morello: There would be some special processes. But the, the, the goal is to not have to completely, you know, retool an entire $10 billion fund. Right.

Andrea Morello: Got it. So are there any other, um, uh, silicon, like silicon on sapphire and other sort of exotic processes? Do you benefit from? No.

Andrea Morello: So there is some success people have found, especially the ones who have access to, you know, proper foundry with the silicon on insulator. Okay. Yeah. Uh, fully depleted silicon on insulator. FinFETs. Okay. So there's some really tiny, essentially silicon nanowires with the wraparound gates. Yep. Those have some very interesting properties that, uh, that can be useful for, for quantum computing that can confine electrons very tightly in the corners of the gates.

Andrea Morello: Okay. Because at the moment you want, you've got to physically deposit the one phosphorus atom within the, essentially within the transistor. Yeah.

Andrea Morello: Yeah. Yeah. Yeah. So that's the standard method by which dopants are introduced in any, in any chip. What we, and in particular, I want to acknowledge my colleagues at the university of Melbourne, they are the ones who do the ion implantation. What they do is to, um, they have an ion implanter and they've integrated that with a atomic force microscope with a tiny little nanometer hole in the tip. Right. It's a very nice design. So we make in Sydney the, the, the silicon device and we can put some alignment marks on it that can be seen. And then the atomic force microscope is a super sharp tip that can actually see if you have an atomically flat image, you can see the individual atoms. Wow. But at the scale, even just seeing nanometer, you know, you can hover it over the surface and you can see our alignment marks and where we put all the, all the circuitry. So you can, they get an image of the, of the circuit and then you can move it with nanometer precision to where you want the atom to go. And this tip acts like a mask. It's like a movable mask and it's got a nanometer hole drilled in it. So you start spraying ions with the ion implanter and then there will be one that goes through the hole. And when it goes through the hole, then we at UNSW make the, um, on chip ion detector, which is again, a modified version of a, of a radiation detector, which when a high energy particle hits, creates a thousand electron hole pairs, they get accelerated and collected by a very sensitive electronic circuit that tells us, boom, one atom has gone in. You blank the beam and you got one atom right there. Wow. You see, again, this is not, it's a modification, but it's not a major modification from the way in which normal chips are made.

Andrea Morello: Is the University of New South Wales doing research on any other types of, well, can you go through the different types of you? You're doing silicon, you're doing phosphorus on silicon primarily. Yeah. What are the other methods that other research teams around the world are doing?

Andrea Morello: So in, at UNSW, we have three ways to do it. Mm-hmm. So I use ion implanted dopants. There is, uh, my colleague, Michelle Simmons, who makes also, uh, dopants as qubits, but she puts them on the chip by a scanning tunneling microscope method. Right. So this is a case where you take a atomically flat silicon surface in ultra high vacuum, you deposit a layer of hydrogen. Mm-hmm. And then with a scanning tunneling microscope, you remove the hydrogen where you eventually want the phosphorus to go. Then you introduce phosphine gas, which is pH three, and by some miracle of surface chemistry, the phosphorus sticks to the surface. Oh, it's just magically. It just magically happens.

Andrea Morello: And the rest just vanishes.

Andrea Morello: Yeah. Oh, wow. And so then you get phosphorus where you want it. And that does not only the, um, the, the qubit, but also the classical circuitry is all made by highly doped phosphorus. Mm-hmm. And then you encapsulate with another layer of silicon. Yeah?

Andrea Morello: Interesting.

Andrea Morello: So that's another way to make dopant-based quantum bits with a different fabrication process, which has a higher positional precision, but it's very different from the way classical computers are made. Right. And then there's my colleague, Andrew Zurach, who makes a really, he doesn't use dopants as the qubits. He uses electrons kept inside the nanoscale transistor. So he makes quantum dots, what they're called. Yeah. They are essentially the super shrunk version of a silicon transistor, shrunk so much that it holds just one electron. Okay. So what he does is arguably the most, you know, CMOS similar, CMOS compatible device you can make. It's really just a super small modified silicon array of transistors. Mm-hmm. And all these three methods have their pros and cons, and we're all making progress. And, and there's a lot of synergy between, uh, between the activities.

Andrea Morello: What about non-silicon methods? Right.

Andrea Morello: So non-silicon methods, um, some of the two probably most, uh, developed and successful ones at this moment are superconducting circuits. Mm-hmm. So that's what Google does. That's what IBM does. And that's what a number of other, also smaller, uh, groups and companies around the world do. So here you have a, um, a circuit that's made from a film of superconducting material. And on this film, you make essentially what is, uh, it's a quantized nonlinear oscillator. So imagine you make an LC oscillator, just an LC circuit, right? Yeah. It will oscillate at a certain frequency. So if you just take a capacitor and an inductor and you put it here on your breadboard. It's a tank resonance circuit.

Andrea Morello: It's a tank resonance circuit.

Andrea Morello: And at room temperature, you will have, you know, billions of microwave photons in it. Yeah. So imagine this circuit resonates at 10 gigahertz. 10 gigahertz is the equivalent of half a Kelvin in temperature. Okay. Right? So if you now cool down this tank circuit to 20 milli Kelvin, how many photons are in there?

Andrea Morello: Zero. Are we talking?

Andrea Morello: Essentially zero. So it's an L, it's an electrical circuit. It's an LC oscillator in its quantum mechanical ground state.

Andrea Morello: Wow. Okay.

Andrea Morello: And then you can introduce one photon in there and it goes in its quantum mechanical excited state. So that's your zero and the one. Yep. The problem is you can't just make a simple LC oscillator because then, uh, all the energy levels are aque-spaced. So, so you don't have only two levels. You have one, two, three, four, you can put as many photons as you want in there.

Andrea Morello: Right up to room temperature.

Andrea Morello: Right up to room temperature and they're all aque-spaced. Okay. So you need something that is non-linear. And this is done with a superconducting circuit trick. It's called the Josephson junction. It gives you essentially a non-linear inductor.

Andrea Morello: Got it. That's what they use for the voltage standard, the Josephine junction. That's right. Right.

Andrea Morello: That's right. Okay. So you lay it out in a clever way and you get what is essentially a non-linear LC oscillator. That has the too low a state with a certain energy difference. And the third state has a different energy gap. So if you apply a microwave photon at this energy, it doesn't then jump up the other level. You know, there's only... Got it. And then, so again, for our electrical engineering friends, these are really just quantized electronic circuits, basically. And the way they talk to each other is by either direct capacitive coupling, right? So the capacitances are talking to each other in a Coulomb way. Or sometimes they talk to each other by sharing the microwave photon across a resonator. So you can make, you know, if you make a 10 GHz half wavelength resonator, it will be about a centimeter long or something. So now you can imagine putting two of these little LC oscillators, nonlinear LC oscillator, at the opposite ends of this centimeter long resonator. And these two guys can share a microwave photon, which acts as the coupler for them.

Andrea Morello: Oh, that's interesting.

Andrea Morello: So you can entangle electronic circuits at, you know, centimeter distance. At centimeter distances? A centimeter distance via this microwave photon that is shared through the resonator.

Andrea Morello: Oh, are there practical applications for that?

Andrea Morello: Well, that's how the most developed quantum computers are built.

Andrea Morello: Okay. Is that like D-Wave? How does D-Wave do it?

Andrea Morello: D-Wave does it in a different... So the example I gave you is what Google and IBM and some others are using. D-Wave does something different. D-Wave uses what's called a flux qubit. So it's also a superconducting circuit, but you have to imagine it's a loop. It's a loop of superconductor. A physical wire?

Andrea Morello: A physical wire, a physical loop. Yeah.

Andrea Morello: Now, in a superconductor, you can have current that flows without dissipation.

Andrea Morello: It's... yeah, it just flows forever.

Andrea Morello: So now you can imagine you could encode a zero or a one in the current flowing clockwise and the current flowing counterclockwise. You create some magnetic flux, and in fact, that flux is quantized. It's called the flux quantum. But so if you put a Josephson junction, in fact, three Josephson junctions across this loop, you can make a qubit where the current is in a superposition of flowing clockwise and counterclockwise. Right. This is another one of the things I love to teach to the students. You know, you kind of take them by the hand, and you try to tell them, quantum isn't weird, don't worry about it. You know, and then when you really do it in a lecture style, you come to this inevitable conclusion that there is a current, which is a sizable current, like microamps, right?

Andrea Morello: Yeah.

Andrea Morello: A current that is in a superposition of flowing clockwise and counterclockwise, which is not the same as saying there is zero current. Right, no. It's a superposition of flowing one way and flowing the other way.

Andrea Morello: And how do they read those?

Andrea Morello: They read them by essentially making it switch to one side or the other. Okay. Right, so they can tell whether their current flows clockwise or counterclockwise. And so if you have a superposition, it will project it into one way or the other. It's the same with the spin. If I prepare the spin in a superposition of up and down, and I go and read it, I will project it into either up or down. So I will half of the time get the spin in the up state, which escapes, and half of the time the spin in the down state that does not escape.

Andrea Morello: This is why, and we probably have to get into this now, is why quantum computers, please correct me if I'm wrong, I'm sure I will be, that, okay, you've got, you know, you're processing all this information, but quantum computers are really only useful if you get like a single output or a limited output. Is that correct?

Andrea Morello: Yeah, yes. Yes. So that is probably the, the simplest way to understand why is it so hard to make useful quantum algorithms, right? Because, so you have this, this exponential density of information you can encode in the quantum computer, but you can't get it out.

Andrea Morello: You can only get it out in a limited form. Yes.

Andrea Morello: So what you need to do is to design an algorithm that converges to a form of, to a kind of information where the quantum bits are not in a superposition anymore, but they are in a sort of equivalent classical state. One easy example, and I'm going to do some shameful advertising here. I have some YouTube videos myself. There's something called, yes, I've seen them, they're very good. Yes, there's the quantum computing concept. There I show a brief example of the quantum search algorithm. Yes. Yes. So that works exactly like that. So let's say you want to search for an item in an unsorted database.

Andrea Morello: Unsorted is important.

Andrea Morello: Oh, so if it were sorted, then of course you'll find it. But if it's unsorted, and the reason this algorithm is intellectually important is because it's one of the very few where we know mathematically that there is a quantum advantage. Right. For example, for the factoring algorithm, the famous one, you know, cracked... We'll talk about that, yep. We don't know of any way to find the prime factor of a large number in a polynomial time. But there is no proof that this way does not exist. Exist, that's right.

Andrea Morello: Okay.

Andrea Morello: Whereas in the search through an unsorted database, you can very simply prove there is no better option than scrolling through the database. Okay. So if you have an old school telephone book, and I give you the phone number of someone, and you want to find who is the person who has that phone number, you have no better way than just scrolling through the whole thing until you find the name. Or you can randomly jump around.

Andrea Morello: Or randomly jump around.

Andrea Morello: Or randomly jump around, but there's no advantage in doing that. Yeah. In a quantum computer, you get there in the square root of the number of steps. So if you have, you know, one million entries like the Sydney telephone book...

Andrea Morello: A thousand efforts.

Andrea Morello: A thousand steps. On average. On average, yeah. Right. So the way it works is this. You create a superposition of all the entries in the database, and then at every step, at every operation, you do an operation, which I won't go into, but that concentrates the probability amplitude on the item you're looking for. Right? So at the beginning, all the entries are equally probable. Yep. And at every step, the other ones shrink, and the one you're looking for rises, pops up. Yep. So after about a thousand steps, on a million, you know, entries phone book, you will have essentially all of the probability concentrated in the one item you're looking for, and nearly zero everyone else. So now you do a quantum measurement of your register, and with very high probability, you'll find, boom, it's in there. Does it ever reach a hundred percent probability? In theory?

Andrea Morello: Does it ever reach a hundred percent?

Andrea Morello: Well, I mean, eventually if you run the classical algorithm, it will. Right. Right.

Andrea Morello: Okay. Yes. Got it.

Andrea Morello: So if you want, if you do a million steps, then eventually you're equivalent to the classical algorithm. Right. And also there, so in the classical case, we say that the runtime is n over two. So on average, you will scan a half a million numbers. But you know, if you're unlucky that the number you look for was the last one, then you need to scroll all a million of them. That's right. It's averages. On average, it's n over two. On average is square root of n.

Andrea Morello: So the advantage in this particular case, which is one of the cases that we know of, one of the few, we're going from half a million to a thousand. Mm-hmm.

Andrea Morello: So what's that? Yeah. So it's only square root, you see. So it's not the kind of thing that really changes the complexity class. So it doesn't go from exponential to polynomial. But it's one case where, you know, if you actually look at how it works, it's reasonably simple. And you see really this probability shrinking everywhere and peaking there. And it tells you, you know, also the typical question is, but how do you read out the quantum register if it's got all these superpositions? The point is, a good working quantum algorithm will not have a superposition as an output. It will have a measurable state.

Andrea Morello: Right. A measurable state.

Andrea Morello: And that's what makes it hard to develop the quantum algorithms that give you a useful result.

Andrea Morello: So even if we had a quantum computer tomorrow, the algorithms would be severely lacking. The...

Andrea Morello: There's a few, like there's hundreds, but... Oh, okay. There's a few hundred. Right. There's a few hundred. There is a webpage that's kept up to date is the quantum algorithm zoo. And there's some interesting ones like solving sparse systems of linear equations. Oh, yes. That's pretty broad.

Andrea Morello: Okay.

Andrea Morello: And one thing that's really interesting, I'm just going to tell you this because it fascinates me. I only recently discovered quantum finance.

Andrea Morello: Quantum finance.

Andrea Morello: Yes. And so there's various ways in which you can do that. So there are people who are looking into using quantum computers for optimization problems, like portfolio optimization, portfolio investment optimization. Basically, you have this much money and you have some constraints on, you know, how many types of different stocks you want to buy. These things become computationally very, very hard, very soon. Right. And so there are suggestions that quantum computers might be able to run this kind of optimization problems in a more efficient way.

Andrea Morello: I wouldn't have thought that there'd be a place for quantum there.

Andrea Morello: No, apparently there is. Okay. And the thing that really blew my mind and only recently discovered is that, now I'm not a finance expert, of course, but apparently there is some model whose name I forgot, some economist, who developed a mathematical model for how to essentially model markets in the presence of arbitrage. Ah, yes. Arbitrage is when you have, for example, exchange rate fluctuations. So, you know, you buy a TV in Australia, but if you went to buy it in New Zealand, you get some advantage because the, you made money that way. You made money that way. Yeah. And it turns out, it's amazing. If you look at the form of that equation, it's like the Schrodinger equation of quantum mechanics.

Andrea Morello: Oh.

Andrea Morello: Where the Planck constant is the degree of arbitrage.

Andrea Morello: Oh, no. Yeah. It pops out, does it?

Andrea Morello: It pops up. So the quantum uncertainty we have in the Schrodinger equation in the economics model is the arbitrage, the uncertainty in the exchange rates.

Andrea Morello: Mind blowing.

Andrea Morello: Right? That's it. It's just a mathematical coincidence. Yeah, yeah. There's nothing quantum about the finance. Oh, okay, there's nothing. There's nothing quantum. It just so happens that the form of that equation has the same form as the basic equation that governs the time evolution of a quantum system. Wow. So then if you want to model how will, you know, your investment in the presence of arbitrage across different markets evolve in time, you can just as well create a quantum system that is subjected to formally the same equation and watch it evolve.

Andrea Morello: Interesting. Oh, that is fascinating. Right? Wow. That's great. Let's talk about cryptography, prime numbers, because everyone freaks out. Oh, quantum computers within five years will be, no. Okay. Yeah. Tell us why you're laughing at the thought of that.

Andrea Morello: So with the present knowledge we have, the most, let's say, plausible architecture of a quantum computer that might be able to break and let's say RSA encoding would require about 200 million physical qubits.

Andrea Morello: Because you need to have...

Andrea Morello: Because of the error correction. So that includes error correction. Right. You need to have a physical qubit.

Andrea Morello: So you can't do it with just your 50 or 100 qubits.

Andrea Morello: No way. No way.

Andrea Morello: Why is that?

Andrea Morello: Because you're lacking the... You're lacking the ability to correct the errors. Got it. So you could do it... Okay, let me think.

Andrea Morello: So we're talking hundreds of millions.

Andrea Morello: Yeah, of physical qubits. So you need a few thousand logical qubits. So if you had a few thousand perfect qubits with zero errors... You can do it. Done. You can do it. But you never will. There will always be errors. And so with some realistic values for the errors we have, for the ways we use to correct them, you have to budget for a few hundred million physical qubits. So that's a long way from where we are.

Andrea Morello: That is. Is there any shortcuts to that that you can foresee?

Andrea Morello: Not that we... Well, you see, that's the thing about this field. This is not like making a better washing machine or a better car, right? This is uncharted territory. And so the typical mistake that people have made in the past, and I'm trying not to make it myself, is to make prediction on the basis of what we know now. Yes. So I can tell you what I think based on what I know, but tomorrow there could be someone that comes up with a better quantum error correction code, with a better quantum algorithm, and you just don't know. This is uncharted territory. Okay? So when you speak to, you know, people in, you know, in the banks, the CTO of a bank, and you try and tell them, you roll your eyes, and say, man, come on, it's 200 million qubits, and it's, no, no, no, you don't understand. We cannot afford the risk that someone...

Andrea Morello: Will do that in the next 10 years.

Andrea Morello: ...do that in the next 10 years. The other thing that, again, I only recently discovered, cryptographic systems for, especially for finance, I mean, for anything, they have a very long time scale. The time it takes to put a full cryptographic system in place is 20 years. Right. It takes a long time. So people really need to think 20 years ahead. And so then if you ask me over a 20-year horizon, can you see this happening?

Andrea Morello: Well, yeah. Okay.

Andrea Morello: Two years? Probably not, but I only say probably, because for all you know, tomorrow morning, some, you know, smarter guy than me will pop up with something. And, you know, big financial institutions or governments or security agencies simply cannot afford to be left unprepared.

Andrea Morello: Which is, it brings one of the questions one of my audience had is, where's the funding coming from? I mean, it's no secret. I've been to your webpage. The funding comes from NSA, like various government agencies. Yeah.

Andrea Morello: So my funding is from the Australian Research Council. I have funding from Australian Department of Defense. And I have funding from the US Army Research Office. And, yeah. So that covers it for now.

Andrea Morello: They're just hedging their bets. Yeah.

Andrea Morello: I mean, the Australian Research Council, of course, funds basic research. So that's their job to do. Of course. The security agencies and the defense agencies are, you know, trying to stay ahead of the curve. And, but now there is private sector funding coming into place. Okay. So I have at UNSW a company that's called Silicon Quantum Computing. That is a partnership of UNSW, of Telstra, of the Commonwealth Bank. Okay. Yes. Of New South Wales government and of Commonwealth government. Right. So there's five partners and they are pushing the scanning tunneling microscope technology I was telling you about before.

Andrea Morello: Right. Because they think that's the most, they just pick that one.

Andrea Morello: Yeah. Big one. There are other companies in Sydney, for example, there is a startup founded by my good friend, Mike Biersuk. And the company is called Q-Control and it's essentially a quantum control software for the most part development company. So they get venture capital, such fully venture capital. India and there are various others, you know, medium to small size startups, mostly in the, in the topic of quantum algorithms, quantum simulation, quantum software, quantum applications. Some also in quantum hardware, there are some in Europe making prototypes, silicone or superconducting qubits or ion traps. Some companies make specialized electronic circuits, classical electronic circuits for controlling quantum, quantum hardware. quantum hardware. So if you come to my lab now, you will see the usual suspects brands of your favorite top shelf, you know, microwave and digital electronics vendors. And that's all very well, but you can design, and some of them are already doing it. You can design dedicated, you know, FPGAs, digitizer, microwave sources with super stable clocks, all the things that you can kind of buy commercially. So if you know what is the application you have in mind, you can make like multi-channel boxes, which are much more economical than just buying many, you know, arbitrary way from generation, stacking them up in a room, you know. So these things are also happening. It's an ecosystem that includes a lot of the classical control and electronics engineering.

Andrea Morello: So at the moment, people don't need to be concerned, that concerned, about quantum computers breaking private number encryption.

Andrea Morello: No, and moreover, I would say the other side of the coin is that quantum technologies also give you the opportunity to secure your data.

Andrea Morello: Oh, it's going to get to the secure part of it. Right?

Andrea Morello: And in many ways, that is more advanced than the cracking side.

Andrea Morello: Yes.

Andrea Morello: So you can already buy, even in Australia from Quintessence Labs, a, you know, two boxes that plug on an optical fiber with which you can transmit quantum encoded data. Right. Right? There's a couple of companies around the world, and one of them is actually in Australia. These things already exist, right?

Andrea Morello: So that's transmission of data. That's not... Well, transmission and encryption.

Andrea Morello: So because they're using entangled photons, so they encode it into entangled quantum states, and then you're transmitting and then you decode it. Got it. Also things like quantum random number generators, which are very important for, you know, various cryptography and security aspects, they, you know, they exist, and they are an improved technology over the classical ones.

Andrea Morello: Because a lot of the existing encryption algorithms rely on a true random number generator. So, yeah. Yeah.

Andrea Morello: So, you know, I would like to remind people that, you know, quantum mechanics can be used for good or for evil, you know, like most things, you know. But the good is arguably more advanced and more developed than the evil. Than the evil. Got it. Right. So, you know, it's okay.

Andrea Morello: So at the moment, we're not that concerned. No. But we... No, but you can't lower the guard. You need to be prepared. No, no.

Andrea Morello: You can't ignore it. Well, some agency like the National Institutes of Standards and Technology in the US, they have already a couple of years ago started this big initiative to make quantum-safe encryption. So they call for proposals and they are scanning various ideas for how to encrypt classical data in a different way that makes it protected from, as far as we know now, quantum attacks.

Andrea Morello: Are they working towards an international standard for that, like an IEEE standard or something for quantum encryption?

Andrea Morello: Yeah, so it's a pretty, they cast the net pretty broad, they call for proposals and I imagine that what they eventually want to get at is a standard that, you know.

Andrea Morello: So quantum encryption is going to happen before quantum decryption? Well.

Andrea Morello: I would hope so.

Andrea Morello: Quantum decryption of existing algorithms. Yeah. Right.

Andrea Morello: Yeah, I would think so.

Andrea Morello: Okay. So there you go, we don't have to worry.

Andrea Morello: That much. Stay alert. That much. Participate in the development, you know. Okay, and this is the other thing I really want to say. There is so much to do in this field, like, you know. Yeah. Not only you don't have to worry, but you have an opportunity to be engaged and to be involved. There is a lot that we can do. There is a lot that we can develop for the good. And there is a lot of room and opportunities for engineers, also classical engineers, to get involved in this field. So if you look at the biggest quantum computing initiatives in the world, you know, the ones of Google and IBM and Intel and Microsoft, and also some of the small startups are, in fact, quite big when you go and look how many people they got on board. There is a lot of, you know, microwave engineers. Yeah. There is a lot of software developers. There is a lot of people who have a classical engineering background and their role is indispensable. Excellent.

Andrea Morello: So you don't have to be a physics researcher.

Andrea Morello: You don't have to be a physics researcher. But I think it will be good if in the very, very, very near future, we train a new generation of quantum engineers. That's actually the other one of my passions and goals in life. Yes, I'm a researcher and I build quantum computing hardware, but I'm also a teacher. And so over the years at UNSW, with some colleagues, I've been developing some essentially quantum engineering courses that we are trying to kind of crystallize and formalize now. It's within the environment of electrical engineering. I call it the microelectronics and microwave engineering of the 21st century. Okay. Right. So even now, even if you don't care about building a quantum computer, I would make the point that unless you abdicate your capacity to really be, you know, involved in the cutting edge of technology, you need to have some understanding of how the quantum world works. Right. The transistor you have in your pocket is, you know, on the tens of nanometer scale. You have no chance of understanding what's going on in there unless you understand some quantum mechanics. And, you know, some people in some places may have a bit of a black boxing attitude. It's like, oh, you know, there's someone out there in Silicon Valley understands that stuff. I just want to make an app to deliver pizzas, you know. And that's okay. But I think it's a bit unambitious, you know.

Andrea Morello: Got it.

Andrea Morello: We can do better than that. And Australia has a strong presence and a strong background in the quantum technologies. And I think we have a real opportunity to be a hub for quantum engineering.

Andrea Morello: Fantastic.

Andrea Morello: Which doesn't mean forgetting all we know about the classical electronics engineering. It just means blending it and upgrading it to what are the nanoelectronics, microwave engineering technologies that are coming up.

Andrea Morello: Do you ever see a point where people will be doing quantum computing, straight quantum computing, agree at university instead of classical? Will there be that much demand that, oh, you know, there'll be whole courses devoted just to quantum computing, learning quantum computing?

Andrea Morello: Yes. I would call it quantum engineering. Quantum engineering.

Andrea Morello: Okay. So, okay.

Andrea Morello: So there might be...

Andrea Morello: Well, when does it become, when does it stop being engineering and becomes just general computing?

Andrea Morello: Well, so, okay. I'm speaking maybe from my, you know, background of electronics engineering. I see mostly the hardware part of things. Mm-hmm. And then we also teach, of course, a bit of software and some of the communication protocols. In the, you know, computer science department, they might think, wow, one day we will have a whole school of quantum software. And in fact, some places like UTS here in Sydney, they have masters. Oh, really? Okay. Interesting. Yeah. So we are working more at the undergraduate level. So at UNSW, we are developing essentially a quantum engineering offering that is at the undergraduate level. Okay. So that would be a... Which is really quite unique.

Andrea Morello: Elective course. Yeah. That they can take as part of their normal degree. Yeah.

Andrea Morello: Okay. That's already there now. Right. And then we are looking at in the near future, really making it an actual stream, an actual degree that you can...

Andrea Morello: Right. Because a lot of engineers will... You know, regular practical electronics design engineers won't believe it until they can buy the chip on DigiKey or somewhere. You know, they won't believe it until they can physically...

Andrea Morello: That TV is called a quantum dot TV for a reason, right?

Andrea Morello: Oh, true.

Andrea Morello: I just bought this TV two months ago. It's a QLED. The Q stands for quantum dots. Quantum dots. I mean, it's not. It's not science fiction. It's on the wall. Okay.

Andrea Morello: Yep. But as we talked about before, you can't... At the moment, you can't really see a practical way of doing it without going to the super low temperatures.

Andrea Morello: That's perfectly fine. There's no issue with that. Right. On the topic, before I forget, there's a very important document that was launched by CSIRO just maybe three weeks ago. It's the quantum technologies roadmap for Australia. Oh. They spent a significant amount of time and effort really mapping out what is the prospect for quantum technologies industry in Australia. And they estimated 4 billion revenues and 16,000 jobs by 2040. By 20...

Andrea Morello: In the next 20 years?

Andrea Morello: 2040, yeah. Okay. Wow. And if you speak to the... They had a webinar sort of launch of this quantum technologies roadmap. And there were a number of people participating there, including a venture capitalist. And he, as an investor, he thought that those numbers were really, really underestimated.

Andrea Morello: Oh, okay. Yeah. So, somebody wanted to know, is there any way that, you know, people interested in this sort of stuff can experiment in their own labs? Is there any, like, or do you really need the, you know, the liquid helium and do you need the, you know, the 40 gig transmitters and the... Okay.

Andrea Morello: So, in the context of this quantum engineering courses that we teach at UNSW, we've actually developed labs for students. Okay.

Andrea Morello: Yeah.

Andrea Morello: So, some colleagues of mine have, in fact, built up this little lab that the students do where you can control and measure a single spin.

Andrea Morello: That's all I'm talking about, just, you know, not doing anything serious.

Andrea Morello: At room temperature.

Andrea Morello: Oh, okay.

Andrea Morello: There is a particular system, it's called the nitrogen vacancy center in diamond. Ooh. It's a color center in diamond. It's one of those defects in the diamond crystal that give it a bit of a color. Yes. And it turns out this defect has a spin. Ah. And this spin has an energy splitting that's created by the crystal field. So, you don't even need a magnetic field for it. If you apply magnetic field, you can shift it a bit. But it has an intrinsic splitting of 2.7 GHz, which is a nice number.

Andrea Morello: I know.

Andrea Morello: Which is doable. It's doable. And then you say, okay, but at room temperature, 2.7 GHz is much less than the thermal energy. Yes. In fact, it works nonetheless, because diamond is a very stiff crystal. Mm-hm. Which means that the thermal, the frequency of the thermal vibrations of the diamond crystal is extremely high. It's like, imagine a massive spring model. It's like having very light masses, because carbon is a light atom. Yeah. And very stiff springs. Right. Whereas this spin has barely any interaction with that crystal lattice. So, the spin essentially doesn't know what the temperature is.

Speaker ?: Oh, interesting.

Andrea Morello: Right? The spin, I mean, it will eventually. Yeah, yeah, right. Like after a couple of milliseconds, you will find out what the temperature is. Okay. But you have a window of time where you can thermally insulate it and you can play with it. Right. Interesting. And so, in fact, my colleagues have published the paper where they describe how this little student's lab is built, and, you know, it's actually on the public archive. Oh. So, if someone wanted to-

Andrea Morello: That's terrific.

Andrea Morello: Somebody wants to have a play with that, they can. Have a play, you know, there's, I don't know how much you would call, there's probably a few thousand dollars off, you know, like, you need some radio frequency sources, you need some digitizers. Yeah. There's a few thousand dollars worth of stuff, but it's not, you don't need a refrigerator in your room, you know. Exactly. And especially if you are an electronics engineer, you may have some of the instruments already in your lab, you know.

Andrea Morello: Yeah, exactly.

Andrea Morello: So.

Andrea Morello: Okay.

Andrea Morello: Yeah, so it's not out of this world. Okay. It's not out of this world.

Andrea Morello: So people can have a little- People can have a little play if they want.

Andrea Morello: Yeah. Interesting. So what they can do is to log on to the existing and openly accessible quantum computers that companies like IBM make available. Oh, they do?

Andrea Morello: They do.

Andrea Morello: So there's something called the IBM Quantum Experience, which is a five qubit superconducting processor that's in the cloud. Oh. So you can go on that website, make yourself an account, and you can program that quantum computer, it will give you back the results. There's lots of people who use it for teaching, for example. Oh. You know, like you can get your students to log on to that, you give them an assignment, and they use that quantum, IBM Quantum Experience for doing these things. Of course, there you don't, you know, you interface with your terminal. Yeah, yeah. It's not the same as making your own thing, but it is a legitimate five qubit quantum computer that you can access remote. Wow.

Andrea Morello: Okay. Fantastic. I've heard that there's, what languages are there? I've heard about QHash. Is that a...

Andrea Morello: Yeah, and then there's QuizKit. There's a couple of languages out there. There's Liquid. There's the one for simulations from Microsoft. There's a couple of ones.

Andrea Morello: And how do they differ from regular procedural or object-oriented?

Andrea Morello: They look a lot like assembler. They're really low level. They're reasonably low level languages.

Andrea Morello: Right. So you just, you know, flip this quantum bit, read this one, that kind of stuff. Yeah.

Andrea Morello: There's even people, there's colleagues of mine who, I don't know at what stage it's at, but some time ago they were telling them about, they were developing some quantum games. It's basically video games where the rules of the game are the quantum rules.

Andrea Morello: Ah, interesting. Okay. At an abstract level, I'm guessing, kind of.

Andrea Morello: Ah, no, there are actual rules. No, there are actual rules. Yeah. Okay. I never had enough time to actually play with them. I wish I could. I wish I was a teenager and I had time to play with it. Yeah, yeah, exactly.

Andrea Morello: You don't have responsibilities, you know. Join the club.

Andrea Morello: All the things I would do if I had the time.

Andrea Morello: Oh, yes. Oh, tell me about it. Give us your best and worst case predictions for, say, the next 20 years. Like, worst cases in, oh no, it all just peters out, no, there's nothing really practical, it doesn't work. And what's the, you know. I know you said you don't like giving predictions. No, no, no, I'm going to be really cheeky here. Please do.

Andrea Morello: You see, I'm fundamental, I'm a legitimate electrical engineer. I've got a degree in electronics engineering and I teach, you know, electromagnetism in third year. So, you know, I'm one of you, okay? Okay. I'm also someone who has a genuine passion for the fundamental questions in science.

Andrea Morello: Yeah.

Andrea Morello: So, for me, you see, quantum computing is a win-win bet. I see two scenarios. Either it works, like it actually works.

Andrea Morello: For, are you talking practically works? For practical things. For practical things.

Andrea Morello: It works now. Yeah. So, either you can actually build a scalable quantum computer, meaning you have figured out the fundamental logic gates, you have an error budget that is under control, you have protocols to detect and correct the errors that keep you below a correction threshold that the thing actually scales up. Right. And you can manufacture it and your manufacturing tolerances are compatible with the errors that you're able to tolerate for your quantum computer and you just make the thing. And the thing works. And you can run algorithms on it that give you interesting and useful and beneficial results. Right. That's scenario one. Yeah. Great, right? Yeah, fantastic. Holy grail. Scenario two, it doesn't. It doesn't. You can't do that. And it doesn't because of some fundamental law of physics we haven't discovered yet.

Andrea Morello: Got it. So, something you can't foresee at the moment. You can't foresee at the moment. But you can't rule it out that it may not be possible.

Andrea Morello: Maybe there is some fundamental constant that limits the quantum information you can encode. There have been some, you know, people have been wondering in the sort of fundamental quantum science community, you know, the problem of how does the microscopic quantum world morph into the classical world we experience at the large scale. Right. And some people think it's a known question. It's just a kind of, you know, stupid thing to even ask. Some people come up with various theories of what it could be. And some people have even imagined that there might be some scale, which could be a size scale, it could be a mass scale, or it could be an information density scale, at which the, for example, the Schrodinger equation, as we know it, which is a linear differential equation, is no longer exact. It gets some correction terms. Okay. And those correction terms have a pre-factor, have a constant in front of it, that if you're making, you know, a 10 qubit quantum computer is irrelevant. If you're making a 10,000 qubit quantum computer starts to affect things. Ah, interesting.

Andrea Morello: Right?

Andrea Morello: I'm not saying that this is what's going to happen, right? No, but it could in theory. But I see two scenarios. Either we're done with physics. I mean, but I don't know.

Andrea Morello: Come on, that's going to come back to haunt you that way.

Andrea Morello: Either, you know, the physics we know is correct enough that we can use it to manufacture a quantum computer that works, or it's not. Or it's not. Or it's not. So for me, both scenarios are fantastic. I would take both. I would take both.

Andrea Morello: Is that Andreas' cat? Is that like a Schrodinger's cat? That could be on Andreas' cat? Yeah. Both scenarios are alive and dead at the same time? Yeah. Yeah, okay. I'll find out.

Andrea Morello: You see, my wild dream is to, you know, have developed the technology to make, you know, quantum bits in silicon, have it all under control, have it, you know, have some deal with some foundry, and we get it all manufactured with the latest equipment, the fanciest technology. And we make a 100 qubit circuit, and you run it, and it works. You make a 1000 qubit circuit, and you run it, and it works. You make a 10,000 qubit circuit. Scales. Scales. And then eventually it just stops.

Andrea Morello: Yeah, that would be disappointing, wouldn't it?

Andrea Morello: That would be amazing. Oh, that, oh. That would be amazing. That would be like to actually watch in an engineered electronic device a new law of physics.

Andrea Morello: Oh, it could pop out. A new law of physics could pop out.

Andrea Morello: Electronics engineering, listen to this, my friends. Yeah. Electronics engineering enabling the discovery of a new law of physics. Wow, that would be something. That's what I work for. That's what I work for.

Andrea Morello: Wow. There's a Nobel Prize in that one.

Andrea Morello: Yeah. If I'm still alive, forget it. Oh, that is awesome. Thank you very much.

Andrea Morello: My pleasure. That is fantastic. Thank you for your time and the very, very clever questions. It was a very pleasant conversation.

Andrea Morello: Awesome.

Andrea Morello: Thank you very much. Catch you next time. Bye.

Andrea Morello: Bye.

Andrea Morello: Bye. Bye.

Andrea Morello: Bye.

Andrea Morello: next time