We're kicking off 2021 with a new interview series: GOTO Unscripted, with our first round of interviews recorded back when we could still meet in person. GOTO Unscripted takes our conference speakers off the big stage and brings them behind the scenes for an intimate conversation on topics they know best.
Get a thorough introduction to what quantum computing is and the real-life problems you can solve using these computers and their quantum algorithms. Jørn Larsen talks to Guen Prawiroatmodjo and Jessica Pointing about all this and much more.
Jørn Larsen: We are here at GOTO Chicago, and we just have talks from some of the best quantum computing experts in the world. So, here with us, we have Jessica Pointing and Guen Prawiroatmodjo. Could you please tell us more about your background?
Guen Prawiroatmodjo: Sure, yeah. So, I studied physics in the Netherlands and I did my Ph.D. in Denmark, Copenhagen, and I'm specialized in experimental quantum physics. I worked at a computing startup called Righetti for about a year and a half and now I'm a software engineer.
Jørn Larsen: Cool. Jessica?
Jessica Pointing: Yeah, I'm originally from England and also live in Denmark for high school, and then came to MIT to study physics and computer science, and then I transferred to Harvard and got my bachelor's there, and then I moved to California to start my Ph.D. in quantum computing at Stanford. That's where I'm at at the moment.
How did you get into quantum computing?
Jørn Larsen: Thank you. And I'm Jørn Larsen, by the way. It was my crazy idea 23 years ago to do conferences and now we're here. So what led you to study quantum computing?
Guen Prawiroatmodjo: I was always very interested in computers in general, my dad was a software engineer so he gave me all his old toys and books and then I started coding. In school, I noticed that I got really annoyed because the physics teacher wouldn't explain to me how the equations work, so I was like, "maybe I should study physics because then I can find out," and turned out it would be true, I learned everything. I was so excited about it.
I want to combine these two passions, physics and computing, so that's how I ended up with quantum computing. Actually, this school I went to, Delft University of Technology has a really big and successful group doing quantum computing and all kinds of different platforms. So, I started doing spin qubits, which is a platform where you have a layered system and you have a layer of electrodes, so you isolate one electrode and you use the way it spins in the magnetic field to encode a qubit. I went from there.
Jørn Larsen: Awesome. And for how many years have you been within the field of quantum computing?
Guen Prawiroatmodjo: It's been almost 10 years, and I'm going to switch to biotech now so I took a break now from quantum computing. I might go back one day, who knows? But, yeah, it's about 8 to 10 years.
Jørn Larsen: A long time. Jessica?
Jessica Pointing: Yes, so when I got to MIT, I always loved physics and computing physics. Then I got into computer science as well and at that point, they were also a separate interest of mine. Then I remember actually, very vividly, I was in the dining hall on my laptop and a friend came over and he was like, "oh, I know you're interested in physics and computer science, you should take this quantum computing course that this professor is going to teach." Seth Lloyd is one of the pioneers of quantum computing. I said, "okay, maybe I'll try." I ended up taking the course, I added it super late, but then it was really fascinating and at the time, I thought, "oh, this is really interesting." So, that's how I got started.
Jørn Larsen: You fell in love with quantum computing?
Jessica Pointing: Yes..
Jørn Larsen: And when was that?
Jessica Pointing: This is my second year at university, so as a sophomore at MIT. So, yeah, a few years ago.
What is quantum computing?
Jørn Larsen: Ok. Thank you. Now we know that you're working on quantum computing, maybe we should talk about what it is. I don't know... who wants to… you can explain how to answer it.
Guen Prawiroatmodjo: Jessica is gonna do a great job.
Jørn Larsen: Ok. Jessica.
Guen Prawiroatmodjo: Well, overall, a quantum computer is a new type of computer that can solve certain types of problems significantly faster. The reason why, it's because underlying it is quantum physics, which describes how the world works at the level of atoms and subatomic particles such as electrons. For example, on a normal computer, you have bits, zeros and ones. I have this to demonstrate [a large, stuffed donut].
Jørn Larsen: Can I have a bite?
Jessica Pointing: For example, this is zero and this is a one. But in a quantum computer, you can have a bunch of bits, which can also be a superposition of zero and one. It's like it's spinning down because if I asked you, "is it zero or one?" You could say, "well, it's a combination of both." We can use superposition and other kinds of effects to give us advantages in quantum computers to solve problems faster.
Jørn Larsen: Cool explanation. So, has this all to do something with what Einstein was working on? Maybe, Guen, can you relate it to Einstein's work?
Guen Prawiroatmodjo: Einstein, it's definitely during that time, so Einstein was actually one of the physicists who had really, like, philosophical and religious sort of problems with quantum mechanics. It's more like people like Niels Bohr.
Jørn Larsen: Another Dane.
Guen Prawiroatmodjo: Another Dane, yes, a famous Dane, who laid the foundation of this work and defined what quantization means and how energy works at the small scales.
The notion of certainty in quantum computing
Jørn Larsen: Yeah, okay. Yeah, so do we have a notion of certainty within the quantum world? Who wants to give a go at that? Because this spinning donut seems a little bit like, "What is it? Is it a zero or is it a one?"
Jessica Pointing: Yes, so there are probabilities associated with these quantum states and the goal is we want to create a quantum algorithm. These are a set of instructions that you give to the quantum computer. Your goal when you're creating this algorithm is to have it so that you have a high probability of getting the right answer. And usually, if we wanted to design it, for example, that there is a 99% probability of getting the right answer and you theoretically know it should be 99% and you run it and it is around 99%, then that's how you can get certainty that you got the right answer. Ideally, we get 100% probabilities, but that's not always possible.
Jørn Larsen: I think we're getting more and more used to probability because also within AI, when you do facial recognition or stuff like that, it's like you have a probability that this is Guen, with 95% certainty. This could also be Sarah or another person, but we think it's Guen, you know? So, we're getting more and more used to computing with uncertainty, right?
Jessica Pointing: Yes.
Guen Prawiroatmodjo: That's right, and I think exactly this question is what also bothered Einstein. He said, famously, "God does not play dice." So, this was really a fundamental problem that people had, it took a while… it actually took the hippie movement for people to actually accept that our world is not deterministic, you have the least probabilities. So, when you ask like, "oh, what is certainty in the whole world?" It's actually a good thing that everything is not predefined, you know? It makes the possibility of free will more plausible, right? Because you have all these interactions, right?
A lot of people say, "oh, it's nature versus nurture and the decisions you make in life are just based on your environment and how you grew up and what you've been exposed to." But the fact that there is this fundamental… on the fundamental quantum level, there is this probability and uncertainty, like, philosophically, it gives you this notion of free will or freedom.
Jørn Larsen: Good stuff, Guen.
Jessica Pointing: Yes, I actually have something like that. It's not all random probability, we can design it so that we know what the probability should be, which is maybe different in doing a neural network, sometimes you may not know what the probability will be. But in this case, we do know, "ok, this will be the probability," we'll get this outcome, and then we can verify.
Jørn Larsen: Okay, actually, when we look at how it's actually working and how we can build computers today.
Jessica Pointing: Yes.
Real-life applications of quantum computing
Jørn Larsen: Yes. So, what are the potential applications of quantum computing, and how does the program look?
Guen Prawiroatmodjo: So, I think this probably more of a Jessica question because she's a quantum computer scientist.
Jørn Larsen: Ok.
Guen Prawiroatmodjo: But, I think there's a lot of potential applications, the most prominent one is quantum chemistry. So, I'm sure you're working in the field as well.
Jessica Pointing: Yes, in quantum chemistry, we want to basically simulate a model of how particular molecules work that can be used in medicine or in materials. It makes sense, right, because a quantum computer is a quantum system, so you want to use that to simulate a quantum system. That's one of the most promising applications of quantum computing.
Jørn Larsen: But there's also this thing about just solving hard problems like cryptography and things like this. Maybe you can explain that a little bit?
Jessica Pointing: Yes. Actually, one of the sort of algorithms that really push the field forward was Shor's algorithm, which can basically bind to the prime factors of a really large number.
That is one potential application you could get. Another one is machine learning and AI, so there's a field called quantum machine learning and it looks at, basically, in machine learning, you have vectors and matrices in a high dimensional space, and in quantum computing, the underlying mathematics is very similar. The idea is you can map the two together and you would use quantum computers to speed up some machine learning algorithms.
Quantum computing hackers: a real concern?
Jørn Larsen: I think now we can also touch on the security thing because many of us fear that once you have these big quantum computers, you can actually hack into any system and steal our money. So, would you be the future hackers, or what are we looking at?
Jessica Pointing: It's so fun.
Guen Prawiroatmodjo: Quantum hackers.
Jørn Larsen: This is fun, yes. So, how do we protect ourselves against kind of the threat of quantum computers hacking our accounts?
Jessica Pointing: Yes, it's funny because there's this guy called Bob Sutor at IBM and he gets this question. The first thing he says is, "maybe companies should just, first of all, encrypt their data just normally, and then they can start thinking about quantum-safe encryption, because a lot of companies don't even encrypt their data in the first place." I think when the company says, "It's probably something that people aren't thinking about," but it is an active area of research in the cryptography side to try and build new methods that could be safe against quantum computers.
And so, people are actively trying to do this and it's important that we do think about it now because it could take a while to actually replace all of our systems. But it's not something… it's not going to happen within the next few months or anything because, in order to actually run the algorithm to, like, break into RSA or the like, you need to actually have a quantum computer that's powerful enough to do that and currently, we don't have that.
Jørn Larsen: Okay, that's good, at least from that point of view.
Jessica Pointing: Yes.
Can you use quantum computing for time travel?
Jørn Larsen: Ok. Here's another cool question: can we use quantum computers for time travel?
Jessica Pointing: Interesting question.
Jørn Larsen: Yes, I don't know what to say.
Guen Prawiroatmodjo: So, I did actually recently stumble upon work where they demonstrated a time-reversal symmetry that's in quantum computing, these quantum circuits. So, that's not really the same as time travel but it does sort of demonstrate this fundamental theory about time in a quantum circuit.
Jørn Larsen: That has more complications because of this entanglement, is it true that if two qubits are entangled over a longer distance then they will be 100% synchronized and not, you know, bound to the speed of light or anything? Can you elaborate a little bit on that?
Jessica Pointing: Yes, actually, I want to go back to the time travel thing. There is something called quantum teleportation, so you can teleport quantum bits… and actually, that's one of the first things I learned in the quantum computing course that I took. So that's also interesting to think about.
In terms of entanglement, yes, so I think there have been experiments done where they separate some quantum bits by far distance, and they're actually able to get this correlation between the measurements of the two qubits.
Jørn Larsen: Awesome. There's a lot of possibilities.
Guen Prawiroatmodjo: Spooky action at a distance is what the early quantum theorists called it.
Jørn Larsen: Ok.
Guen Prawiroatmodjo: So, it's spooky because we don't know what it is because we can't actually visualize it mentally and, like, what's going on, but there have been experiments done, where they're using photons, so something they called flying qubits.
The qubits that you work on when you do the quantum computation, they're actually solid-state qubits, so electrons or microwave photons. But you also have these flying qubits where you take, for example, a lattice irregularity in diamonds and then entangle that with a photon and that photon can just, like, travel large distances. There has been an experiment done in China, I believe, where they actually bounce the photon off of a satellite over, like, a large distance to entangle them.
Jørn Larsen: Cool stuff. Let's look at what we have. Let's get down to earth. So, it’s so cool what we have on making these computers nowadays, let's have a look at that.
Jessica Pointing: Well, you have worked at one.
Companies building quantum computers
Guen Prawiroatmodjo: I have worked at one. So there's a startup called Rigetti at Berkeley and they're building solid-state quantum computers. There is IBM in New York and they're also building similar technology and then there's Microsoft and they are building something called Majorana qubits. It's a very different type of… fundamentally a different type of qubit, which is still very early in development but its prominence is really big because once you can create a special qubit, you don't have the problem of quantum decoherence anymore. Then Google is also developing solid-state qubits.
Jørn Larsen: Any new smaller startups?
Jessica Pointing: Yes, there are actually lots of startups. Surprisingly, there are like 100 startups across the world that are working on this, and some of them are focusing just on the quantum algorithms, and they call themselves quantum software companies, and other ones are actually beginning to try to physically build the quantum computer. So, there's a lot of...
Jørn Larsen: Can you buy them today?
Jessica Pointing: No. If you have a lot of money, yes.
Guen Prawiroatmodjo: Yes, if you can put the time on the...
Jørn Larsen: Time, yes so it's like a child offering.
Jessica Pointing: That's right.
How do you build quantum computers?
Jørn Larsen: Yes, quantum and forecast. Okay. So, how are they built?
Guen Prawiroatmodjo: So, how do you build a quantum computer? So, a quantum computer, essentially what it is, it's a quantum device or an integrated circuit that can encode quantum information.
To be able to actually do that, once you have the chip which you can create using lithography, so conventional lithography method. You also need very high resolution, so you typically use electron-beam lithography for that. And once you have that chip to actually be able to store the quantum information and retain it, you need to cool it down to almost absolute zero.
Typically, these systems are cooled down using something called a dilution refrigerator, which is a really big machine, it's about the size of a closet, maybe a little bigger, that cycles liquid helium in a diluted phase to cool down the system to these low temperatures.
You need a few different things. You need somebody to design the qubits, something to actually fabricate the qubits, and they need an experimental team to cool down the system and tune it and then you need a whole software stack on top of them to be able to program, to be able to create custom registers that can talk to these quantum registers.
So, you need a bunch of hardware that can interface with this quantum chip and send the right signals to it and you need CPUs, so you need a regular computer to be able to control this hardware.
The different architectures for building quantum computers
Jørn Larsen: Can you add something to that, the different architectures of building quantum computers?
Jessica Pointing: Yes, so there are many different ways of building a quantum computer. You mentioned one way and, for example, we can use photons, which is light, we can use trapped ions, we can even use diamonds as I mentioned. So, this actually is quite an exciting time because there are many different ways to physically build a quantum computer and people are trying to basically scale out these quantum computers and see which one is the best. At the moment, each one has its own advantages and disadvantages. Some will be more reliable but others not or vice versa.
Clustering quantum computers
Jørn Larsen: Interesting. Ok. So, there is this thing that quantum computers are still very small. Guen mentioned earlier today between 50 and 72 qubits quantum computers, and we would like to have billions of qubits to really solve super hard problems. Is there a way to just cluster a number of smaller quantum computers? Because if we can build like a 50 qubit computer, maybe you can just build 100 of them. How would that work?
Jessica Pointing: The main thing is, if you have multiple quantum computers, and you want to connect them together, you're probably connecting it classically, right?
There's classic communication. But really, the power of quantum computing is the ability to have these quantum effects entangle them together. And so, the problem with just taking multiple quantum computers and just connecting them classically is that you basically lose the quantum effects, you lose the power that it gives you if you had an actual quantum computer with all of these qubits.
Guen Prawiroatmodjo: Yes so you need something called a quantum link. As Jessica mentioned, right now the only way we can actually connect them is classically and classically meaning through Ethernet cables or through a direct cable or signal cables between them. But in order to actually have a full range 1000-qubit lattice, you need to be able to take five qubits from one chip and connect them quantum-mechanically to five qubits in the other chip.
Jørn Larsen: That's a big challenge.
Guen Prawiroatmodjo: Yes, exactly, so to do that, you need a quantum link.
There's research ongoing to create links between different systems, so we mentioned there's the diamonds, there's the superconducting integrated circuit qubits, and there's the spin qubits. All these systems have different ways of interacting between each other, so there's a lot of research groups working on creating a link between, a solid-state qubit and more of a photon-based system such that you can entangle a quantum state in the solid-state circuit to, like, photons, which you can then send over a fiber cable to another quantum computer. But this is really difficult because the problem doesn't scale very well, right? Because then you have 50 cables going to 50 connectors and imagine scaling that up. So, yes, that's still a big problem of how you can actually distribute that to people.
Progress made on optimizing quantum algorithms
Jørn Larsen: Ok. So, now, we already talked about some of the challenges and I guess, in talking about the algorithms, maybe we can talk a little bit about them because you can also fix the algorithms so they don't need this giant quantum computer. Can you explain some of the developments there?
Jessica Pointing: Yes, so one of the challenges in quantum computing is, if we go back to the qubit.
Jørn Larsen: Yeah, I love the donut.
Jessica Pointing: So, it's spinning but you see, eventually, it does stop spinning and it's sort of similar to quantum computing, eventually, you lose the quantum effects. This means that if you think of the computation as sets of operations, that means we can't do any operations because it just ends up being a mess at the end, so that means we can only do a short number of operations. As a result of that, researchers have come up with new types of algorithms, in particular, they're called hybrid quantum-classical algorithms. The idea is you use a quantum computer and the classical computer together.
What you do is on a quantum computer… because we can only do a short number of operations, we do that short number of operations and then we take this outfit to a classical computer, we do some processing on the classical computer, and then it spits out new things that should go into the quantum computer. Then we kind of do this loop between them and the idea is to take advantage of the fact that we have a quantum computer, but it has a small number of operations. It's also an open question whether this is actually useful or whether we can do things that, like, solve really interesting problems significantly faster but these types of algorithms have been built.
Gates in quantum computing
Jørn Larsen: Ok. I think we should talk a little bit about the different gates that we have, because I think that's really interesting. Because once you hear about the spinning things and then you think, "ok, how do you actually build stuff like that?" And I think a lot of our audience, they are used to like OR gates, AND gates, XOR, and what have you, but in the quantum computing area, you call it something else. Maybe you can explain the different kinds of gates that we have and how they work.
Jessica Pointing: Yes, I can explain the gate and you can say how you actually implement it.
Jørn Larsen: Ok.
Jessica Pointing: So, there are different types of...
Jørn Larsen: Maybe you need your donut.
Jessica Pointing: Yes, actually I do.
Jørn Larsen: The first thing is about reading the state, right?
Jessica Pointing: Yes. As I mentioned, during the actual algorithm, you want the qubits to be in superposition and in entanglement, but the thing is, at the end, information is just zeros and ones. So, how can we actually get that? Well, we can do a thing called a measurement. So, if it's in superposition, for example, I can make a measurement and force it to land on either side and we can force it in such a way that the probability of it landing on the correct side is the one we want. There are different types of gates that we can do before that.
For example, you can have the equivalent of a NOT gate — it's called an X-gate in quantum computing, but it's zero and you can flip it to one. You can have a gate called the Hadamard gate, where it's zero and then you can put it through the gate. I don't have a gate, but you can imagine putting it through a gate, and then it ends up spinning.
Jørn Larsen: The spinning superposition, it keeps it.
Jessica Pointing: Yes, by putting it through the gate. Then I think Guen can actually explain how you actually implement that on a quantum computer.
Guen Prawiroatmodjo: I love the donut, by the way. That's so great. It's cool. Yeah, so maybe it's easier to think of a quantum processor that works differently than a regular CPU, right?
In a regular CPU, what you have is you have these… well, they're literally gates, right? They're, like, open or close, right? Transistors and so on. How that works is you have the data flow through the gates, right? Type ones and zeros and they basically go around in, like, circuits. And it's like a black box, right? You put it in a 1 and you've got something out like a 0 or you can put in a 111 and you get a 000.
However, in a quantum chip, it works fundamentally differently. So, instead of having the data flow through the gates, the data is the chip. So, in the CPU, the gates are the chip, but in a quantum processing unit, QPU, the data is the chip, right?
You have the gates actually go to the data, so it's kind of a flip in how to think about it. So, to answer your question, like, "how do you implement these gates?" It really depends on the system. For example, in the microwave circuit, so in the superconducting qubit system used by Rigetti, Google, and IBM, is you send microwave pulses. So, I have a little antenna that… it's a little bit similar to RFID, are you familiar with, like, the RFID technologies?
Jørn Larsen: Yes.
Guen Prawiroatmodjo: Yes, so you have a card and you beep it, right? So, it's really like a small transceiver, so you send a signal and you get a signal back, that's similar to how these superconducting qubits work except that they're much, much smaller and they're much much colder. So, what happens in the normal resonant circuit is you have different frequencies, right? They have different harmonics that you can send, like, you can send like… similar to a guitar string? So they can have one note and then they have a higher harmonic of all these energies.
But in a quantum circuit, because it behaves inherently quantum-mechanically, you can actually see much more well-defined levels of these harmonics and you can isolate it. So, in a transmit qubit, I use a nonlinearity such that you can actually isolate two of these levels and those are my quantum information. What you do is you send the microwave pulse to excite that transition. Does it make sense? So, this is kind of more like a physics-y way of saying it but...
Jørn Larsen: You play the qubits.
Guen Prawiroatmodjo: You play a program on the qubit. So, with the donut example, you give it a whack, right? So, that's literally how you would do it and instead of giving it a physical whack, you give it a whack with a microwave pulse.
Jørn Larsen: Oh, cool. Let's see what's going on. Ok, so right now, you said one of the challenges is that we have a qubit in superposition and you want it to spin forever until you do something with it or you read it. So, how long will they live today or spin today? How long will they maintain the superposition?
Guen Prawiroatmodjo: So in the current systems, I think the best ones measured are in the range of like 100 to 200 microseconds but those are really difficult to reproduce, so we still have to figure out how to reproduce them. So, actually, reproducibility-wise, they're more in the range of like 50 microseconds and I think Jessica Pointing can give a better insight into how and what that means for circuit depth.
Jørn Larsen: I mean, your donut spins longer.
Jessica Pointing: Yes, it does.
Jørn Larsen: Maybe we just need to go back to that thing.
Jessica Pointing: Yes.
Guen Prawiroatmodjo: Donut completely.
Jessica Pointing: Yes, so as I said, it's already on the order of microseconds, so it's extremely short. So, the idea is how many like operations can you fit in in that amount of time? Then you can imagine, that's why it's a very small amount of operations that we can do in those few seconds.
Implementing quantum computers
Jørn Larsen: Ok, good. So what about the implementation of these physical computers? And how do you deal with the errors and imperfections of those computers?
Guen Prawiroatmodjo: Yes, so there's a whole field… like the one I mentioned, a theory called quantum error correction, which I don't understand. It's a holy grail and it deals with how to fix quantum errors. I don't know if you've worked with it in the field.
Jessica Pointing: Yes, I can totally talk about that. So we got actually a qubit, and when I talk about the quantum algorithms as a qubit, I'm not sure if they're a logical qubit, so I was assuming this qubit is perfect.
When I put it through the Hadamard gate, it ends up spinning 100% of the time, when I put it through a flip gate, it ends up flipping it. But what happens is, when you actually build this on the hardware it doesn't happen like that. What we want to do is we can actually add more qubits, so what quantum error correction does is we're basically… let's say, we now have three qubits, and these three qubits actually sort of represent one qubit.
So, it allows us to basically detect an error if it happens and then correct the error. And there are a whole bunch of errors, for example, if I put it through the X-gate, an error might happen that when I put it through the X-gate but it ends up still zero, that's an error because it should be flipping it. Or there are other types of errors like the ones that have just to do with noise, which is this idea of eventually, it's just going to stop spinning and so if you do a computation farther down the line, there's like areas where it just ends up not being fully perfect.
What can you do with quantum computing today?
Jørn Larsen: Okay, thank you. So, what already works today? I mean, what can you actually do with quantum computers today? What is the most practical application that you can do?
Guen Prawiroatmodjo: It's a great question. I don't know, actually. So, I think the most prominent application I've seen so far, is understanding classical algorithms better. But it sort of sounds like it defeats the purpose, but I don't think it actually does. So, there is this really talented researcher, her name is Ewin Tang, who shook up the quantum community and people are upset about that. She wrote two papers about two different… or about quantum algorithms but implemented them classically and she outperforms the original classical equivalence using another classical algorithm that was inspired by a quantum algorithm, right?
So, you can take this negatively and say, "oh, yeah, we don't need quantum computers, this is nonsense," right? Or you could say, like, "look, actually, studying these systems gives us better insight into how to solve classical problems." We should look at it from a very different perspective where basically you take problems that you would normally just walk through one by one, and you actually map them onto something like a wave function, like a probability, so you think about the problems differently. I think that's now the most practical application.
Jørn Larsen: I think you just have to give one example of what you can actually do with it.
Jessica Pointing: People are asking, well, you can actually run a lot of different types of algorithms on a quantum computer, you can run machine learning algorithms on a quantum computer. The only thing is, it's, like, limited, right? Like, you can run Grover's search, which is the one where you can search for a number in the square root of the number of entries that you have. You can run a very small example of Shor's algorithm, I think there's actually a recent announcement that they hit the larger limit in terms of the prime factors.
You can actually run a lot of algorithms, the only thing is, it's all limited in terms of… you have a limited number of qubits and so you can't do anything that you cannot do on a classical computer at the moment. People are working on this to try and actually do something that would be very difficult to do on a classical computer but do it on a quantum computer.
Jørn Larsen: So, what do you call the moment where you actually can beat a classical computer calculating a problem?
Jessica Pointing: So, there are actually multiple terminologies for this. There's a quantum speedup, which says that, for example, if you have an algorithm running in exponential time and you run it in polynomial time, then you get an exponential speed up, the quantum speed up. You can also say quantum advantage where that is you get quite a bit of advantage from a particular algorithm. And there's another terminology called quantum supremacy and this is a bit more specific but it says, "if we were to run something on a quantum computer, would it outperform the best classical hardware running this task?"
This is something that Google, for example, is trying to get out there because they said, "oh, we're gonna do this quantum supremacy experiment," and people are waiting, "when is it coming?" But the thing is that might not actually be a useful problem, and I think the problem they're working on is not actually usable? People didn't know any applications of it, but it would probably be possible within, hopefully, like next year or, like, months, I don't know.
Jørn Larsen: So, you're talking about a few months or years before that coming?
Jessica Pointing: Quantum supremacy, but the thing is it may not be actually useful if that makes sense? It will be a very, like, small mathematical problem that we could do on the quantum computer in just a matter of seconds, which may take a classical computer, I don't know, hours or months or days, but whether it's actually useful as another thing.
What are the problems quantum computing is solving now?
Jørn Larsen: So, what are you actually working on right now? I mean, in the field of quantum computing, what are the problems you really need to solve right now?
Jessica Pointing: Yes. So, I actually used the acronym MATHS, M-A-T-H-S, MATHS instead of MATH. We're using MATHS there. But it stands for M, "milestone," so this is the milestone I talked about, outperform the best classical hardware running the best classical algorithm. Then A stands for applications, so we need to continue to find new applications of quantum computing, design new algorithms to do this. Then T stands for theory, so, theoretically, if we know that there are more problems that can be hard classically but easy quantumly, that can drive a lot of research because they're reassured that, "Okay, like, there's something there that we can go for it,".
Then H is hardware, we need to improve… like increase the number of qubits as well as the quality of the qubits and reducing the error. And then S is finally shifting our mindset, as you said, from this classical world to the quantum world, which is interesting, because yeah, it helps us actually develop better classical algorithms possibly, but I think it would help us develop even better quantum algorithms.
Jørn Larsen: Interesting.
Guen Prawiroatmodjo: I can add to the hardware part of it. So, sort of the holy grail right now in superconducting qubits is actually the third dimension. So, you have these integrated circuits, right? But that forces you to make a flat structure of qubits. Well, if you can do, like, vertical integration, you can connect, like, qubits in three dimensions and that's sort of one of the biggest problems that the field is trying to solve.
Jørn Larsen: Interesting. So, is there a Moore's Law in quantum computing? Have you been tracking the number of qubits over time?
Guen Prawiroatmodjo: Yes, there's something called Rent's rule.
Jørn Larsen: Rent's rule?
Guen Prawiroatmodjo: Rent's rule, yes. So, there's a paper on that actually written by a collaboration between Intel and Delft that tells you that the number of connections you can actually make to this chip is limited fundamentally. So that's also why people are trying to solve this vertical integration problem. I think that's sort of the equivalent of Moore's law.
Jørn Larsen: Ok.
Jessica Pointing: One other thing to note is we can have more resistance but in qubits. The number of qubits is important, but also the quality of the qubits. You could just add, like, hundreds of qubits, fine, but they could all be, like, very poor quality and then it wouldn't give you much. So, actually, it would be much better to have, you know, like, 20 really, really good qubits than hundreds of real qubits. And so, that's an important thing to think of when we track qubits. But from my understanding, IBM released their, like, 5 qubit quantum computing, I think, in 2016 and so now, they said they have 50 qubits.
Jørn Larsen: It's good growth.
Jessica Pointing: That is doubling every year but, like, will that continue or what is the quality of these qubits is up to debate.
Guen Prawiroatmodjo: Yes, there's actually a website that just shows how many qubit… I think it's called howmanyqubitsinthequantumcomputer.com or something like that, but it just tracks how many qubits are announced, but as Jessica says, it doesn't really tell you anything if you don't know what the quality is.
What are the new types of algorithms we have in quantum computing?
Jørn Larsen: So, what are the new types of algorithms that we have in quantum computing?
Jessica Pointing: I think we talked about hybrid quantum-classical algorithms.
Jørn Larsen: Yes. That's what we want to do for now until we have bigger quantum computers?
Jessica Pointing: Yes.
Guen Prawiroatmodjo: Yes, and I think the Watson machine learning is one of the new latest and greatest that people are working on.
How to get started with quantum computing?
Jørn Larsen: So, you probably get asked a lot, "how do you get into this field? How can you get started?" If you watch this interview or one of your talks and you say, "I want to do it now, you know, how can you get started?"
Jessica Pointing: Yes, so it really depends on what you want to sort of get out of it. There are actually a lot of exciting resources out there right now. If you want to just play around with a quantum computer, I recommend you to Google "IBM Q Experience," you can literally drag and drop gates and try it out. Or you can use an API, Rigetti has an open-source API too where you can actually write Python code to actually code on a real quantum computer. If you want more of the actual theory, the most famous book is called "Quantum Computation and Quantum Information," by Isaac Chuang and Michael Nielsen. This is used in all introductory to quantum computing courses, and this will give you the mathematics and it will give you a good solid foundation for actually how it works.
Guen Prawiroatmodjo: Yes, I can add to that, like, in terms of jobs, if you're interested to work in the field, if you're like an electronics radio person like a microwave engineer, there's a lot of jobs for FPGA programming, that kind of stuff. If you're more of a software engineer, I recommend starting with what Jessica Point mentioned and just, you know, start coding and coming up with your own algorithms.
Jørn Larsen: Ok. Thank you, both of you for coming.
Jessica Pointing: Thank you.
Jørn Larsen: I think we all learned a lot here and I hope you have a great conference here in Chicago.
Jessica Pointing: Thank you.
Guen Prawiroatmodjo: Thank you so much.