In BriefHow do you program a quantum computer? What could they be used for? When might one be in your pocket? Expert Mikhail Lukin chats with Futurism about these questions and more.
The science and tech world has been abuzz about quantum computers for years, but the devices are not yet affecting our daily lives. Quantum systems could seamlessly encrypt data, help us make sense of the huge amount of data we’ve already collected, and solve complex problems that even the most powerful supercomputers cannot – such as medical diagnostics and weather prediction.
If you still don’t understand what a quantum computer is, what it does, or what it could do for you, never fear. Futurism recently spoke with Mikhail Lukin, a physics professor at Harvard University and the senior author of one of those papers, about the current state of quantum computing, when we might have quantum technology on our phones or our desks, and what it will take for that to happen.
This interview has been slightly edited for clarity and brevity.
Futurism: First, can you give me a simple explanation for how quantum computing works?
Mikhail Lukin: Let’s start with how classical computers work. In classical computers, you formulate any problem you want to solve in the form of some input, which is basically a stream of 0s and 1s. When you want to do some calculation, you basically create a certain set of rules depending on how this stream should actually move. That’s the process of calculation — addition, multiplication, whatever.
But we’ve known for more than 100 years that our microscopic world is fundamentally quantum mechanical. And in quantum mechanics, you can have systems. Your computer, for instance, or your chair can be placed in two different states at once — that’s the idea of quantum superpositions. In other words, your computer can be simultaneously both in Boston and in New York. So this quantum superposition, even though it sounds very weird, is allowed by the laws of quantum mechanics. On a large scale, like the example that I gave, it is clearly very strange. But in the microscopic world, like with a single atom, creating this kind of superposition state is actually quite common. So by doing these scientific experiments, scientists proved that a single atom is in two different states at once.
The idea of quantum computers is to basically make use of these rules of quantum mechanics to process information. It’s pretty easy to understand how this can be so powerful. In classical computers, you give me a certain input, I put it in my computer, I give you an output. But if our hardware was quantum mechanical, rather than just sequentially providing some input and reading out the answers, I could prepare the computer register in the quantum superpositions of many different kind of inputs.
This means that if I then take this superposition state and process it using the laws of quantum mechanics, I can process many, many inputs at once. It could be potentially an exponential speedup, compared to the classical programs.
F: What does a quantum computer look like?
ML: If you were to walk into a room with our quantum machine in it you would see a vacuum cell or tube and a bunch of lasers which shine into it. Inside we have a very low density of a certain atom. We use lasers to slow down the atomic motion very close to absolute zero, which is called laser cooling.
F: So how do you program the thing?
ML:. To program a quantum computer, we shine a hundred tightly-focused laser beams into this vacuum chamber. Each of these laser beams acts as a optical tweezer, grabbing one atom or not. We have these atom traps, each of which is either loaded or empty. We then take a picture of these atoms in these traps, and we figure out which traps are full and which are empty. Then we rearrange the trap containing single atoms in any pattern that we wish. This desired arrangement of single atoms, each individually held in and easily controlled, are positioned basically at will.
Positioning these atoms is one way that we can program it. To actually control the qubit, we gently, carefully, push the atoms from their lowest energy state into a higher energy state. We do this with carefully chosen laser beams that shoot to one specific transition. Their frequency is very tightly controlled. In this excited state the atom actually becomes very big and, because of this atom size, the atoms start interacting or – in other words – talking to each other. By choosing the state to which we excite the atoms and choosing their arrangements and positions, we can then program the interaction in a highly controllable way.
F: What kinds of applications would a quantum computer be most useful for?
ML: To be honest, we really don’t know the answer. It’s generally believed that quantum computers will not necessarily help for all computational tasks. But there are problems that are mathematically hard for even the best classical computers. They usually involve some complex problems, such as problems involving complex optimizations in which you try to satisfy a number of contradictory constraints.
Suppose you want to give some kind of collective present to a group of people, each of which has its own niche. Some of the niches might be contradictory. So what happens is, if you solve this problem classically, you have to check each pair or triplet of people to make sure that at least their niche is satisfied. The complexity of this problem grows in size very, very rapidly because the number of classical combinations you need to check is exponential. There is some belief that for some of these problems, quantum computers can offer some advantage.
Another very well-known example is factoring. If you have a small number, like 15, it’s clear that the factors are 3 and 5, but this is the kind of problem that very quickly becomes complicated as the number grows. If you have a large number that is a product of two large factors, classically there is pretty much no better way to find what these factors are than just trying numbers from one, two, three, and so on. But it turns out that a quantum algorithm exists, called Shor’s algorithm, that can find the factors exponentially faster than the best known classical algorithms. If you can do something exponentially faster than using the alternative approach, then it’s a big gain.
F: It sounds like your mission, and that of others in your field, is to help us advance and understand this technology, but the applications are sort of secondary and will come when you have the tools. Does that seem about right?
ML: I will answer your question with an analogy. When classical computers were first developed, they were mostly used to do scientific calculations, numerical experiments to understand how complex physical systems behave. Right now, quantum machines are at this stage of development. They already allow us to study complex quantum physical phenomena. They are useful for scientific purposes, and scientists are already doing it now.
In fact, one significance of our papers [published in Nature] is that we have already built machines, which are large enough, and complex enough, and quantum enough to do scientific experiments that are very difficult to impossible to do on even the best possible classical computers — essentially supercomputers. In our work, we already used our machine to make a scientific discovery, which had not been made up until now in part because it’s very difficult for classical computers to model these systems. In some ways, we are now crossing the threshold where quantum machines are becoming useful, at least for scientific purposes.
When classical computers were being developed, people had some ideas of which algorithms to run on them. But actually it turned out that when the first computers were built, people were able to start experimenting with them and discovered many more practically efficient, useful algorithms. In other words, that’s really when they discovered what these computers can actually be good for.
That’s why I’m saying that we really don’t know now the tasks for which quantum computers will be particularly useful. The only way to find these tasks is to build large, functional, quantum machines to try these things out. That’s an important goal, and I should say that we are entering this phase now. We’re very, very close to a stage when we can start experimenting with quantum algorithms on large scale machines
F: Tell me a little bit about your Nature paper. What actually is the advance here? And how close are we to being able to start discovering the algorithms that could work on quantum computers?
ML: So first let’s talk about how one could quantify quantum machines. It can be done along three different axes. On one axis is the scale — how many qubits [a “quantum bit,” the unit that makes up the basis of quantum computer the way “bits” do in classical computing] it is. More is better. Another axis is the degree of quantum-ness, that is, how coherent these systems are. So eventually, the way to quantify it is that if you have a certain number of qubits, and you perform some calculations with that, what’s the probability that this calculation is error-free?
If you have a single qubit, you have a small chance to make an error. Once you have a lot of them, this probability is exponentially higher. So the systems described in our paper, and also in the complementary paper, have large enough qubits and are coherent enough so that we can basically do the entire series of computations with fairly low error probability. In other words, in a finite number of tries, we can have a result that has no errors.
But this is still not the complete story. The third axis is how well you can program this machine. Basically if you can make each qubit talk with any other qubit in an arbitrary fashion, you can also encode any quantum problem into this machine. Such machines are sometimes called universal quantum computers. Our machine is not fully universal, but we demonstrate a very high degree of programmability. We can actually change the connectivity very quickly. This in the end, is what allows us to probe and to make new discoveries about these complex quantum phenomena.
F: Could a quantum computer be scaled down to the size of a phone, or something vaguely portable at some point?
ML: That is not out of the question. There are ways to package it so that it can actually become portable and potentially can be miniaturized enough maybe not to the point of a mobile phone, but perhaps a desktop computer. But that cannot be done right now.
F: Do you think, like classical computers, quantum computers will make the shift from just scientific discoveries to the average user in about 30 years?
ML: The answer is yes, but why 30 years? It could happen much sooner.
F: What has to happen between now and then? What kind of advances need to be made to get us there?
ML: I think we need to have big enough computers to start really figuring out what they can be used for. We don’t know yet what quantum computers are capable of doing, so we don’t know their full potential. I think the next challenge is to do that.
The next stage will be for engineering and creating machines that could be used maybe to target some specialized applications. People, including [my team], are already working on developing some smallscale quantum devices, which are designed to, for example, aide in medical diagnostics. In some of these applications, quantum systems just measure tiny electric or magnetic fields, which could allow you to do diagnostics more efficiently. I think these things are already coming, and some of these ideas are already being commercialized.
Then maybe, some more general applications could be commercialized. In practice quantum computers and classical computers will likely work hand-in-hand. In fact, most likely what would happen is that the majority of the work is done by classical computers, but some elements, the most difficult problems, can be solved by quantum machines.
There is also another field called quantum communication where you can basically transfer quantum states between distant stations. If you use quantum states to send information, you can build communication lines that are completely secure. Moreover, through these so-called quantum networks, sometimes called quantum internet, we should be able to access quantum servers remotely. That way, I can certainly imagine many directions in which quantum computers can enter everyday life, even though you don’t carry it in your own pocket.
F: What’s something that you wish more people knew about quantum computers?
ML: Quantum computing and quantum technology have been in the news for some time. We scientists know that it’s an exciting area. It’s really the frontier of the scientific research across many subfields. Over the last five to 10 years, most people assumed that the developments have been very futuristic. They assumed that it will take a long time before we create any useful quantum machines.
I think that this is just not the case. I think we are already entering the new era with tremendous potential for scientific discoveries, which might have wideranging applications for material science, chemistry — really anything that involves complex physical systems. But I also feel that very soon we will start discovering what quantum computers can be useful for in a much broader scope, ranging from optimization to artificial intelligence and machine learning. I think these things are around the corner.
We don’t yet know what and how quantum computers will do it, but we will find out very soon.