IBM Research Explains the Functioning of Quantum Computing and Its Importance

While technological advances codified by Moore's Law are slowing down, computer scientists are turning to alternative computing methods, such as superconducting quantum processors to generate computational gains in the future.

Jeffrey Welser, vice president and laboratory director at Almaden's IBM Research, spoke about quantum computing at the 49th edition of the Semicon West chipset in San Francisco last week. I caught up with him to get the opinion of a layman on quantum computing.

IBM also exhibited part of its IBM Q system at the show, giving us an idea of ​​the amount of refrigeration technology to be built around a current quantum processor to ensure accurate calculations.

Binary numbers – ones and zeros – are the basic components of information in conventional computers. Quantum bits, or qubits, are built on a much smaller scale. And the qubits can be in a state equal to 0, 1 or both at any time. These computers can handle extremely complex calculations in parallel, but to be precise, they require high manufacturing accuracy. IBM is working to improve it, and it may take years before the improvements materialize and quantum computing has a chance to beat conventional computers, Welser said.

In a quantum processor, superconducting quantic bits, or quantum bits, process quantum information and return the results of calculations through the system via microwave signals. The entire device around the processor is designed to cool it as much as possible. The quantum processor must sit inside a shield to protect itself from electromagnetic radiation.

Here is a transcript of our interview.

VentureBeat: The usual question is: what is quantum computing?

Jeff Welser: Quantum computing is a computer form that takes advantage of certain quantum effects that we believe can achieve some types of algorithms much more efficiently than conventional methods. The basic unit of a quantum computer is what we call a quantum bit, a qubit. We all know ordinary bits, a one or a zero. This is what we use for a normal calculation. A qubit can also be one or zero, but since it is a quantum bit, it can be superimposed on both one and one zero. It has a probability of being one or the other.

In addition, you can entangle two qubits, or hundreds or thousands of qubits, and each time you perform an operation on one of them, the state of each of them. is instantly determined, due to entanglement. In a sense, it gives you the ability to do a massively parallel calculation. For algorithms or related problems, you can do things exponentially faster or better than you can with a conventional system.

Examples of things that can do that – chemistry and materials themselves, of course, are based on quantum chemistry. These are all quantum effects. You can simulate these atoms or interactions with a quantum computer much more precisely and at much larger scales. The example I gave in the keynote speech, think of the caffeine molecule. It is an important molecule for us every day. It contains about 95 electrons, so it's not a particularly big molecule, but if you want to simulate it exactly on a typical computer, you need to have 10 to 48 bits of conventional power. For reference, the planet Earth has about 10 to 50 atoms of power. Obviously, you will never do that.

With a quantum system, if it was a very robust and fault-tolerant quantum system, you could only do it with 160 qubits. The system presented here is a model of our 50-qubit system. We are not so far from 160 today. If you visit the IBM Q website, you can access a 16-bit system with which you can play for fun. In a way, we still have a few years left and we will have something very precious compared to conventional systems, but it's not as far as we thought.

VentureBeat: What kind of physical space are we talking about?

Welser: If you look at the system, the reason it is structured this way is really: you first have to isolate the chip. The chip is in the lower part where these cables are embedded. This is the real chip of quantum computing. If we used it, we would have a cartridge and all around to isolate it, so that you could not see it, but we discovered it. When it is clogged, this whole system goes at low pressure, but also at low temperature, which really matters.

The summit is about 40 degrees Kelvin, then it goes down to four Kelvin, 100 milli Kelvin, and so on. Basically, it is 15 milli Kelvin, which corresponds to 15 thousandths of a degree above absolute zero. For reference, the space is about two to three Kelvin. It's two hundred times colder than space when you go there.

The reason it must be so cold is that you have to isolate it from any kind of interference, including thermal interference. Any thermal energy will drop the qubits of the overlay state we wish. Despite all this isolation, the qubits will only maintain their superposition for about 100 microseconds. It's really very good. We are proud of this number. But it's still very short, obviously. You must do all your calculations during this period before generating an error.

VentureBeat: Is this a demonstration unit now?

Welser: It's a demo, yes. The components are all there. In theory, you could run it. But it lacks vacuum systems and other things around. The ones that currently work are in the basement of our Yorktown Heights lab in New York. We have several systems in place. These are the ones you can access in the cloud. We put the first online in May 2016. It was a five qubit system. As I said, we now have a 16-bit system that you can use for free, and we have a 20-bit system for people who join our network. We have a network of companies and universities, more than 70 currently, who also have access to 20 qubit system.

We have also put in place a complete open source software infrastructure called Qiskit. It gives people the tools they need to try to program that. As you can guess, one of the challenges is programming very different from the one we have a habit of living. Qiskit allows you to manipulate qubits individually, if you understand this part. Over time, we introduced libraries. A chemist could therefore use a library of quantum algorithms. They would understand what the high-level algorithm does and this would result in a run on a quantum computer.

VentureBeat: What do people find helpful right now?

Welser: Most people who consult it are in three main areas. One is the discovery of chemistry or materials. For example, JSR, a major producer of semiconductor polymers, is a member. Samsung is a member. They use it immensely – they believe that when they have large enough systems, it will help them discover new materials with different properties, no matter what application is needed. Materials generate a large part of what happens in consumer goods, cars, batteries, etc. This is one of the areas in which we believe that in three to five years we will have large enough systems for real benefits. For now, these are just experiences.

The next is optimization. We have J.P. Morgan Chase and Barclays as members. They plan to use it to perform very large quantum Monte Carlo simulations or other optimization problems for bond pricing or to predict the behavior of very complex financial systems. Today we do it with very large supercomputers, but that's one of those things where, as with the caffeine problem, you can only simulate a lot. It's actually five years before your system is big enough.

The other is about AI and machine learning. We believe that some machine learning problems that can be solved on quantum systems will allow you to create much larger sets of parameter and function spaces than on standard systems. We just published an article about it about six months ago. This one, again, is three or five years old, maybe five years old.

The one I did not mention and which most people think is factorization or cryptography, the idea that quantum computers can potentially factor very large numbers, and could therefore break the Internet, break the encryption that we use. It is true that if you had a system large enough, you could factor very large numbers and the current types of encryption that we use on the Internet would be vulnerable. But to get there, you'll probably need a system with thousands of qubits, or even millions. These should be very robust qubits, very error-free, that we do not have today. We have at least 10 years, if not 15 or 20 years, before we have a system that is strong enough to do it. No immediate concern there.

In the meantime, there are already known encryption methods that we could use today on conventional systems that are not properly mapped to a quantum computer. Even with a very large system, they would not be vulnerable. A form of cryptography called network cryptography, for example. We have a lot of time to implement that kind of thing. In fact, one of the topics we talk about with our customers, because many of our clients are major players in industry or government, is that it's too early to worry about anything. that breaks the Internet.

If you're archiving data or want to keep data secure and private for the next 10, 15, or 20 years, think about your tape archives of all your stored data, that's not too much. It's too early to think about encrypting that and using something like network cryptography, which is very feasible. In 15 years, you will not want to go back and re-encrypt all the data in your archives when quantum computers arrive. It is not too early to think of that.

VentureBeat: What is the effort deployed by IBM right now?

Welser: It's a strong goal. We have a lot of work in our Yorktown Heights lab, as well as software in the Albany lab, as well as in the Zurich lab. The creation of this vast network of universities and companies is explained in particular by the fact that we need a lot of people to work in different spaces. We will continue to advance the hardware part, of course, while continuing to activate the algorithm and software part, but we want a lot of people to build applications, because that is how we will find how to use this.

VentureBeat: How long have you been working on it?

Welser: It has probably been working since 1981. In 1981, a very famous meeting took place between physicists. He was co-sponsored by MIT and IBM. Richard Feynman, a rather famous physicist, is where he invented the idea of ​​quantum computing. He said that he thought it would be logical to think about using quantum effects to perform calculations. He also pointed out that it might be necessary to use them if you have always wanted to do chemical simulations.

This is where the idea started to make its way. People began to gather some ideas about what should be done to build it. David DiVincenzo, a physicist working for IBM in the 90's, put together a set of five criteria needed to build a quantum computer. In the late '90s, we built our first seven-qubit system using trapped ions – a completely different technology – just to prove it was even possible. It was not particularly usable, but it proved the concept.

As for the version you see here, we started working on this version rather like six or seven years ago, to determine how you can build – they are based on superconducting transmons, that is the real device located at the bottom. We started building that six or seven years ago and, as I said, in May 2016, we launched the first one.

IBM Q System One, our first commercial-grade version, will soon be up and running. It will be for many people who want a more robust system. We hope this will continue to extend the work to more companies that are not so much at the heart of quantum computing, but are more generalist.

VentureBeat: There were a lot of skeptics about it at first. What are the key steps you have overcome that have overcome this skepticism?

Welser: We see it progress very gradually. The skepticism comes largely from the fact that there are only two known algorithms: it has been theoretically proved that it is faster on a quantum computer. There is the Shor algorithm, which is the factorization, and the Grover algorithm, which is a type of search algorithm. But everything else was more speculative than whether it would be really faster.

We are starting to see published articles now where people are showing, "Hey, I just did it, and if you increase that up to a number of qubits, that's more than you could possibly do on a conventional system. "start running simulations and show that you can do it.That breaks some of the skepticism.

The other thing is that we started our own quantum volume increase roadmap, we call it. In other words, find ways to reduce the rate of error while increasing the number of qubits. This shows that you can make circuits deeper and deeper, algorithms more and more complex. All of these things are starting to make people think: "It looks much more real. Nobody knows where we'll go in the end, but people are starting to see that if you combine it with conventional smart computing, you can get something that might seem workable.

VentureBeat: Is there any benefit of Moore's Law to draw from this?

Welser: Not directly. There is probably no direct analog. However, we want to double the quantum volume each year, as Moore's law does to double the number of components. But it's a more complicated problem, because to double the quantum volume, you have to not only increase the number of qubits – it's pretty easy, because they're more important than what we do – they work in the range of 40 nm, as opposed to Moore's law. below 10nm today. We can easily make more qubits. This is not a problem. But if we do not improve the error rate on qubits, then having more qubits does not help. You must reduce the error rate.

We hope to find ways to continue to improve this error rate regularly, to allow the quantum volume to improve the Moore's law model. But physics is very different now.