What is the future of Google’s quantum computer

(ORDO NEWS) — Google specialists, who created a prototype of the Sycamore quantum computer, plan to use it for solving both theoretical and practical scientific problems. Among them are algorithms for autonomous transport, the study of quantum chaos and the internal structure of black holes. One of the developers of this computer, Vadim Smelyanskiy, told TASS about this.

What is a quantum computer?

Quantum computers are computing devices based on the principles of quantum mechanics. Unlike ordinary computers, in which bits are used to transmit and process data – units of information that contain either 1 or 0, quantum computers operate with qubits – memory cells and primitive computational modules that can store both zero and one at the same time. … Thanks to this, quantum computers can process large amounts of information many times faster than ordinary computers – even if they are supercomputers with enormous computing power.

There are two approaches to the development of such devices – classical and adiabatic. In the first case, scientists are trying to create a so-called universal quantum computer, which is similar in principle to an ordinary modern computer. An adiabatic computer is closer to analog computers. It is easier to create them, but such devices are able to solve only a limited number of tasks.

Google engineers and scientists have been developing a device for almost five years that should combine the merits of both approaches. Its first version consisted of nine qubits. The main task of the company’s specialists was to achieve the so-called quantum superiority, that is, the moment when no classical computer can achieve the same results as a quantum one.

Last November, Google officials said they had achieved quantum superiority in a math problem that boils down to iterating over a set of pseudo-random numbers. To solve it, the scientists used a prototype quantum computer, Sycamore, which consisted of 53 superconducting qubits. He found several million solutions in less than four minutes, while the most powerful supercomputer would have taken more than 10 thousand years.

In 2020, scientists from the China Science and Technology University reported a similar achievement. Their quantum computer, Tszyu Zhang, is based on 73 photonic qubits. He solved another problem of generating random numbers in 20 seconds, which a conventional computer would complete in several billion years. As with Sycamore, these claims sparked a lot of controversy about what constitutes an unsolvable problem and how it can be calculated on a quantum computer.

A revolution in science

Such advances made scientists wonder what practical problems a quantum computer can solve. One of the directions, according to Vadim Smelyanskiy, can be the simulation of physical systems. “Even now, even before the advent of error correction systems, quantum computers can be used to study the phenomena associated with quantum chaos and the disappearance of information. Another Google team will use quantum circuits to build models of quantum gravity,” the scientist explained.

The researchers also plan to use similar systems to study how some semiconductors and metals can turn into insulators due to the interactions of electrons within them. Scientists recently discovered that similar processes, the so-called Anderson transitions, can occur in a wide variety of environments, including inside crystals, electromagnetic waves or clouds of atoms.

Smelyansky and his colleagues hope that quantum computers will help understand the nature of these transitions and find out what role they play in the formation of the properties of a particular type of matter. According to the scientist, it is not necessary to use Sycamore and other superconducting quantum computers for this – other machines with a suitable number of qubits and a high level of their connectivity will be enough.

“All these phenomena can be studied in quantum systems of various types, including neutral atoms and ions. However, superconducting qubits are still maximally suited for solving these problems. Compared to their ‘competitors’, they are strongly connected to each other and work quickly. it is important, for example, when studying quantum chaos and various transitions between similar states, “the physicist continues.

No streak

In the future, as noted by Smelyansky, Google plans to use quantum computers to create scientific problems on which artificial intelligence systems will be trained, as well as for chemical calculations at the level of individual atoms. However, this will become possible only after scientists solve another important problem.

“There are still problems of quantum chemistry, but we do not yet understand whether they can be solved without creating error correction systems, in contrast to the algorithms that I have already mentioned. In fact, we will move into a new dimension when this milestone is reached , since we will be able to pose completely different, more ambitious and complex problems for ourselves, “the physicist noted.

Using error correction algorithms, physicists and chemists will be able to calculate how individual atoms and electrons interact within the most complex molecules. Thanks to this, scientists will be able to accurately predict what properties this or that – even not yet existing – compound has, and also purposefully synthesize substances with specific properties.

According to Smelyansky, two things are still missing for this. On the one hand, the number of qubits in quantum computers is still too small for the existing algorithms of quantum error correction to work on them. On the other hand, qubits make too many random errors during simple computational operations.

Because of this, scientists are not yet able to create large enough quantum computers that could independently correct errors in their work. As a result, quantum computers are in a kind of “no-man’s land”. They can be used to solve a limited number of complex practical problems, but so far they cannot be turned into general purpose computers.

“We assume that the accuracy of our logic circuits, which can now be achieved, is most likely not enough for conducting quantum chemical calculations. On the other hand, while this question is open. It is possible that we will be able to implement some simpler an approach for error correction that does not require qubits that live indefinitely, suppression of quantum noise and other complex tricks. We will be doing this for several years, “Smelyansky said.

Quantum acceleration

At the same time, according to him, scientists should focus their efforts not on increasing the accuracy of the qubits, but on creating algorithms that could suppress interference, as well as lightweight circuits with which it would be possible to correct errors.

“Even if we significantly reduce the level of errors and improve the quality of the linked chains of qubits, this will still not be enough to solve the problems of quantum chemistry. Therefore, progress in the area of ​​error suppression is critically important for us. Without this, we cannot do anything. And if we achieve all of the above, we still have no guarantee of success, “the physicist emphasized.

According to Smelyanskiy, in the future, scientists will improve existing and sufficiently developed approaches, including superconducting circuits, semiconductor circuits, cold atoms or ions. To create fundamentally different systems for quantum computing, including those based on atomic nuclei, will require colossal efforts and time. The scientist does not yet see any promising, but not yet studied directions.

“Our Russian colleagues are now saying that they are planning to develop all four directions within the framework of the Rosatom quantum project. However, in this case, we are talking about the fact that they will study what others have done. laboratories are needed for this. As I understand it, they will do this until they decide to focus on one thing, “added Smelyansky.

Not all of these efforts, according to a Google specialist, will bring results. In particular, he noted that he considers the adiabatic approach, which is gaining popularity among scientists and investors, much less promising and interesting than the creation of universal quantum computing systems.

“Investing resources in such projects is a big mistake. Quantum anneallers (as the prototypes of adiabatic quantum computers are called, – note TASS) can be used for physical experiments, but they are unlikely to be suitable for solving optimization problems. Recently, our group showed that adiabatic approaches in all realistic cases, they come to the answer no faster than the classical quantum Monte Carlo algorithm. In fact, quantum acceleration is not observed here, “the researcher explained.

New perspectives

Google itself, according to Smelyansky, plans to use universal quantum computers not only to test the fundamental laws of physics, improve the operation of information retrieval systems, train artificial intelligence and other abstract IT problems, but also in practice.

In particular, scientists are going to use quantum computers to create algorithms that control the operation of autonomous vehicles, as well as to improve the properties of components of lithium batteries and other types of batteries.

Sycamore plans to recruit the “heirs” to help biologists and biochemists from Google, as well as their partners at Harvard and other American universities, in order to significantly accelerate the search for various new materials and catalysts, as well as drugs for cancer and other diseases.

However, when and how all this will be implemented is still difficult to say, since scientists have not yet solved the complex technical issues related to the creation of a universal quantum computer with thousands of logical qubits.

“Now we are working and looking at these problems in the format“ I’m not fat, I would live. ”Therefore, our primary goal is simply to build a machine that would be able to solve such problems. Only then we will think about how all these parameters will be influence its work and develop metrics by which these changes can be assessed, “summed up Smelyansky.


Contact us: [email protected]

Our Standards, Terms of Use: Standard Terms And Conditions.