Serious quantum computers are ready to use. What they can do?

In a small laboratory in the lush countryside about a hundred miles North of new York city from the ceiling hangs a complex tangle of tubes and electronics. Is a computer, albeit messy. And it's not a regular computer. Perhaps it destined to become one of the most important in history. Quantum computers promise to perform calculations far beyond the reach of any conventional supercomputer. They can produce a revolution in the field of creation of new materials, allowing to simulate the behavior of matter down to the atomic level. They can bring cryptography and computer security to a new level, breaking the hitherto impenetrable codes. There is even a hope that they will bring artificial intelligence to a new level, will help him more effectively sift and process the data.

And only now, after decades of steady progress, scientists are finally close to creating a quantum computer, powerful enough to do what regular computers can't do. This beautiful landmark called "quantum supremacy". The movement to this landmark is headed by Google, followed by Intel and Microsoft. Among them — well-funded startups: Rigetti Computing, IonQ, Quantum Circuits, and others.

And yet nobody can match IBM in this area. 50 years ago, the company has made progress in materials science, which laid the foundations for the computer revolution. So last October, MIT Technology Review went to the Research center's Thomas J. Watson at IBM to answer the question: what is a quantum computer be good? Is it possible to build a practical, reliable quantum computer?

the

Why do we need a quantum computer?

This research center located in Yorktown heights, a bit like a flying saucer as it was in 1961. It was designed by architect Eero neofuturists Saarinen and built during the heyday of IBM as the Creator of large mainframes for business. IBM was the largest computer company in the world, and over ten years of building the research center she became the fifth largest company in the world after Ford and General Electric.

Although the corridors of the building looking over the village, the design is such that none of the offices inside with no Windows. In one of these rooms and found Charles Bennett. Now he is 70, he had big white sideburns, he was wearing black socks with sandals and even a pencil case with handles. Surrounded by old computer monitors, chemical models and, unexpectedly, a small disco ball, he remembered the birth of quantum computing as if it was yesterday.

When Bennett joined IBM in 1972, quantum physics was already half a century, but the calculations still relied on classical physics and mathematical theory of information that Claude Shannon developed at MIT in the 1950s. What Shannon quantified the information by the number of "bits" (the term he popularized, but not invented) required for its storage. These bits, 0 and 1 a binary code formed the basis of the traditional calculation.

A year after arrival in Yorktown heights, Bennett has helped to lay the Foundation for quantum information theory, which challenged the previous one. She uses bizarre behavior of objects at the atomic scale. In such scales the particle can exist in "superposition" of multiple States (many positions) at the same time. Two particles can be "entangled" so that changing the state of one instantly responds to the second.

Bennett and others understood that some types of computations that take up too much time, or even impossible, can be effectively carried out with the help of quantum phenomena. A quantum computer stores information in quantum bits, or qubits. Qubits can exist in superpositions of ones and zeros (1 and 0), and confusion and interference can be used to search for computing solutions in a huge number of States. To compare quantum and classical computers are not quite right, but, figuratively speaking, a quantum computer of several hundred qubits can produce more calculations simultaneously than there are atoms in the known universe.

In the Summer of 1981 by IBM and MIT, organized a landmark event called "the First conference on physics computing". It was held at the hotel Endicott House, a mansion in the French style near the campus of MIT.

In the photo, which Bennett did during the conference, on the lawn, you can see some of the most influential figures in the history of computing and quantum physics, including Conrad Susa, which developed the first programmable computer, and Richard Feynman, who made important contributions to quantum theory. Feynman kept on a key conference speech in which he raised the idea of using quantum effects for computation.

"The big push of quantum information theory has received from Feynman," says Bennett. "He said: nature is quantum, her mother! If we want to simulate it, we need a quantum computer."

A Quantum computer, IBM is one of the most promising of all, is right down the hall from the office of Bennett. This machine is intended for creation and manipulation of an important element of a quantum computer: qubits that store the information.

the

the Abyss between dream and reality

IBM Machine uses quantum effects that occur in superconducting materials. For example, sometimes the current flows clockwise and counterclockwise at the same time. The IBM computer is using superconducting circuits, in which the qubit consists of two different electromagnetic energy States.

Over head transmission line approach has many advantages. Hardware can be created using well-known established methods, and to control the system, you can use a regular computer. Qubits in a superconducting circuit easily amenable to manipulation and are less delicate than individual photons or ions.

In the quantum lab IBM engineers are working on a version of computer with 50 qubits. You can run a simple simulator of a quantum computer on a conventional computer, but with 50 qubits would be virtually impossible. And that means IBM is theoretically approaching the point beyond which a quantum computer can solve problems inaccessible to conventional computer: in other words, the quantum superiority.

But scientists from IBM will tell you that quantum superiority is an elusive concept. You will need to have all 50 cubits worked perfect, when in reality quantum computers suffer from errors. Also incredibly hard to maintain the qubits over a given period of time; they tend to "decoherence", that is to loss of its delicate quantum nature, like a smoke ring dissolves at the slightest breeze. And the more qubits, the more difficult to cope with both tasks.

"If you had 50 or 100 qubits, and they really would work well enough, and was totally free from errors you might make incomprehensible calculations, which it would be impossible to reproduce on any classic car, not now, not then or in the future," says Robert Selikoff, a Yale University Professor and founder of Quantum Circuits. "The other side of quantum computing is that there are an incredible number of opportunities for error".

Another reason for caution is that not quite obvious how useful it would be even a perfectly functioning quantum computer. It not only accelerates the solution of any task that you throw up. In fact, in many types of computing, it will be disproportionately the "dumber" of classic cars. Not many algorithms has been defined to date in which a quantum computer will have an obvious advantage. And even with them, this advantage may be short-lived. The most famous quantum algorithm developed by Peter Shor of MIT, designed to find Prime factors of the integer. Many well-known cryptographic schemes rely on the fact that this search is extremely difficult to carry out normal computer. But cryptography can adapt and create new kinds of code, not relying on factorization.

That's why, even approaching 50-cubitas milestone, IBM researchers are trying to dispel the hype. At the table in the hallway, which overlooks the lush lawn outside, is Jay Gambetta, a tall Australian, exploring quantum algorithms and potential applications for IBM equipment. "We are in a unique position," he says, carefully choosing her words. "We have a device that is the most difficult that can be simulated on a classical computer, but it is not controlled with sufficient precision to carry out through it certain algorithms".

What makes all ilimskom the hope that even an imperfect quantum computer can be useful.

Gambetta and other researchers started with an application that Feynman foresaw back in 1981. Chemical reactions and properties of materials are determined by the interactions between atoms and molecules. These interactions are controlled by quantum phenomena. A quantum computer can (at least in theory) to model them as can't conventional.

Last year, Gambetta and his colleagues at IBM used semicubical machine for accurate modeling of the structure of beryllium hydride. Consisting of only three atoms, this molecule is the most complex of all, which was simulated with the use of a quantum system. Eventually, scientists will be able to use quantum computers for the design of efficient solar cells, drugs or catalysts that convert sunlight into clean fuels.

These goals are, of course, still unimaginably distant. But, says Gambetta, valuable results can be derived from paired quantum and classical computers.

the

for the physics of the dream, engineer's nightmare