The future has come: when without quantum computers it will not be possible to manage
The team of Mikhail Lukin created in 2017 one of the most powerful quantum computers. With the help of the scientist, RBC understands what are the criteria for success in the quantum race and when it is worth waiting for quantum superiority
Twenty years ago, quantum computers were considered fantastic, and soon they will surprise us no more than a regular PC. "I think that in five or ten years already in many areas of human activity without quantum technologies it will be impossible to get by," says Harvard professor Mikhail Lukin, whose team in 2017 created one of the most powerful quantum computers.
Mikhail Lukin left for America about a quarter of a century ago. In 1993, a graduate of the Faculty of Physical and Quantum Electronics, MIPT, invited Marlan Sculley, a world-renowned researcher in the field of quantum optics, to the postgraduate course at the Texas A & M University. In Texas in 1998, Lukin defended his thesis on the use of lasers to control the environment. But his main scientific experiments, Mikhail Lukin did in the next decade at Harvard University. Here he became a professor of physics, then co-director of the Harvard Center for Quantum Physics and the Center for Ultracold Atoms.
"I was very lucky: at Harvard, I was on special terms. An ordinary postdok (a scientist who recently received a PhD, which roughly corresponds to the Russian candidate of science - RBC ) must work in one scientific group and engage in a specific, highly specialized project. I also had complete freedom, "Lukin told RBC.
Lukin says that he and his colleagues have many times been invited to work for corporations involved in the race to create a quantum computer, but he invariably refuses: "I would say that until now the most creative activity in this area is still going on in universities."
In the atmosphere of "working permissiveness" during the last 16 years, the scientist and his team conducted experiments that hit the scientific world: like stopping light or creating photonic molecules - matter similar to light swords from "Star Wars" - and temporary crystals, structures, before existed only in theory. During these years, he also nurtured the idea of an experiment on quantum computing, which in the summer of 2017 glorified Lukin and his laboratory around the world.
Quantum Informatics
Even in the early 1990s, nobody took seriously the idea of creating quantum computers even in the scientific community, Lukin said: "But then there were two, if I may say so, revolutions."
In 1994, American Peter Shore developed a quantum factorization algorithm, later named after him. "Multiplying two prime numbers, even very large ones - is simple, and finding a prime factor divides a large number is a very difficult task for a computer. Factorization underlies all modern cryptography, "explains Lukin.Conventional computers are capable of hacking modern cryptographic systems, but they take so much time and resources for it, that the result is useless. A quantum computer can solve such problems almost instantaneously, and the Shore algorithm became the first proof of the practical meaning of creating such devices. "Secondly, at the same time, there have been major shifts in experimental physics: scientists have learned to cool atoms well, isolate individual particles," Lukin continues.
In the same turning point for quantum computers in 1994, a scientific article by two European physicists, Peter Zoller and Juan Ignacio Sirak, appeared in which they described a quantum computer using an ion trap. "Quantum informatics was just emerging, other researchers had only abstract ideas of quantum computers, no one seriously even thought about whether it can be done or not. The publication of Zoller and Sirak changed everything: it became clear that it was possible to build a quantum computer, and even a specific proposal appeared, "Lukin recalls.
With the authors of the article, Mikhail met in the early 2000s: "They were already famous people, and I - young beginning scientists. But it turned out that our ideas are very similar. We joined forces and wrote a series of articles that theoretically described the ideas that formed the basis of our current practical work. "
In the 2000s, many scientific groups began to conduct experiments on superconductors-materials that at low temperatures completely lose electrical resistance. Lukin's group, in turn, decided to try to focus on "cold atoms" - particles cooled to almost zero and placed in optical traps created by lasers. If the necessary conditions are met, they can be used as sufficiently stable quantum bits (qubits).
To make a real quantum calculator Lukin in the middle of the 2000s did not dare: the project seemed too risky, there was not enough technological base. For several years his group at Harvard studied other ways to make qubits for a quantum computer - for example, from impurities in a diamond. Other practical projects appeared from such studies: for example, former students of the professor came up with how to make quantum sensors for medicine from diamonds.
In 2010, quantum computing ceased to be discussed exclusively in the laboratories of scientific centers - they were seriously interested in large IT companies.
The present quantum
Several years ago, not only IBM, which had been studying the field for a long time, but also Google, Intel and Microsoft that had not been seen before, announced its intention to build working prototypes of quantum computers.
In this case, the Canadian company D-Wave since 2011 has already produced and sells "real quantum computers" - first with a power of 16, then 28, and after a couple of years - 512 kbits. Today the company offers already 2000-qubit computers. D-Wave has a serious pool of customers: Google, NASA, Lockheed Martin, Volkswagen Group. An uninitiated person may think that the quantum future has already come - and yes, and no.
D-Wave produces so-called adiabatic computers - to understand their differences from full-fledged quantum computers, one has to read at least a short course of quantum physics. In the applied sense, the difference lies in the fact that D-Wave computers can solve only a very narrow range of tasks related to optimization. In Google, for example, for a computer, D-Wave picked up one task, which the adiabatic computer decided millions of times faster than the classic one. But it was not possible to derive real benefit from this, and for solving other problems the machine is not intended.
Successes in the field of creating "real" quantum computers are more modest: until recently, their capacity did not exceed 17-20 qubits, and Lukin says that a couple of years ago he did not believe in the possibility of creating a device with more power. But in the summer of 2017, Lukin's group reported the creation of a working prototype of a quantum simulator for 51 qubits, and just a month later a group of Professor Christopher Monroe from the University of Maryland announced the creation of a simulator of 53 qubits. Devices and the results of the first experiments conducted on them are described in an article published in the journal Nature in late November.
Atoms in optical traps and superconductors are today two advanced technologies for the creation of quantum computers, Professor Christopher Monroe told RBC. "Both approaches are now at a stage when we already have a clear idea of how to build quite large devices, and there are ideas how to scale them," he said. "Superconductors are still showing lower performance, but since the qubits here are printed on a chip, they are easier to scale." It is easier to work with atoms, because each atomic qubit is identical by definition. There are other similar technologies that catch up with us, including the neutral atomic qubits that the group of Mikhail Lukin is doing. "
Race for qubits
The number of qubits seems to be a simple and understandable criterion of success, but in quantum physics nothing is simple and understandable. The number of qubits is only one of the three "axes" on which a quantum computer is built, explains Professor Lukin. The second is coherence, the ability of the qubits to be in a state of superposition (think of the Schrodinger cat), to be both zero and unit at the same time - the whole theory of quantum computation is based on this phenomenon of quantum mechanics.
This ability determines the time during which the machine can work: the longer the coherence time, the more calculations the computer is able to perform. "If you have a million qubits, but you can not do enough operations on them, then you will not get a quantum computer. For example, in D-Wave computers, each of the original qubits has such low coherence that it is not clear whether there are any quantum properties there at all, "Lukin said.
Finally, the third "axis" is the degree of programmability, it describes how many tasks of a different type can be solved with the help of a quantum computer, Lukin continues. "Our simulator has good enough coherence and a fairly large number of qubits, but all this is in other systems. What is important is that we managed to make the system highly programmable, "he says.
The difference between a quantum simulator and a universal quantum computer is that the first one can be programmed to perform only a certain type of tasks, explains Professor Monroe: "But the beauty is that the simulator can be turned into a universal computer in the future." True, it is not always possible to draw a clear line between them, Lukin adds.
"A quantum simulator, which can be programmed arbitrarily, becomes universal. It turns out that the boundary between the computer and the simulator is very blurred, and now it is unclear whether it is possible to determine it at all. But this is normal, we are now literally at the forefront of science, and this is happening with all the new phenomena, "the scientist explains.
Optimism without evidence
Even scientists are not yet going to outline the whole range of tasks in which a quantum computer will surpass the ordinary one. "Shor's algorithm is unique in some ways, because this is one of the few tasks that we know for sure that a quantum computer will cope with it better than usual, it's proven. There are many other very promising algorithms, including those for the same combinatorial optimization, for which there is as yet no evidence, "Lukin says with his hands.
On the one hand, it is the Shore algorithm and the inevitability of quantum cracking of cryptographic information security systems that attracted large state money to this sphere. Leading in this sense is China, which recently promised to invest $ 11.5 billion in the construction of a new quantum center. On the other hand, the decoding of codes will become an important, but a small part of what quantum computers can do, Lukin hopes. "I do not like Shor's algorithm, that he has basically destructive power. However, I am sure: before it is implemented, a quantum computer will have time to bring a lot of benefits to humanity, "he says.
In an article published in the end of November in the journal Nature, scientists said that they were able to see the formation of quantum crystals - a material that can be used to create quantum memory in quantum computers. "What we did, it's impossible to directly simulate on classical computers, from this point of view, we can say that quantum superiority has already been demonstrated," Lukin said. "This is important for science: we have already entered the limit when quantum computers are beginning to benefit."
It is believed that quantum superiority will be achieved when quantum computers will cope with practical tasks better than classical supercomputers. The power of classic computers is constantly growing, but there is a class of tasks that they will not be able to cope with anyway, and this can not be corrected by simply building up computing capabilities, Lukin explains. Among them, for example, the problems of combinatorial optimization, which exist in any field.
"A classic example is the traveling salesman's problem. Let's imagine that Aeroflot wants to optimize the routes of flights so as to spend less fuel and at the same time cover a large territory and make departures convenient for passengers. A classic computer does not cope with this type of tasks, they are too complex for it, too many answers. All that he can - in turn to sort out different options, it takes a huge amount of time and requires a lot of power, "- explains Lukin.
A quantum computer is able to sort out these variants not in series, but in parallel, which fantastically speeds up the calculation process - literally minutes instead of years. An effective solution to such problems is extremely important for modern fields of informatics, for example, for artificial intelligence or machine learning, Lukin adds.
Among other possible applications of the quantum computer physics are the modeling of new materials with specified properties and different chemical processes. "Even simple chemical reactions are very difficult to model on classic computers, because there are so many options for their flow," explains Lukin. "Quantum computers probably will be able to do this." And increasing the effectiveness of any chemical reaction is literally a couple of percent capable of creating a new industry. " He agrees with Monroe: he sees the main prospects for quantum computing in logistics, the creation of new materials and medicines in pharmaceuticals, as well as in a wide variety of optimization.
Quantum Internet
One of the main problems that physicists and engineers have to solve is the scaling of quantum computers. "Today, we do not exactly know how to scale these systems beyond about 1 thousand qubits. There are different ideas, the most promising of them, in my opinion, is the idea of modular architecture, "says Lukin. "Instead of adding more qubits to a single machine, we create a network of quantum computers." Each calculator with a power of a couple of hundred qubits is connected to something like a "quantum Internet". Several groups now work on similar concepts, including Lukin's group, but all are at relatively early stages.
About 30 people work in the Harvard group of Mikhail, but over the quantum simulator - much more: it was created by the joint efforts of three scientific laboratories. In total, according to Lukin, there are about ten such centers in the world where development is at the forefront of quantum technologies. Most of them now go from pure physical experiments to practical developments, and the role of corporations is growing ever more. "In addition to pure science, it is now necessary to solve engineering problems that can be clearly put, and this is much faster and more efficiently done in companies, and not in universities," says Lukin. "We already know how to build a sufficiently large quantum computer, now we need to make it so that the system does not work at the level of" only the graduate student will understand, "but at the level" came, turned on, works. " It is in this,
In the next five years, a lot of working quantum machines will be created, Monroe says. And in ten years there will be a full-fledged quantum computer programmed by people who do not know and do not particularly care about how it is arranged inside, he says: "It is then that the search for his real practical applications will begin." Now universal quantum computers for several dozen qubits can work only with artificially created algorithms, Monroe continues: "And this is not so interesting, because such a small system can easily be modeled on an ordinary computer."
Quantum computers are at the same stage as the first classical computers at one time, Lukin says: "This is often said by Peter Shore himself: then there were also some ideas about algorithms that might work efficiently, but maybe not". When the first classical computers became real devices, scientists and engineers began testing these algorithms on them, and many of them turned out to be very effective, Lukin says: "I think the same will happen with quantum algorithms."
Will a quantum computer become the same device as a normal PC? While nobody knows this, everything will depend on concrete examples and applications that can become a part of our life, Mikhail Lukin answers. "Who would have thought even 20 years ago that it would be a real computer," he concludes, pointing to the cell phone in front of him.
Post a Comment