“A classical computer will decompose a number into 2,048 bits in 1,000,000,000,000 years. A quantum computer - in 10 seconds"
—
— The creation of a quantum computer is one of thefundamental problems of physics of the XXI century. Last week, even the Nobel Prize was given to physicists for demonstrating quantum entanglement, the principle behind quantum computers. If you know about Moore's law (the number of transistors on an integrated circuit chip doubles every two years - ed.), then in recent years it has ceased to be fulfilled, and even microprocessor manufacturers have moved away from such a thing as a technical process. Nanometers, which everyone is talking about now, are more of a marketing thing.
Now there is a new development branch in lithography -extreme ultraviolet, where they shine at a wavelength of 13.5 nm. This is a record wavelength that can be obtained stably and make chips in the 2-3 nm limit, reducing the diffraction limit with various optical tricks. But what to do next is unclear. A dead end is possible in the reduction of transistors on the horizon of 5-10 years.
Danila Shaposhnikov
This is where the fundamental difference can help.quantum and classical computing. Classical ones are sequential, and quantum ones naturally allow you to do completely parallel calculations. That is, each quantum bit can compute in parallel with the other quantum bits of the system. In this case, a bit can have several states at the same time - be both zero and one. Or even a multi-level system, but the mainstream now is a qubit, it has two levels. Computing power grows exponentially with the addition of qubits to the system (2n). And in the usual system, it grows quadratically (n2).
Modern science is in the stage of understanding,what is quantum mechanics. All the laws of particles, the interaction of atoms among themselves is described by the laws of quantum mechanics. This science is different from what came before it. For example, in quantum mechanics there is the superposition principle, due to which the dimension of the state space grows exponentially.
A classical computer just can't do it.simulate. A quantum computer itself is built on such phenomena and is able to work with such systems. Plus, in the quantum mechanical system there are probability amplitudes with complex numbers - ordinary computers do not have this.
If we take the problem of expanding some number into2048 bits, then the classical algorithm will decompose it in a thousand steps and in 1,000,000,000,000 years. And Shor's algorithm, if there were a quantum computer with the right number of qubits, would do it in 107 steps - about 10 seconds. So far, there are no such quantum computers, but those that are already able to do what a classical computer would take a huge amount of time to do.
- Will quantum computers justify the hopes that have already been placed on them?
Let's first understand what it takes to create a quantum computer. Physicist David di Vincenzo correctly articulated five basic criteria:
- Define what a qubit is. They are different, today there are several well-known platforms - on atoms, ions, superconductors, photons.
- Be able to introduce a qubit into a superposition.Understand how to make a qubit be both zero and one at the same time. In each of the platforms, the introduction to superposition is a separate task, and this can be done by different physical principles.
- It is necessary to create qubits and quantum entanglement between them, to be able to control them, to build gates based on them.
- Maintain this coherent state as long as possible.
- Make measurements on our quantum computer.
Behind each of these phenomena there is a lot of engineeringdifficulties. For example, if you measure a qubit, its state will change and it cannot be cloned. Or noise, electromagnetic waves, particles have a bad effect on the system, so most platforms cool the entire system to low temperatures to minimize the impact of noise and dust. But working in cryogenics is much more difficult. All this complicates the creation of quantum computers, so now there are a maximum of about 130 qubits. For example, IBM released a 128-qubit system.
There are many engineering complexities behind every step in the development of a quantum computer.
But there are not only physical, but also logicalqubits. What is the difference? The accuracy of quantum computing should be about 99.9999999999999% - then we consider it to be very high. But today it floats from 90 to 99%, these are very low parameters, it is difficult to calculate exactly with their help, the percentage of errors will be high. To achieve the desired level, they make logical qubits, that is, they make one logical qubit from a large number of physical qubits, program error correction protocols, an algorithm on it, and it turns out that this is one qubit with a high accuracy rate.
Therefore, if we return to physical qubits,of which a quantum computer should be made - the industry is at an early stage, approximately at the level of ten logical qubits. In the coming years, we expect that a level of one hundred logical qubits will be achievable. This will already allow doing interesting things - route optimization, clinical tests, synthetic creation of clinical data, proximation of quantum simulations, optimization of financial portfolios. For comparison, to crack RSA algorithms, you need about a thousand logical qubits.
Here we need to make a small digression andTo say that today in quantum computing there is one more difficulty in a row - until quantum memory is invented. Therefore, in the next 10 years, quantum computing will work in conjunction with classical computers.
The strategic long-term goal is to create a universal quantum computer. This requires more than 10,000 logical qubits, reliable control of multi-qubit gates, and quantum memory.
What will quantum computers change?
— They can solve a huge range of problems— for example, for biosciences. Currently, we cannot model even moderately complex molecular compounds. That's why scientists make synthetic molecules and constantly experiment. Simulations are severely limited by the size of molecular systems and accuracy parameters. Because of this, it takes ten years to create a new drug. And a quantum computer that can simulate a quantum mechanical system will radically speed up the process.
Or they are trying to do protein folding nowX-rays, tricky magnetic resonances. And if there is a quantum computer, it will be able to simulate this system, and we will simplify our life in creating drugs. The development of new materials for space flights, engines, and superconducting systems will also accelerate. New electrolytes for batteries will appear, which have been at the level of 200-250 Wh per kilogram in terms of energy density per mass for 20 years. We can’t do better, because we don’t model well yet.
It’s impossible to even list everything in one interviewthose applications of quantum computers that can be imagined. Even if he can simply speed up a few processes of important operations (such as the Fourier transform), this will already be serious progress. And this is only one step towards creating a universal quantum computer. That's why there's such hype.
— But they can only be used within the bounds of science?
- No, in any kind of optimization - for example, where graph theory is used. They are already being used to optimize financial portfolios, routes, and optimize AI algorithms.
“Qubits are good, but this does not mean the speed and accuracy of the calculation”
- Are there any other problems that it is not clear how to solve? What can stop progress?
- The main one is the creation of qubits in a largenumber and their binding, the lifetime of the entire system. For example, if the system lifetime is 0.001 seconds, then you may not have time to calculate something important. We need to think about how to maintain the quality of calculations and scale them.
Let's take the company IonQ - they invested in itrespected investment funds from all over the world, it even went public. They make systems with ions, and the problem is that there are ion traps, but there is a limit to the number of ions that can be caught. And we need to come up with a mechanism for linking traps to each other. There are still big problems with this - it greatly hinders the scaling of the system. Other platforms have similar severe problems.
There are still problems with the equipment - sometimes underquantum computers need to invent new devices. For example, special optics, lasers, vacuum equipment, cryogenic chambers. There are many problems, but this is the path of development - microelectronics has already passed it. This is normal: the industry adapts to each new process and invents new conductive metals and other discoveries. It's just that the whole system is still at an early stage of maturity.
The main problem in creating quantum computers is the creation of qubits in large numbers and their binding, the lifetime of the entire system
- As non-specialists who are interestedquantum computers, to understand whether a new discovery is really a step forward for this industry or another news for the sake of clicks? What to pay attention to? For example, is the number of qubits an indicator?
- It's better to try to figure it out on moredeep level. If you don’t understand at all, these benchmarks will very superficially reveal the essence of progress, and sometimes even mislead you. As, for example, with the number of qubits - in fact, this is good, but it does not say how much the system can calculate and with what accuracy.
For me, the number of interconnected logical qubits, the accuracy of the calculation, the lifetime of the system, and the ability to calculate practical algorithms are important.
— The development of quantum computers is a long time,expensive and difficult. Therefore, it seems that a very limited number of organizations are doing this. Doesn't this mean that such devices will work only for the benefit of corporations and states?
- Those who made a more or less working machine,usually open to it cloud access. And you can write your own quantum circuits and calculate algorithms. Each developer is interested in increasing the number of practical tasks that can be done on their quantum computer, so the cost is reduced.
Based on the number of investments in the sector, one can makethe conclusion is that there is progress. This is an indirect parameter - if hundreds of investors invest and the industry grows, this speaks volumes. And since 2019, the number of investments has been growing - from $300 million to $2.3 billion. Apparently, we are close to solutions that will become practical.
But at the same time there are only 80 organizations thatmake quantum computers. But the figures say that 1.5 billion were invested in hardware. Of these, 12 companies took the lion's share. Specialists are needed here in quantum physics, mathematics, engineers are in great demand. An interesting fact: the Soviet school is considered strong here. We talked to many of the 260 active companies in this field - 20% of them have Russian engineers, physicists or mathematicians.
“The number of qubits does not say how much the system can calculate and with what accuracy”
“Russian scientists are 3-5 years behind the world scientists”
— And what about quantum technologies inside Russia?
- Not really.Russia has a program and a roadmap for the development of quantum technologies with a budget of about $1 billion until 2024. The program is divided into several roadmaps - quantum computing (supervised by Rosatom), communications (Russian Railways and the Center for Metrology) and sensors (Rostec). Gazprombank is also in this whole game, because they are the main investor in the quantum center. For example, a special quantum communication line between Moscow and St. Petersburg has already appeared - this is the main protocol for quantum cryptography today.
Probably the main players in quantum computing are the RCC, FIAN, and Moscow State University.
What developments do they have worth talking about?
- According to the roadmap, they make quantumcomputers on different platforms - atoms, ions, photons, superconductors. According to my feelings, they are 3-5 years behind the world companies. But they have serious staff and approach - they will definitely develop something useful.
— Researchers are afraid that the technology will get out of control? Are they trying to regulate it already?
- We are still on the way to regulation, while everyone is concernedcreation of hardware. As soon as something serious appears, it will come to restrictions. But everyone is afraid for their data. For example, it is now possible to secure data with quantum encryption and reduce the likelihood that a quantum computer will be able to crack it. But if someone has copied the data and is waiting for a quantum computer to appear, he will be able to decrypt it later. Now this is the main concern.
Read more:
Catapult sends NASA satellites into the sky
A giant magnetic storm is approaching Earth
Recreate the Sun on Earth: how physicists solved the main problem of thermonuclear fusion