Quantum physics was born in 1900, when Max Planck suggested that energy is not absorbed continuously, but in separate portions - quanta. His idea was further developed: the Einstein photoelectric effect, the theory of Bohr atom, Rutherford experimentally showed what the atomic nucleus looks like, Louis de Broglie erased the boundary between waves and matter, Heisenberg and Schrödinger developed quantum mechanics.It’s hard to understand quantum physics — its mathematical apparatus is almost impossible to translate into “human” language. But to “touch” its manifestations in everyday life is quite real: lasers, flash drives, compact discs, integrated circuits or graphene - all these technologies appeared thanks to quantum physics. It is logical that they decided to use it for calculations - in quantum computers.Quantum computers are fundamentally different from ordinary computers: they process information an order of magnitude faster, and their memory is more exponential. Already now, experimental samples solve some problems faster than the most powerful supercomputers. The prospects for introducing quantum computers are beckoning. With their help, you can create new drugs, composite materials are stronger than titanium and lighter than plastic, superconductors that work at room temperature, achieve absolute encryption security or develop universal artificial intelligence. But in reality, everything is not so rosy. This is because we do not yet understand what a quantum computer really can do .Anatoly Dymarsky (Skoltech) is a theoretical physicist who works in the field of quantum systems physics. Anatoly will tell how the quantum computer differs from the usual one and what the possibilities of the IT industry promise.How does a regular computer work?
To explain what a quantum computer is and how it works, you need to start from afar and tell how a regular computer works. The operation of a conventional computer is determined by two parameters: memory, speed of computation.Memory is the main characteristic of a computing system. A computer can read, write and process information that is stored in memory.The computer performs the simplest operations: multiplication, subtraction, addition of numbers. If you perform these operations a lot and quickly, then you can combine them into a program that processes information. This is how databases, search, or neural networks work. Speed of calculations or speed of execution of operations (FLOPS) is important here .There is a third (additional) parameter - determinism,general characteristic for all computing systems. It means that all machines work according to a program that is unique - zero is always zero, and a unit is definitely a unit. No other interpretations are provided and there is no element of uncertainty.Uncertainty can be introduced only at the input level, for example, by random numbers. The input may be random, but the program always unambiguously processes all incoming data.How does a quantum computer work?
It works differently - by intuitively incomprehensible logic. Like the usual one, it performs calculations, but it is based on the laws of quantum mechanics .The classical world and classical mechanics are deterministic. This means that the value of any memory register in the computer is always 0 or 1, and the plate is always either whole or broken.In a quantum-mechanical system there is no such clarity, but there is a probability that determines its essence. The right question here is: what is the probability that the plates are broken or intact, what is the probability that the values of the register are 0 or 1?Probability is the first important concept in quantum mechanics . From the point of view of quantum mechanics, the Schrödinger plates are both whole and broken. There is a certain probability that they are whole, and some probability that they are broken. This uncertainty reflects the real physical world.At the classical level, uncertainty disguises our ignorance. For example, when we buy a Sportloto lottery ticket, we are likely to win because we do not know the winning number.For classical physics, a lottery is not a probabilistic process. You can always describe the movement of the hand that launches the drum, the speed and trajectory of each ball. Theoretically, you can guess the winning number (although in practice it is difficult). In quantum mechanics, even theoretically, one cannot guesswhat will happen the next second. We can only predict this in terms of probability.The second concept is the principle of superposition . A regular bit is found only in the values 0 or 1. In quantum computers there are no ordinary bits, but there are quantum bits - qubits . The quantum bit is in state 0 or 1 with some probability. A qubit can be simultaneously in these states, moreover, in different combinations - in a superposition of these states.When the system (qubit) is simultaneously in the state 0 or 1, we can only talk about probabilities. If there are many states, the system is simultaneously in all possible states, but with a lower probability for each. This is fundamentally important.In a classical program, at each particular moment in time, each line of the program works with a specific memory cell. In quantum mechanics, you can work with all memory cells at the same time ."Memory" of a quantum computer
What is the main difference between quantum and classical computer memory? On a regular computer, we write numbers in binary code. For example, the number 8 in the binary system looks like 00001000, and 4 bits are enough to write it.In quantum computers, qubits are in state 0 or 1 with some probability. Probability is a number. To write a single number with infinite precision, you need an infinite number of bits. Therefore, in theory, one qubit is a physical system with an infinite amount of memory .In practice, measurement methods have limited accuracy. We assume that it corresponds to the usual machine (float). It turns out that the qubit contains two numbers: the probability that the qubit is in state 0 and in state 1.Note: for simplicity, we ignore that the sum of the probabilities of a qubit in state 0 and 1 should be equal to one. The main conclusion does not depend on simplification.One qubit corresponds to two real numbers (float). This is a big win, because for two real numbers on a regular computer you need two machine words - 128 ordinary bits, and we managed with one quantum. It may seem that a quantum computer is 128 times better than usual. But this is not so.A quantum computer is exponentially better than usual.
One qubit is 2 real numbers. Two qubits - 4 real numbers. But eight qubits is 256 potential configurations of eight zeros and ones - two to the eighth power.For one qubit, the gain is 128 times, and for eight qubits it is much larger - 256 * 128. A system of n qubits in memory is equivalent real numbers.The capacity of quantum memory is growing exponentially.
The memory of an ordinary laptop is equivalent to 15 qubits, 40 qubits are equal to the memory of the most powerful data centers, and 50-60 qubits exceed the total memory of all data centers around the world.Three to four qubits is equivalent to an increase in ordinary classical memory by 10-20 times. Quantum memory is much more capacious than any other classical methods of recording information. This is the main potential of quantum computing.But an exponential increase in the capacity of quantum memory causes a dimension problem . Because of the curse of dimension, it is difficult to describe such a quantum system on a classical computer — more and more memory is required.What tasks can a quantum computer solve?
If the quantum world operates at a level of uncertainty, how is it possible to calculate anything at all? Quantum mechanics has a probabilistic nature, and we need an exact answer. How will everything work if you just need to multiply two numbers?I will explain by the example of problems of the class NP , that is, solvability problems whose solution cannot be found in polynomial time - in any case, under the assumption. However, the correctness of the solution in polynomial time can be verified. This is similar to breaking a closed lock: we do not know how to use master keys, but we can quickly check any key, if any.Thanks to the principle of superposition, the quantum system is immediately in all states and is looking for the best option. The system does not give a definite answer, but it increases the likelihood that the best option is a solution. When the system stops at some solution, we can quickly check it for correctness.If it turns out that the answer is incorrect, start the quantum computer again. The probability of getting the right answer is more than 50%, and often much more. So, in 2-4 starts of the quantum algorithm, we get the correct answer.We will not have a definite answer, but only the probability of getting the right answer. But this probability is very high. In fact, we are guessing, but not on the coffee grounds, but on the scientific one. In a few iterations, we will find the answer and verify that it is correct.Quantum Computer Parameters
A classic computer has two quality parameters: the amount of memory and the number of operations. With a regular computer, we assume by default that we have access to all memory locations for writing and reading.In the quantum case, there are three parameters.The amount of memory or the number of qubits . The more memory, the better? For a quantum computer, no - when we increase the number of qubits, the complexity of the quantum system grows. The system becomes difficult to maintain in an isolated state.Operating time or number of sequential operations (coherence). The system must be maintained in an isolated state - in physics this is called coherence. If we allow the quantum system to interact with the environment, then this will destroy the state of the cells of quantum memory. Instead of zeros and ones there will be just noise.We try to keep the system isolated as long as possible. But the more quantum operations we carry out, the more time is spent on them, which means that it is more and more difficult to maintain the system in an isolated state.Note: here the number of operations is not per second, but for the entire system operation time.A paradox arises: the more qubits, the less operations are available . Therefore, the time during which you can keep the system isolated and perform a certain number of operations is an important parameter.Imagine a regular computer in which there is no cooling. Until the computer overheats, he has time to count something, and then he turns off. Roughly the same thing happens in a quantum computer. It does not have a “fan”: the more it works, the more it heats up until it collapses. Therefore, there is a limit on the number of operations.Universality. In a classic computer, any operations are available: multiplication, division, subtraction. Theoretically, in quantum too. But in practice, it is much easier to conduct operations only with neighboring qubits, which are located on a straight line, in a rectangular or square array. To work with all qubits, a very complex architecture is required - in practice, they still do not know how.
All three areas conflict with each other. We can improve one, but this will happen due to the deterioration of the other two. Now that the technology is in its infancy, several prototype platforms can be distinguished, and each of them is trying to improve the performance of one direction at the expense of the other two.Prototypes
I will highlight three prototypes that large companies are working on. Google, IBM, Intel, Microsoft are investing in the development of quantum computers. Together, they have invested more than $ 500 million in development, laboratories and research centers.The first classic computers occupied entire rooms, worked on vacuum tubes and were so heated that they needed separate powerful cooling. Quantum computers are very similar to them - these are cabinets 3 meters high, most of which are occupied by cooling systems. Computers cool to a temperature close to absolute zero so that quantum systems can perform their computing functions.Universal quantum computers
These are universal machines from Google and IBM with about 20 qubits of memory. They perform any operation, because complete universality is available with a relatively small number of qubits, and then a practical limitation arises. Perhaps in a year people will learn how to work with 30-40 qubits.Universal quantum computers are able to implement arbitrary quantum algorithms, for example, Shor and Grover algorithms.Modern cryptography is based on the decomposition of numbers into prime factors. It is currently unknown whether a polynomial non-quantum algorithm exists for the factorization problem. However, 25 years ago, Peter Shore published an article explaining how a quantum computer can decompose a very large integer into prime factors.The quantum computer algorithm does not work deterministically, but guesses simple factors with a probability of a correct answer of more than 50% and finds simple factors exponentially faster than a regular one.With the spread of quantum computers, all modern encryption methods will be vulnerable, and this is the main motivation in the development of quantum algorithms over the past 25 years. But for now, it is still difficult to apply the Shore method, because the algorithm requires a large quantum computer. Small ones solve the problem only for small numbers.Another example that demonstrates the potential of quantum computing is the Grover Algorithm for the task of searching or finding a solution to an equationwhere some kind of complex function.In addition to the Shore and Engraver algorithms mentioned above, there are a large number of other quantum algorithms. Any physical system wants to go into equilibrium - quantum is no exception. From a scientific point of view, it is more correct to speak not about equilibrium, but about the basic state of the system. The classic analogue is the state of rest. The system always seeks to go into a state of rest with minimal energy. In terms of computational problems, it is an optimization task of minimizing energy. A quantum computer can just solve such problems.The entire field of applicability of quantum algorithms and computers is not yet understood. But there are already dozens of different optimization problems that quantum computers and algorithms can handle, and new ones are found.Quantum Simulators of Limited Versatility
This is another direction: universality is limited, but isolation (coherence) is maintained. These are computers at the turn of 50-70 qubits, which in the sense of memory is already more than any supercomputer.At this boundary, the capabilities of a specialized quantum computer are superior to the capabilities of a classical one — quantum superiority arises . This means that quantum computers can solve some problems that ordinary (even supercomputers) will take tens, hundreds or thousands of years to complete.In October 2019, Google announced that it had achieved quantum superiority. The news appeared in all leading newspapers and magazines, the corresponding scientific article was published in Nature. Featured articles have been published by many newspapers, even the New York Times and the Wall Street Journal, which are far from science.In reality, Google has developed a quantum processor with limited versatility. It has a fairly large number of qubits, and it can perform some narrow tasks better than any classic computer. Another question is that these are very narrow and artificial tasks.Incoherent processors with the number of qubits from 2 thousand
If you forget about universality and coherence, you can add 2 or even 3-4 thousand qubits. The D-Wave company from Canada is engaged in this area. They have processors with a thousand qubits, but without coherence.Possible applications of quantum computers
One big potential application is cryptography. The second is optimization tasks that arise in a variety of areas.The science. Quantum computing can help predict particle behavior, model DNA molecules, or develop new drugs. For example, they are trying to apply quantum computing in pharmacology. To do this, you need to understand what form different proteins take (which you can think of as microscopic quantum objects). We don’t know how they will behave, but the easiest way to understand this is to simulate their behavior on a quantum computer. This scientific task has a huge business potential: new drugs, supplements, antibiotics.New materials.In the science of materials, the main thing is to understand the interaction of atoms, which can be modeled on quantum computers. This is also a scientific task, but having created new material, it can already be sold.Machine learning and artificial intelligence. Machine learning is a complex process that requires a huge amount of computation. While there is no practical benefit from quantum computers, because they are now at the wrong level of development. But in the long run, quantum computers can speed up standard algorithms. In some cases, this looks revolutionary, because you can reduce the training time of a neural network by tens of times.Transport, energy, logistics.In these areas there are many optimization problems. For example, in the energy sector, the main problem is the distribution of electric energy throughout the country. The price of electricity in different regions is different, while during the transmission part of the energy is lost, and with it the profit. To make more money, the business is trying to optimize the transfer. This is one of those tasks that is in the NP class. It is hard to find the right solution, but a quantum computer can help.Business applications. In business, only large companies and corporations are involved in quantum computing. Giants have money and resources, for example, Google, D-Wave or IBM (the leader of the field with great achievements).On the website of the company D-Wave it is written that already in 150 business applications, quantum computing is used. IBM has released a brochure that discusses what can be done with a quantum computer. These are dozens of different industries and potentially hundreds of business solutions. So it looks on paper.
In reality, everything is a little different. The development of technology is not yet at the level to put it into practice.What does the quantum revolution mean for the IT industry?
Nothing so far. We are in the so-called NISQ era - Noisy Intermediate-Scale Quantum technology . This means that now there are no such quantum devices that could compete with classical computers. It is not yet possible to create a quantum system that in all respects surpasses the classical: rather small, universal and isolated. So far, only systems have been obtained that perform highly specialized tasks of a certain kind better than a computing cluster. Quantum technology is not yet practical. I would like to use this huge potential for my daily tasks, but I don’t know how to do it.Quantum technology has a huge "disruptive potential." If you learn to solve well at least one of the optimization problems mentioned above, this will change one specific industry, at least. I hope that in 5-10 years the situation will change in some areas.Many companies create prototypes of real quantum computers - they already know how to do something, but so far this is not enough.In Skoltech, we are trying to answer the main question - how and why you can use a quantum computer. With my colleagues Vladimir Antonov and Oleg AstafievWe are working on a project in which we are working on a small quantum computer. Unfortunately, some of the architectural and design issues have not yet been resolved, because we are still not sure which tasks this computer will have to solve. If you are interested in this question, I invite you to discuss it .The interest with which the HighLoad ++ participants received the report on quantum computers and nuclear power plants prompted us to pay more attention to such topics at our conferences. Therefore, at RIT ++ in May online we will have sections of the scientific field and the application of IT in related fields. And this is only a small part of the novelties of the festival “Russian Internet Technologies” - for more details, see the website and the newsletter .