"A cool computer will spread the number of 2,048 bits in 1,000,000,000 years, and a quantum computer in 10 seconds."
Why is everyone talking about quantum computers?
The creation of quantum computers is one of the fundamental problems in the physics of the twenty-first century. Last week, even the Nobel Prize was given to physicists for demonstrating quantum complexity, the principle underlying quantum computers. If you know about Moore's law, it has stopped in recent years, and even the producers of microprocessors have moved away from the notion of the process. The nanometers that everyone is talking about now are more of a marketing thing.
Now there's a new branch of development in the lithography, an extreme UV, where the wave is 13.5 nm long. It's a record wave length that you can get stable and make chips at a limit of 2-3 nm, reducing the diffusion limit with various optical tricks, but what to do next is not clear. There may be a dead end in reducing transistors on a 5-10-year horizon.
So every quantum bit can be calculated in parallel with other quantum bits of the system. The bit can have several states at the same time, be both zero and unit. Or even a multi-level system, but the majors are now cubit, it has two levels. The computational power increases exponentially with the addition of cubics to the system.
Modern science is at the stage of understanding what quantum mechanics is: all the laws of particles and the interaction of atoms between them are described by the laws of quantum mechanics; this is different from what was before it; for example, quantum mechanics has the principle of superposition, which increases exponentially the dimension of state space.
The classic computer simply can't model it, and the quantum computer itself is built on these phenomena and can work with these systems. Plus in the quantum system, there are amplitudes of probability with complex numbers -- conventional computers don't have these.
If you take the task of degrading a number of 2048 bits, the classic algorithm will spread it over a thousand steps and 1,000,000,000 years, and Shora's algorithm, if there was a quantum computer with the right number of cubits, will do it in 107 steps, about 10 seconds. As long as there are no quantum computers, but those that do already know how to do what a classical computer will take a huge amount of time.
- Quantum computers will live up to the hopes they've already given them?
First, let's understand what it takes to create a quantum computer.
Each of these phenomena is worth a lot of engineering difficulties. For example, if you measure a cube, it will change and it cannot be cloned. Or noises, electromagnetic waves, particles are bad for the system, so most platforms cool the whole system down to low temperatures to minimize the effects of noise and dust. But working in cryogenics is much more difficult. All this complicates the creation of quantum computers, so there are now as many as 130 cubes as possible. For example, IBM has released a 128-bit system.
But there's not only physical but also logical cubes. What's the difference? Quantum calculations have to be accurate in the order of 99.999999999999% -- then we think they're very high -- but today they're going between 90 and 99%, they're very low parameters, they're difficult to compute precisely, the percentage of errors will be high. To reach the right level, they're making logical cubes -- they're making one logical cube out of a large number of physical cubes, they're programming error correction protocols, the algorithm, and it's going to be one cube with a high accuracy.
So if you go back to the physical cubes on which the quantum computer is supposed to be made, the industry is at an early stage, at about ten logical cubes. In the coming years, it's expected that a level of 100 logical cubes will be achievable. This will already make it possible to do interesting things -- route optimization, clinical tests, synthetic clinical data creation, encapsulation of quantum simulations, optimization of financial portfolios. To compare: to hack into RSA algorithms, you need about a thousand logical cubes.
We need to make a little retreat and say that there's another problem in quantum computing today -- until quantum memory is invented -- so quantum computing is going to work with classical computers in the next 10 years.
The strategic long-term challenge is to create a universal quantum computer, which requires more than 10,000 logical cubes, reliable management of multi-cube geites, quantum memory.
- What changes quantum computers?
They can solve a huge set of problems -- for example, bioscience. Now we can't even model medium-sized molecular compounds. So scientists do synthetic molecules and experiment all the time. Modelling is very limited to the size of molecular systems and precision parameters. This takes about ten years to create a new drug. And a quantum computer that can model quantum mechanical systems will radically accelerate the process.
Or squirrel folding is now trying to make X-rays, tricky magnetic resonations, and if it's a quantum computer, it can model this system, and we'll make it easier for ourselves to make drugs. And it's gonna accelerate the development of new space mission materials, engines, superconductors, new battery electrolytes that have been standing at 200-250 Wh per kilogram mass density for 20 years. We can't do better because we're not modeling very well.
For one interview, you can't even list all the quantum computer applications that you can think of. Even if it can just accelerate the number of processes that are important, it's going to be a big advance. And that's just one step towards creating a universal quantum computer. That's why it's such a hip.
- But they can only be used within the limits of science?
No, in all types of optimization, for example, where graph theory is used, it's already used to optimize financial portfolios, routes, I.I. Algoritms.
"Qubits are good, but that doesn't indicate speed and accuracy of computation."
- Are there any other problems that we can't figure out how to solve?
The main thing is the creation of cubes in large quantities and their binding, the time of life of the whole system, and let's say that if the time of life of the system is 0.001 seconds, you can't figure out what's important, and you have to think about how to keep the quality of the calculations and scale them up.
IonQ is a company that has been invested in by respected investment funds from all over the world, and it's even become public. They make systems on ions, and the problem is that there are ion traps, but there's a limit to the number of ions that you can catch, and you have to come up with a mechanism to tie traps between you. And so far, the big problem is that it's very difficult to scale the system. Other platforms have similar serious problems.
There are also problems with equipment -- sometimes quantum computers need to invent new devices. For example, special optics, lasers, vacuum equipment, cryogenic cameras. There are many problems, but this path of development -- the microelectronics have passed it. It's normal that every new process, industry adapts and comes up with new conductive metals and other discoveries. It's just that the whole system is still at an early stage of maturity.
- How can specialists who are interested in quantum computers understand whether a new discovery is really a step forward for this industry or just another news for clicks? What should we look at? For example, the number of cubes is an indicator?
If you don't understand at all, these benchmarks are gonna be very superficial, and sometimes even misleading, like the number of cubes, it's actually good, but they don't tell how much the system can compute and how accurate it is.
What matters to me is the number of connected logical cubes, the accuracy of the computation, the lifetime of the system, and the ability to compute practical algorithms.
The development of quantum computers is long, expensive and difficult, so it seems like a very limited number of organizations do it. Doesn't that mean that such devices will only work for corporations and nations?
And you can write your quantum diagrams and count algorithms. Each developer is interested in increasing the number of practical tasks that you can do on their quantum computer, so the cost is cheaper.
In terms of the number of investments in the sector, it can be concluded that there is progress. This is an indirect parameter: if hundreds of investors invest and the industry is growing, it means a lot. And since the 2019s, the number of investments has been increasing, from $300 million to $2.3 billion. We seem to be close to solutions that will become practical.
But there are only 80 organizations that make quantum computers, but the numbers say that 1.5 billion have invested in hardware, and of these, 12 companies have taken the lion's share, and they need specialists in quantum physics, mathematics, engineering, and the interesting fact is that the Soviet school is considered strong here, and we've talked to many of the 260 active companies in this field -- 20 percent of them have Russian engineers, physicists, or maths.
"Rossian scientists are three to five years behind the world"
Probably the main players in quantum computations are the RCC, FIAN, MGU.
- What kind of research do they have to talk about?
On the road map, they make quantum computers on different platforms -- atoms, ions, photons, superconductors -- and I feel like they're three to five years behind the world's companies, but they've got serious footage and an approach -- they're definitely going to develop something useful.
Are the researchers afraid that the technology will get out of control?
We're on our way to regulation, so long as everyone's concerned about building a hardware, and as soon as something serious comes along, it's gonna go down, but everyone's afraid of their data. For example, now you can protect the data with quantum encryption and reduce the likelihood that a quantum computer can hack it, but if someone copied the data and waits for the quantum computer to appear, then they can decipher it. Now that's the main fear.