Physicists from Google were able to demonstrate that existing quantum computers will be able to solve a problem that cannot be solved by any “ordinary” supercomputer.
Ok, Googlebut… in 2019, scientists from the company’s quantum artificial intelligence laboratory already made a similar optimistic statement. True, then their dreams were shattered by the harsh non-quantum reality. If an ordinary classical computer, even with the prefix “super,” operates in its processors and registers with only two abstract states – 0 and 1, then a quantum computer, built on quantum mechanical laws that are not entirely obvious to the “ordinary” world, can, as it were, “hold” in your mind – the qubit – both zero and one at the same time. And due to this, some calculations can be performed at a speed unattainable by conventional computers. Very figuratively, this difference can be imagined as if you tried to measure the area of your room by painting millimeter by millimeter squares on the floor with a pencil and carefully counting them in a notebook, and your smart “quantum” neighbor would use a tape measure for the same task and simply multiplied the length of the room by its width.
No supercomputer can compete with a quantum one in terms of computing power, but only under one condition – if the quantum computer is ideal. In the real world, quantum computers are not perfect. Their main elements – qubits – are not isolated in a perfect vacuum, but interact with the outside world, which introduces interference and noise, destroying the fragile quantum “homeostasis”. And the more powerful a quantum computer we want to assemble, the more qubits we want to connect with each other so that they calculate something useful to us, the more difficult it is to achieve. Related to this is the fact that despite periodic statements from various laboratories about the construction of the next quantum computer with another record number of qubits, it is still not possible to achieve that same quantum superiority from this computer. This happened in 2019, when Google’s Sycamore quantum processor was practically “beaten” by the Summit supercomputer of another corporation, IBM. However, IBM’s quantum developments subsequently suffered no less.
And so, as they write in a recent article in Nature researchers from the same Googlethey succeeded by experimenting with a more advanced version Sycamoreto determine the limits of the “noisiness” of a quantum processor at which it will definitely be able to receive the world’s first medal “for quantum superiority.”
“The introduction of noise into a processor makes it less quantum and more like a classical one, and the researchers were able to establish a noise boundary for sequences of random samples (Random circuit sampling) of different depths, and show that the boundaries of quantum supremacy are akin to the boundary between ice and water – for given initial conditions, you can always say where the transition will be, and it will be of high quality. Below the boundary, a quantum computer is always faster, above it is slower.”,” comments Alexey Akimov, scientific director of the Russian Quantum Center.
Interestingly, near the threshold value, even a slight decrease in the noise level can radically change the “quantum power” of the computer. From here we can draw a simple conclusion that it is not so important, relatively speaking, to chase the number of qubits in a quantum computer, but to improve the quality of operations and their error-free operation on already existing and working installations. So, look, quantum supremacy will come faster.
Based on materials from the press service of the Russian Quantum Center.
Source: www.nkj.ru
