MITiS earned the first prototype of a quantum computer in Russia. The device in two qubits performed the specified algorithm, exceeding the previously known accuracy limit by 3%. Superconducting materials were taken as the basis for qubits.

Work on the creation of a quantum computer within the framework of the Advanced Research Foundation project has been carried out at the MISiS NITU since 2016 under the leadership of Valery Ryazanov, the chief scientific associate of the University’s Superconducting Metamaterials Laboratory. The design involves the use of superconducting materials as the basis for qubits.

Qubits (quantum bits) – the real power of a quantum computer, an analogue of the “bits” of a conventional PC, only much more advanced. If the familiar computer “thinks” and counts as zeros and ones, that is, each bit of information can be encoded either as “0” or as “1”, then the qubit has the property of the so-called superposition, the ability to be in both states simultaneously. This opens up enormous prospects, because with such computing resources a quantum computer will be able to overtake the most powerful computing devices by orders of magnitude.

A quantum computer based on superconducting materials is a more advanced system than its counterparts. For example, other scientific teams are developing qubits on individual atoms (which can be “lost” due to a negligible size) and on ions (they can be built exclusively linearly, which is physically inconvenient). The qubits created at NITU MISiS are made of aluminum, have a size of 300 microns, they cannot be “lost”, and you can also build non-linearly.

During the experiment, a two-qubit quantum computer solved the Grover algorithm – an enumeration algorithm for a function. A quantum computer, thanks to the principle of superposition, in the ideal case can find the correct value of x in solving this problem in one call to the function f (x) with a probability of 100%.

“Grover’s two-qubit algorithm is a very important step towards creating a quantum computer. We are not the first in the world to demonstrate its work, but here we are talking primarily about technological achievement. We have shown the possibility of implementing all necessary for logical operations for a universal quantum processor: initialization, single-qubit and two-qubit operations, and reading, and with an error level satisfactory for small algorithms, ”said Ilya Besedin, an engineer at the Superconducting Metamaterials laboratory, one of the project participants.

The biggest difficulty in creating a useful quantum processor is errors. Unlike classical computers, which can work for years and always produce reproducible and predictable results, quantum computers are affected by noise, which distorts the calculation results. Despite the fact that the two-qubit processor created at NUST “MISiS” is too small to solve applied problems, it successfully “crossed” the threshold of 50% probability of a correct answer, reaching 53%.

The whole algorithm consists of initializing two qubits, four one-qubit operations, two two-qubit operations, and reading two qubits; errors in any of which reduces the probability of a correct answer in the answer.

A chip for a quantum computer was manufactured at MSTU. Bauman, and his design and launch of the device was already done at NUST MISiS, where a unique set of equipment with cryostats was built in the Superconducting Metamaterials laboratory, which allowed operation at extremely low temperatures to -273.14 degrees Celsius, which is close to absolute zero.

“Nevertheless, we still have a long way to go,” adds Ilya Besedin. – More recently, the Google article, which has not yet been officially published, got into the press and managed to implement the “quantum superiority” algorithm on a 53-qubit superconducting quantum processor. The task of “quantum superiority” is the most favorable task for a quantum computer, which is very difficult to accomplish on a classical computer. And if we have to overcome the “classic” limit – this is still a fundamental result, then the Google result is already closer to the practical side: they were able to formulate and solve a task that their processor can complete in minutes, and a powerful supercomputer has been tested for weeks. “

And even so, Google has not yet managed to get close to ensuring that a quantum computer solves some practically useful task more effectively than a classical one. However, while theoretical predictions regarding the computational superiority of quantum computers are confirmed by experiments.

The next important steps towards creating a useful quantum computer are a demonstration of reduced to several tens of qubits versions of “useful” quantum algorithms (for example, a simulator of a chemical reaction or the ground state of a molecule) and a demonstration of quantum error correction. Exactly for error correction, by the way, superconducting qubits are best suited: they can be organized into a two-dimensional lattice with local interactions and parallel gates, which is necessary for the “surface code” – the simplest from the point of view of requirements and accuracy of operations.

“We also want to move in this direction, but from my point of view in quantum computing it is important not only“ more ”, but also“ better ”: the superconducting qubits that we use now turn out to be quite expensive and give a lot of mistakes. And before doing hundreds and thousands of qubits, in my opinion, it’s worth working on the most basic unit – qubit, ”summarizes Ilya Besedin.