Scientists from the Ural Federal University taught neural networks to solve problems on quantum computers.
Scientists are trying to use quantum computersto solve quantum problems, since the computing power of conventional computers and even supercomputers is not enough for this. However, there is one problem: a quantum computer interacts with its environment, because of which its state is constantly changing. For this reason, the results of the calculations may not meet expectations. To cope with the problem, scientists are trying to minimize the impact of external noise either by setting up the equipment, or by software.
Researchers of the Ural FederalUniversity went the second way. They developed an algorithm capable of determining the phase of a magnetic material and its properties. The neural network can solve the problem even when exposed to external noise and still comes to the best of the possible values. In fact, the neural network itself adapts to the current state of the quantum computer.
Scientists hope to use a neural network to predict the properties of new materials.