Russian and Italian physicists have used a special machine learning system to improve the understanding of the properties of metallic hydrogen at high pressures. This was reported by the press service of MIPT.
The approach used is universal and suitable for a wide range of tasks where direct quantum calculations are too cumbersome. However, the potential needs to be "trained" for a specific task by collecting statistics of calculations based on the fundamental principles of quantum mechanics.
For a long time, scientists have been выясняли under what conditions metallic hydrogen can occur in the laboratory or in massive natural objects, such as giant planets. To obtain such data, quantum mechanical calculations were used, requiring significant computing resources and time.
Russian scientists have developed a new approach that allows accelerating these calculations without significantly reducing accuracy, combining machine learning methods and classical molecular dynamics.
Physicists were able to calculate changes in the viscosity of hydrogen and its ability to diffuse at ultrahigh pressures and temperatures. The scientists also confirmed that liquid hydrogen behaves like plasma at high pressures and explained the anomalous increase in the diffusion coefficient with increasing density of the conductive hydrogen liquid.
Earlier www1.ru reported that in Moscow one-atom electrocatalysts have been created for the production of high-purity hydrogen.
Read materials on the topic:
An innovative method for producing hydrogen has been developed in Russia
The first prototype of a hydrogen-powered electric vessel in Russia was tested on the Neva