Mirror Laboratory

The Laboratory for Computational Physics, together with the G.A. Krestov Institute of Solution Chemistry of the Russian Academy of Sciences (Ivanovo), is conducting joint research within the Mirror Laboratories project: ‘Predicting the Properties of Molecular Systems: Combining Machine Learning and Classical Modelling Methods.’ The project aims to develop the concept of generative design in chemistry and materials science. The primary goal is to apply and develop modern methods for computer modelling of molecular systems, based on the achievements of modern theoretical physics and applied mathematics, in order to model and predict the properties of molecular systems and materials based on them. As such, the primary goal of the Mirror Laboratories is to conduct cutting-edge research in the fields of computational chemistry and the integrated design of new materials based on big data processing of their practically significant properties, including physicochemical and biochemical ones.

Today, supercomputers and computer modelling methods, including classical approaches from applied mathematics, theoretical physics, and deep machine learning, are powerful tools for studying molecular systems and materials. They are used to characterise the properties of known materials and to predict new materials with desired properties. These methods are based on processing large amounts of physical and chemical data using machine learning techniques, including neural network technologies.

The development and implementation of these methods in various fields has led to the emergence of generative design—a new tool for design and modelling. Generative design allows for the delegation of a huge volume of operations and calculations to a computer. Thus, scientists, engineers, and designers can become generators of ideas or principles, and optimal paths to achieving results are generated using advanced neural networks and powerful supercomputers.