Computer-aided design of inorganic substances and materials

For the first time the investigations on application of methods of artificial intelligence to a prediction of new inorganic substances with predefined properties were initiated in IMET RAS in the sixties.

The application of methods of artificial intelligence (methods of computer learning, etc.) for the analysis of the information from DB on properties of inorganic substances and materials has allowed to find periodic regularities of formation of compounds of various types and to realize the prediction of thousand of new compounds of various composition and the estimation of some of their properties. Many of their predicted compounds are promising substances for development of new materials with magnetic, superconducting, semiconductor, electro-optical, acousto-optical, nonlinear-optical, ferro-electric and etc. properties. The experimental checking of the predictions which have been carried-out in IMET RAS and in other R&D institutes, shows, that the average reliability of prediction exceeds 80%.