Neural networks as tools for predicting materials properties

Computational synthesis is a method for predicting the properties or for designing materials possessing desired properties. This method entails a computational paradigm for making rapid and accurate estimations of physical properties for molecular systems. The method uses a set of descriptors as a numerical representation of the chemical structure for a given compound and relates these to a set of properties with the use of computational neural network. Compared to other methods for obtaining quantitative structure-property relationships (QSPR), the neural network method offers advantages both in ease of use and accuracy. Finally, the method can be operated in reverse direction with the aid of genetic algorithms.