Artificial Neural Network based Modelling and Simulation to Identify new candidates for hosting Skyrmions

Abstract The premise of this work is to identify new candidates for hosting exotic topological excitations known as skyrmions. The tool used in this research is artificial neural network. The strategy is the statistical analysis of chiral structures contained in the ICSD database, with the additional information about the group number of constituent elements in the periodic table. The model described in this study predicts ideal chiral crystal and proposes a new chiral crystal design direction. Skyrmions are topologically protected materials with an unusual spin composition that are structurally asymmetric. This research introduces a deep learning method for L1 and L2 regularization with solution to a loss minimization problem for identifying the chiral crystals in skyrmion material of thin films. This paper introduces an approach to building a probabilistic classifier and an Artificial Neural Network(ANN) from a true or false chirality dataset consisting of chiral and achiral compounds with elements of the type ‘A’ and ‘B’. A quantitative indicator is demonstrated for the accuracy of chiral crystal formation. Through contrasting with probabilistic classifier method, the viability of the ANN approach is checked comprehensively. We present deep learning algorithm with Artificial Neural Network(ANN)modeling and simulations. Through out the manuscript we employ statistical techniques as well as deep learning strategies to identify chirality property for chiral crystal design. This research work opens the way to build sophisticated crystal design software methodology.