Machine learning models to predict the stress–strain behavior of silver nanowires

Metallic nanowires are the most extensively used and most promising component for next-generation nanoscale devices due to their increasing efficacy in various fields and most importantly in electronic devices. Computational methods are preferred for studying nanostructures because they offer atomic-level insights in a cost-effective, controllable and scalable manner beyond the limits of experimental techniques. Molecular dynamics simulation is a powerful method to understand the mechanical properties of nanostructures, with the help of virtual experimentation. Machine learning provides a very powerful approach for the analysis of previously recorded data to predict the future. Machine learning is a growing approach to model systems and predictions. This work offers a computational framework for the analysis and augmentation of molecular dynamics simulation data for the Silver Nanowires. The parameters like coefficient of efficiency, root mean squared errors and mean squared errors are evaluated for the machine learning models like Fine Tree, Fine Gaussian SVM, Bagged Trees and Rational Quadratic GPR. The dependence of the predictor variables on the response variable has also been discussed with Shapley importance values.