Electron density mapping of boron clusters via convolutional neural networks to augment structure prediction algorithms

Determination and prediction of atomic cluster structures is an important endeavor in the field of nanoclusters and thereby in materials research. To a large extent the fundamental properties of a nanocluster including its chemical, optical, magnetic, mechanical and transport properties are mainly governed by the molecular structure it adopts. Traditionally, structure elucidation is achieved using quantum chemical (QC) calculations that are usually tedious and time consuming for large size clusters. Various structural search algorithms, implemented in software codes, have been reported in the literature. Although they tend to accelerate the structure exploration, they still require the aid of QC calculations of energies for structure evaluation. This makes the structure prediction process using these approaches a computationally expensive affair. In this paper, we report on the creation of a convolutional neural network model based on a machine learning, which can give relatively accurate energies for the ground state of nanoclusters from the total electron density on the fly and could thereby be utilized for aiding structure prediction algorithms. We have built up a dataset consisting of pure boron clusters for the purpose of training our model.