Artificial intelligence approach for calculating electronic and optical properties of nanocomposites

Abstract The optical and electronic properties of quantum dots (QDs) are critical points in the design a QD-based optoelectronic device that can be computed using numerical methods, for which surrogate-based optimization procedures can reduce the computational cost and time. In this paper, an artificial neural network-based Multi-Layer-Perceptron (ANN-MLP) approach is developed to calculate the optical and electronic properties of a hybrid nanocomposite. The predictive capability of the methodology was measured in terms of Mean Square Error (MSE), Relative Square Error (RSE), and Correlation Coefficient (R 2) on the test dataset. The ANN-MLP model has produced the best RSE, MSE, and R2 values of 6.35 × 10−5, 6 × 10-6, and 0.9999 in predicting energy levels and 0.0116, 3.5 × 10-4, and 0.9887 for the absorption spectrums of QDs, respectively. The comparison of the results shows the high accuracy of the ANN-MLP as a practical algorithm for predicting energy levels and optical absorption spectrum of QDs.