Destek vektör regresyonu yöntemiyle karbon nanotüp benzetim süresinin kisaltilmas

Abstract

Density Functional Theory (DFT) is one of the most important application of Carbon Nanotubes (CNTs). Because of the chemical and physical characteristics of carbon, CNTs play an important role in the field of nanotechnology. The most difficult part of CNT simulations is DFT calculations that take hours or even days. In this study, a Support Vector Regression (SVR) model that reduces the atomic coordinate calculation of CNT simulation duration has been proposed. u, v, and w coordinates which obtained from CNT simulations are predicted with high accuracy using the SVR method within minutes. A dataset containing 10721 samples was created using CASTEP software for the prediction model. The dataset consists of the atomic coordinates and chiral vectors. To evaluate the accuracy of the proposed model, Mean Square Error (MSE), Mean Absolute Error (MAE), Standard Error of the Estimate (SEE) and Correlation Coefficient (R) metrics were used. The dataset is studied separately with and without using 10-fold cross-validation. The results obtained from this study can be used in two ways: 1) The atomic coordinates can be predicted with a low-error without using a simulation program, 2) The estimated results can be used as an initial value of simulation software for reducing duration of the atomic coordinate simulation seriously.