Systematic study of complete fusion suppression for 6,7 Li nuclei using artificial neural network

TL;DR

This paper employs artificial neural networks to accurately estimate the suppression factors in complete fusion reactions involving weakly bound 6,7 Li nuclei, demonstrating superior performance over other regression methods.

Contribution

The study introduces ANN-based methods to estimate fusion suppression factors, showing improved accuracy and providing a new computational approach for analyzing weakly bound nuclear reactions.

Findings

ANN achieved low NMSE of around 1.85-6.48% for training and testing.

Suppression factors estimated as 0.68 for 6 Li and 0.74 for 7 Li reactions.

ANN outperformed Support Vector Regression, Random Forest, and Gaussian Process Regression.

Abstract

In recent decades, there has been a significant increase in the measurement of complete fusion cross-sections for various reactions, with particular emphasis on the reactions involving weakly bound projectile. It has been well established that the complete fusion cross-section involving weakly bound nuclei is suppressed at above barrier energies due to the breakup effect. Accurate determination of suppression factor is essential to understand the effect of breakup on complete fusion suppression. In this study, Feedforward Artificial Neural Network (ANN) methods based on Multilayer Perceptron is utilized to estimate the complete fusion suppression factor for reactions involving 6,7 Li projectile from the comparison of ANN predicted reduced fusion functions (F (x)) with the Universal Fusion Function(F0 (x)). Average suppression factor has been estimated as 0.68 and 0.74 for 6 Li and 7 Li induced reactions respectively. Normalized Mean Squared Error(NMSE) has been calculated as 1.85% for training and 1.92% for testing data of ANN for 6 Li case, while for 7 Li cases, the same is for 3.73% and 6.48% respectively. Present results have been compared with three other alternative methods Support Vector Regression, Random Forest Regression and Gaussian Process Regression. Relevant performance matrices has been estimated. It has been observed that, ANN algorithm is giving the better accuracy as compared with other. This results indicate that ANN might be an alternative tool for estimation of complete fusion suppression factor for weakly bound projectile.