THE APPLICATION OF THE GENETIC ALGORITHM-BACK PROPAGATION NEURAL NETWORK ALGORITHM IN THE HIGH-ENERGY PHYSICS

Abstract

Multiparticle production mechanism is one of the most phenomena that the high-energy physics concerns. In this work, the evolutionary genetic algorithm (GA) is used to optimize the parameters of the back-propagation neural networks (BPNN). The hybrid evolutionary-neuro model (GA-BPNN) was trained to simulate the rapidity distribution 1/N(dN/dY) of positive and negative pions p-Au, p-Ag and p-Xe for p-Ar, p-Xe interactions at lab momentum Plab =100 GeV/c. Also, for total charged, positive and negative pions for interactions at Plab = 200 GeV/c. Finally, total charged particles for p- Pb collision at center-of-mass energy sqrt(s) = 5.02 TeV are simulated. An efficient ANN network with different connection parameters (weights and biases) have been designed by the GA to calculate and predict the rapidity distribution as a function of the lab momentum Plab, mass number (A) and the number of particles per unit solid angle (Y). Our simulated results have been compared with the experimental data and the matching has been clearly found. It is indicated that the developed GA-BPNN model for rapidity distribution was more successful.

Keywords

• high and ultrahigh energy physics
• hadron-nucleus (h-A) interactions
• rapidity distribution
• modeling and simulation
• hybrid evolutionary-neuro model