论文部分内容阅读
人工神经网络方法已被引入高能物理实验领域并被广泛地应用于夸克胶子喷注的鉴别、电子强子分辨、顶夸克和Higgs粒子的寻找等等。本文采用了一种改良的共轭梯度优化算法并应用于高能物理实验中粒子的鉴别。在该应用中,此算法既能实现每步迭代时在搜索方向上获得最优步长,又能避免目标函数陷入局部收敛点,从而使目标函数快速收敛,提高了算法的有效性。分析结果表明,我们改进后的BP算法显著地提高了粒子物理数据分析中的粒子鉴别能力。
Artificial neural network method has been introduced into the field of high-energy physics experiments and widely used in the identification of quark gluon jets, the resolution of electronic hadrons, the search of top quarks and Higgs particles, and so on. In this paper, an improved conjugate gradient optimization algorithm is applied to the identification of particles in high-energy physics experiments. In this application, this algorithm not only achieves the optimal step size in the search direction during each iteration, but also avoids the objective function falling into a local convergence point, so that the objective function converges rapidly and the efficiency of the algorithm is improved. The analysis results show that our improved BP algorithm significantly improves the ability of particle identification in particle physics data analysis.