机器学习在原子核物理中的应用

Machine learning applications in nuclear physics

机器学习是人工智能研究领域最成功最活跃的研究方向,正在被广泛应用于大数据分析、自动驾驶、自然语言处理、互联网技术、医学诊断、市场分析等领域.同时也被广泛应用于粒子物理的实验和理论研究.本文针对核物理领域列举了一些值得被关注的机器学习技术.通过几个不同案例,介绍了机器学习在核物理研究的不同场景的应用.其中包括深度卷积神经网络结合理论模型用于提取原始数据中的团簇结构信息,无监督学习方法应用于实验数据分析提取原子核液气相变信息,用神经网络量子态(NQS)作为试探波函数求解薛定谔方程,贝叶斯神经网络应用于原子核半径的拟合与预测,图像识别技术应用于活性靶时间投影室探测器事件鉴别.展示了机器学习技术在核物理领域的广泛的应用价值.

Machine learning is the most successful and active research direction in the field of artificial intelligence.It is being applied in various fields,such as big data analysis,autonomous driving,natural language processing,Internet technology,medical diagnosis,and market analysis.Accordingly,it is also widely used in the experimental and theoretical research of particle physics.This article discusses several machine learning techniques that are particularly relevant to nuclear physics.Additionally,the application of machine learning in various contexts of nuclear physics research is demonstrated through several different cases,including the deep convolutional neural networks combined with theoretical models to extract information about the nuclear cluster structure from low level data,unsupervised learning methods applied to experimental data to extract information about nuclear liquid-gas phase transitions,and neural network quantum state(NQS)as a trial wave function to solve the Schrodinger equation.Bayesian neural networks are applied to fit and predict the nucleus radius,and image recognition is applied to identify events in the active target time projection chamber detector.It shows the wide application of machine learning technology in the field of nuclear physics.