利用机器学习研究原子核四极形变在中能重离子碰撞中的效应

Application of machine learning to study the effects of quadrupole deformation on the nucleus in heavy-ion collisions at intermediate energies

本文基于极端相对论量子分子动力学(UrQMD)模型,在模型初始化部分考虑弹靶核的四极形变,研究形变对不同入射能量(每核子0.4,1.0和1.5 GeV)下的^(238)U+^(238)U碰撞的影响.利用机器学习中的LightGBM决策树算法训练初始化过程中有无形变效应所给出的反应末态的自由质子、带电碎片(质量数A>1)以及带电π介子的横动量(p_(t))与快度(y_(0))的分布,使决策树算法具备从逐事件给出的末态粒子分布中识别原子核形变的能力.利用LightGBM算法中的重要特征分析发现弹靶快度区的带电碎片数敏感于弹靶核的四极形变,可以作为用重离子碰撞探测原子核四极形变的探针.

The influence of quadrupole deformation on the projectile and target nucleus in^(238)U+^(238)U collisions at beam energies of 0.4,1.0,and 1.5 Ge V/nucleon is investigated using the ultra relativistic quantum molecular dynamics(Ur QMD)model.The effect of quadrupole deformation is investigated using the Light GBM decision tree algorithm.Light GBM’s ability to identify nuclear deformation from particle spectra is demonstrated on an event-by-event basis by learning the twodimensional(transverse momentum and rapidity)distributions of free protons,charged fragments(mass number A>1),and chargedπmesons produced in^(238)U+^(238)U collisions with and without deformed initializations.Using the important characteristic analysis of Light GBM,the yield of charged fragments around target/projectile rapidities is found to be sensitive to the quadrupole deformation of both the projectile and the target.It can be used as a promising observable to probe quadrupole deformation with heavy-ion collision.