神经元网络法重建具有大丢失能量的共振态粒子质量

Mass Reconstruction of Resonance Particles With Large Energy Loss by Neural Network

一个前馈式神经元网络系统被设计用来重建具有大的丢失能量的共振态粒子质量．以在ＬＨＣ（ｌａｒｇｅｈａｄｒｏｎｃｏｌｌｉｄｅｒ）上ｐｐ对撞产生的Ｈｉｇｇｓ粒子的衰变模式Ｈ０→τ＋τ－→ｅμｘ为例，神经元网络法正确地重建了Ｈｉｇｇｓ粒子质量ＭＨ的峰值及比常规方法好的宽度．该法同时具有抗本底事例干扰能力，对共振态粒子质量的精确测量及新粒子的寻找具有实用价值．

A feed--forward neural network is designed to reconstruct the mass of resonance particles with large energy loss.For the Higgs particles produced at LHC pp collider decaying through Ho→τ+τ-→eμx,this neural network correctly reconstructs its mass with the right peak position and better width than does the conventional method. The network alsoposesses the capability of suppressing background events.This kind of neural network can be widely used in the search for new particles and in precise mass measurement of resonance potticles.