A Statistical Tool for Finding Non-Particle Events from the AMANDA Neutrino Telescope
A Statistical Tool for Finding Non-Particle Events from the AMANDA Neutrino Telescope
复制标题
从阿曼达中微子望远镜寻找非粒子事件的统计工具
DOI：
--
复制
发表时间：
2004
期刊：
影响因子：
--
通讯作者：
A. Pohl
中科院分区：
文献类型：
--
作者：
A. Pohl
阿曼达是一个位于南极的切伦科夫探测器，被设计成中微子望远镜。它的探测器部分是一个阵列的677 PMT在冰在1500 - 2000米的深度。它表明，事件有时包含脉冲，不产生的粒子，甚至不是间接的，但从外部电磁辐射和/或检测器故障。由非粒子源触发或影响的事件称为非粒子事件（NPE）。大量的NPE已经通过了第一过滤阶段，有些甚至是最后阶段。为了避免NPE被用作粒子事件，并避免它们在数据分析过程中扭曲基本变量的分布，已经开发了一种统计工具。该工具由九个变量（“指标”）组成，用于计算每个事件中不同类型的奇怪特征。所有奇怪的特征也出现在看似正常的事件中，接近伽马分布。这是通过将指标标准化来告诉某个值有多不寻常来处理的。从正常事件中，归一化指标都得到相同的指数分布。可被怀疑为NPE的事件具有高指示值，通常超过指数分布结束的指示值。在正常数据过滤期间，指标分布并不完全不变，但它们似乎足够稳健，可以设置指标限值，以确定事件是否为可疑NPE。NPE检查器已被应用于一个主要的数据分析，在那里它被用来恢复一个基本的分布。
AMANDA is a Cherenkov detector situated at the geographical South Pole and is designed to be a neutrino telescope. Its detector part is an array of 677 PMT in the ice at a depth of 1500 2000 m. It is shown that events at times contain pulses that do not arise from particles, not even indirectly, but from external electromagnetic emission and/or detector malfunction. Events triggered or affected by the non-particle sources are called Non-Particle Events (NPE). Large amounts of NPE have passed the first filtering stages, and some even the final stages. A statistical tool has been developped in order to avoid that NPE are used as particle events, and to avoid that they distort distributions of essential variables during data analysis. The tool consists of nine variables (‘indicators’) that count different types of odd features in each event. All odd features also appear in seemingly normal events with a close to gamma distribution. This is handled by normalizing the indicators into telling how unusual a certain value is. The normalized indicators all get the same exponential distribution from normal events. The events that can be suspected to be NPE have high indicator values, typically beyond that where the exponential distribution ends. The indicator distributions are not entirely unchanged during normal data filtering, but they are seemingly robust enough to enable setting indicator limits which decide whether an event is a suspected NPE. The NPE checker has been applied to one major data analysis, where it was used to restore an essential distribution.