Parkes observations for project P1072 semester 2020OCTS_02

The Parkes telescope has been used to discover more than half of the known pulsars. The techniques used continue to be optimised, but the basic search strategy has not changed in decades. In particular the algorithms produce more candidates than can be searched by eye and so machine learning algorithms are used to sift through the candidates. However, the properties of the candidates were designed for human viewing. Here we propose to record data streams from the highly-versatile Parkes UWL system in order to develop pulsar search algorithms that have been developed, from the ground-up, explicitly for machine learning algorithms.

帕克斯望远镜（Parkes telescope）已助力发现超过半数已知的脉冲星（pulsar）。尽管相关探测技术仍在持续优化，但核心搜索策略数十年来并未发生实质改变。尤为关键的是，算法生成的脉冲星候选体数量远超人工目视筛选的能力上限，因此研究人员借助机器学习（machine learning）算法对候选体进行甄别。然而，当前候选体的特征参数是为人工观测场景设计的。本文提出将记录高通用性帕克斯UWL系统（Parkes UWL system）的数据流，以开发专为机器学习算法从头构建的脉冲星搜索算法。