面向海量数据的并行天文交叉证认

发布时间：2018-03-25 16:04

本文选题：天文交叉证认　切入点：HTM　出处：《计算机应用》2010年08期

【摘要】：交叉证认是实现多波段数据融合的关键技术,天文数据的海量性使这一问题必须要依靠计算机技术加以解决。按照PCAM并行设计模型设计了并行交叉证认算法。针对交叉证认在数据I/O访问方面存在的性能瓶颈,通过调整划分的粒度、过滤空白区域、优化数据加载、计算流程等方法,协调了数据读取量与计算量间的关系。实验表明该并行方法对交叉证认计算的效率提升明显。另一方面还考虑了对HTM、HEALPix两种最常用天文数据索引方式的支持,并通过实验对比了两者的性能,为我国天文数据主题库、虚拟天文台等项目提供了技术参考。
[Abstract]:Cross identification is the key technology to realize multi-band data fusion. Because of the magnanimity of astronomical data, the problem must be solved by computer technology. According to the parallel design model of PCAM, the parallel cross identification algorithm is designed. By adjusting the granularity of partition, filtering blank area, optimizing data loading, calculating flow, etc. The experimental results show that the parallel method can improve the efficiency of cross-identification computation obviously. On the other hand, the support for two most commonly used astronomical data indexing methods, HTMHEALPix, is also considered. The performances of the two are compared through experiments, which provides a technical reference for the subject database of astronomical data and the virtual observatory in China.
【作者单位】：天津大学计算机科学与技术学院;中国科学院国家天文台;
【基金】：国家自然科学基金资助项目(10978016) 天津自然科学基金资助项目(08JCZDJC19700) 天津市科技支撑重点项目(09ZCKFGX00400)
【分类号】：P112

【相似文献】