A Study of High Performance Data Analysis for Large Format Astronomical Image Data

我们研究了大格式天文数据的高性能数据分析。我们试图寻找适合从原始图像数据中快速提取天文信息的方法和环境。我们进行了第一阶段分析的实验，即缩减过程，这个过程花费的时间很长，I/O量很大。我们使用IRAF作为简化软件，这是光学/红外天文学中最流行的数据简化/分析软件包。我们发现磁盘的配置和设置、I/O总线的效率和数量以及主存的数量是影响处理速度的关键因素。这些结果将作为系统配置的指导方针，以实现高性能数据缩减。对哈勃太空望远镜（HST）、斯隆数字巡天（SDSS）、冈山天体物理天文台188cm望远镜和斯巴鲁望远镜数据分析中存在的几个问题进行了分析。我们研究了与数据归档系统、图像浏览器和天文数据库的结合，以促进数据分析。我们与许多天文学家和其他领域的研究人员讨论了数据分析的未来。

We studied high performance data analysis for large format astronomical data. We tried to find methods and enviroments suitable to quick extraction of astronomical information from raw image data.We performed experiments of the first stage analysis, i.e. reduction process, which spends a lot of time with large amount of I/O.We use IRAF as the reduction software which is the most popular data reduction/analysis package in the Optical/IR astronomy. We found that the configuration and settings of magnetic disks, the efficiency and number of I/O buses, and quantity of main memory are critical factors for the process speed. These results will be a guideline of the system configuration for high performance data reduction.We also examined several problems in the analysis of the data of Hubble Space telescope (HST), Sloan Digital Sky Survey (SDSS), Okayama Astrophyisical Observatory 188cm telescope, and Subaru Telescope. We investigate the combination with data arhcive systems, image browsers, and astronomical database, which promote data analysis. We discussed with many astronomers and reserchers of other fields on the future of data analysis.