Evaluation of Potential LSST Spatial Indexing Strategies

Nikolaev, S ; Abdulla, G ; Matzke, R

Lawrence Livermore National Laboratory

关键词: 99 General And Miscellaneous//Mathematics, Computing, And Information Science;    99 General And Miscellaneous//Mathematics, Computing, And Information Science;    Implementation;    Sky;    Optimization;   

DOI  :  10.2172/895409
RP-ID  :  UCRL-TR-225827
RP-ID  :  W-7405-ENG-48
RP-ID  :  895409

美国|英语

来源: UNT Digital Library

【 摘 要 】

The LSST requirement for producing alerts in near real-time, and the fact that generating an alert depends on knowing the history of light variations for a given sky position, both imply that the clustering information for all detections is available at any time during the survey. Therefore, any data structure describing clustering of detections in LSST needs to be continuously updated, even as new detections are arriving from the pipeline. We call this use case ''incremental clustering'', to reflect this continuous updating of clustering information. This document describes the evaluation results for several potential LSST incremental clustering strategies, using: (1) Neighbors table and zone optimization to store spatial clusters (a.k.a. Jim Grey's, or SDSS algorithm); (2) MySQL built-in R-tree implementation; (3) an external spatial index library which supports a query interface.