Introduction Possibilities Implementation Results Conclusion Photometric Surveys: Catalog generation with HEALPix and VO Jiˇ r´ ı N´ advorn´ ık Astronomical Institute Academy of Sciences, Czech Republic June 15, 2015 J.N´ advorn´ ık ASU CAS CZ Virtual Observatory
18
Embed
Photometric Surveys: Catalog generation with HEALPix and VO...IntroductionPossibilitiesImplementationResultsConclusion Photometric Surveys: Catalog generation with HEALPix and VO Ji
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Catalog inaccuracies, duplicitiesCan’t produce light curves for objects not in catalogCannot produce transientsBest result with PPMXL: cca 70 % of data used (with errors)
Integration with GAVO DaCHS package for VO publication
Clustering problem
NP-Hard in general (quantum computer needed)Best K-means fails for thousands of clusters (vs. 6 million)We know cluster sizes ⇒ We can define overlapsHEALPix for parallelization
Cca 2 h per whole dataset for linear algorithm (12 cores, 32GB RAM)K-means by order slower, but scales linearlyVery good scalabilityExceptional quality with linear algorithmAcceptable memory complexity cca 8 GB per 100 mil.observations