Atlas Metadata Interface James Walder Lancaster University http://ami.in2p3.fr/ AMI • • • • http://ami.in2p3.fr/ The Atlas Metadata Interface Used for ‘dataset discovery’: • • Necessary for Monte Carlo searches, It is faster than RQT, if you know what you are after; • Doesn’t perform explicit data quality or conditions queries Dataset are also produced with specific Athena releases, geometry tags, conditions tags. These are coded into special config tags that appear in the dataset name. data09_900GeV.00142406.physics_L1CaloEM.merge.DESD_COLLCAND.r988_p62/ • Eg: r988_p62 means reconstructed under tag r988 from which a DPD was made using tag p62 • AMI also contains the tag nomenclature ( r, p, e,s,f,m... ) • And the exact meaning of each individual tag definition: e.g. r988. 2 James Walder Data processing chain Reconstruction Detector & trigger Digits ESD Digits Monte Carlo James Catmore Physics analysis AOD & TAG building ESD AOD TAG dAOD/dESD building ATLAS offline software tutorial, CERN, 21st - 23rd October 2009 3 Physics analysis James Walder Data Formats • • • • • RAW -> Originates from the detector. Large, specialised usage. ESD -> Still large -> used for performance and detector studies. DESD; As ESD, but skimmed in some way; performance studies • DESDM: As DESD, but includes trimming / thinning / slimming AOD; For use with analysis. ESD format important for first data. DPD; Derived Physics Data. • • Number of varieties exist. some are in pool.root Athena format Names and definitions recently changed. • • • DAOD: As AOD but skimmed according to some algorithm DAODM: As DAOD, with additional trimming / thinning / slimming DAODM2: new name for D2DPD. pool.root format data made by groups • Others; the D3PDs are ROOT-ntuples produced by individual groups for common analysis within the group. • All will exist on the Grid; • • RAW and ESD will in general not be easily accessible. Useful formats for first data: • DESD_COLLCAND. As ESD, but only events that satisfy collision timing cuts. • To be phased out. 4 James Walder Anatomy of a dataset name • Dataset names hold much of the immediately useful metadata within it’s own name. Project name: data / mc 08/09... energy: 900,GeV 2TeV Type: physics, calibration File Type: RAW, ESD, DESD, AOD,DAOD Stream data09_900GeV.00142406.physics_L1CaloEM.merge.DESD_COLLCAND.r988_p62/ Run number or MC dataset number • File status: merged / recon Production/ reconstruction tags: Quickest way to find your data is through AMI, using specific parts of the dataset name, and wildcards. 5 (Container) James Walder AMI – Search examples • https://ami.in2p3.fr AMI Dataset Search • • • Dataset search through wildcarded name search, or through use of keywords. Also will translate the meaning of the reconstruction/production tags. Use % as wildcard. • • eg: mc08%AOD% data09%Min%AOD%r988% 6 James Walder AMI: tag interpretation • Need to know the meaning behind the production tags? • Key information: Geometry, Athena Releases, Conditions tags 7 James Walder Advanced Search • Advanced search is a more ‘intelligent’ method of searching for specific sets of data. AMI 8 James Walder Tasks in this session • • • • • • • • • Find the datasets you are interested in, and investigate their metadata For this tutorial choose AOD format from the Minbias stream • Try and find its provenance, locations where it is stored in the world, and the number of events it contains. • What (minimum) Athena release is needed to read it? Look for similar datasets: • • of the same stream (e.g. the same dataset name, but with a different set of tags). of the same set of config tags (eg the r988_p62). What are the reasons for the differences? Find out what information the config tags (eg. r988) hold and what they mean? How many types of config tags are there - do you know what they all stand for? Some Monte Carlo data has been recently produced, it can be found with mc09_900%minbias%AOD% • Try to understand the tags that are provided here and the provenance related to MC. Further details on how to use AMI are at the tutorial page: http://ami.in2p3.fr/opencms/opencms/AMI/www/Tutorial/FastTrackTutorial.html 9 James Walder