1 / 14

Report from Metadata Working Group

Report from Metadata Working Group. ILDG6 file format was proposed and approved QCDml1.1 had been stable ILDG7 propose an update of QCDml improved management part of QCDml marked up new lattice actions summary of on-going discussions observables, validity of schemata

atara
Download Presentation

Report from Metadata Working Group

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Report from Metadata Working Group • ILDG6 • file format was proposed and approved • QCDml1.1 had been stable • ILDG7 • propose an update of QCDml improved management part of QCDml marked up new lattice actions • summary of on-going discussions observables, validity of schemata • other issues ILDG7 (Dec.08,2005) T.Yoshie for MDWG CCS,Tsukuba

  2. management archiveHistory revisions elem elem elem revision revisionAction numberConfigs Management of QCDml1.1 • strategy of QCDml1.1 • we record all events of “addition/removal of configurations” to/from the ensemble, using management part of IDs. • <revisions>: total number of revisions • <revision>: sequential number (=<revision> in config XML) • <revisionAction>: add/remove of configurations • <numberConfigs>: #config processed in this revision • <elem> includes other info (person, date,…)

  3. Management: requirements • new requirements: we want to • write an ensemble XML ID when config generation starts • avoid frequent change of ID suppose we add configuration one by one. long list of archiveHistory • store ID written by users to MDC without further changes for QCDml1.1, a tool to generate archiveHistory consistently has to be developed, and the user ID has to be processed by the tool. • Good properties we don’t want to change: • #configs belonging to an ensemble can be counted. • replace/remove of configs can be detected by users • modification of ID can be detected by users • a kind of time stamps is recorded • change of existing IDs has to be minimal

  4. management archiveHistory revisions elem elem elem Management: solution (1) • ensemble XML • <revisions> and all <elem>’s in <archiveHistory>are optional numberConfigs revision revisionAction • when one records optional archiveHistory elem’s: <revisionAction> refers to a revision of XML ID add: addition of ID to MDC, replace: replace of ID in MDC remove: remove of ID from MDC <numberConfs> is removed, i.e. add/remove of configs to/from an ensemble is not recorded in the ensemble XML ID.

  5. management archiveHistory revisions Management: solution (2) • configuration XML • one <elem> in archiveHistory (generate) is mandatory • others optional elem elem elem revision=0 revisionAction = generate includes info. of person, date • when optional archiveHistory elem’s are recorded <revisionAction> add/replace/remove refer to either ID or config.

  6. Management: solution (3) • new management parts satisfy “new requirements”. It also keeps old important properties, even if optional elements are not marked up. • write an ensemble XML ID when config generation starts • avoid frequent change of ID Yes, because coupling between ensemble IDs and config IDs was removed • ID written by a user is not changed when it is stored in MCD Yes, if one does not record archiveHistory. One may authomatically update the archiveHistory, when IDs are stoared in MDC. • #configs belonging to an ensemble can be counted. count #config IDs with markovChainURI as a key • replace/remove of configs can to be detected by users when replaced, crcCheckSum and hence config ID will be modified • modification of ID can be detected by users users have to keep copies of ensemble/config IDs • a kind of time stamps is recorded revisionAction=generate with a time stamp is mandatoy in config ID

  7. classify ensembles lover level classification (optional) Management: more improvement • <ensembleLabel> is proposed by MILC collab. Ensembles generated by a project are usually classified by e.g. lattice spacing (coarse, fine, superfine), lattice size (small, medium, large). <ensembleLabel> enables to mark up such lower-level grouping

  8. gluon plaquette sixLink plaq. rect. chair 3d New action: gluon • Convention to mark up sixLink actions iwasakiRG • c3 was missing in QCDml1.1 • we mandate c3, though redundant DBW2 LuescherWeisz tpLuescherWeisz treeLevelSymanzik

  9. quark KS overlap wilson wilsonTm clover domainWall npClover improved KS askTad tpClover New action: quark Requests of marking up other actions, write to us

  10. QCDml1.2.0 release major/minor version QCDml1.2.0 • modification from the old version is minor • the new one is not compatible with the previous QCDml1.1 • compatible: if IDs written based on the old schema are still valid under the new one. • strategy to name and treat new QCDml schema • if not-compatible, count-up major/minor version number • if compatible, count-up release number. All releases (with the same version number) belongs to the same name space, e.g. QCDml1.2.x has targetNamespace="http://www.lqcd.org/ildg/QCDml/ensemble1.2"

  11. On-going: observables • a request to markup observables e.g. lattice spacing • useful for users to search ensembles • conceptually (qualitatively) different from what we have marked up. • Ensemble ID: how the ensemble was generated • Observables: properties (consequences) of the ensemble • Thus, deliberate action is necessary • summary of discussions up to now • Some kind of observables are not property of ensemble. E.g. Lattice spacing, if it is determined by chiral extrapolations. • Dimension-full quantities are dangerous. They depend on inputs. • Values of observables change, when more configurations are added. • what kind of observables are worth marking up. pion prop maynot be. • candidate: pi-rho ratio and sommer scale in lattice units • plan to come to conclusions within a month or so.

  12. Validity of Schemata • current version of QCDml is not valid under W3C definition • strategy to express hierarchical structure of lattice actions • action: complexType, derived action: substitution of complexType • content model of the new type has to be restriction/extension of that of the old one. Current QCDml violates this rule • validation of schema does not work properly with XMLspy5 • QCDml does not pass XMLspy2006 and W3C… validators. • causes no serious problem for usual use of QCDml • validation of ID works correctly • Xpath(1.0) query does not refer to schemata • working to write a new valid schema • not completed yet, but expect required changes to be backward compatible • inheritance of actionTypes is valid (action hierarchical tree unchanged) • elements in IDs are unchanged (users don’t have to do anything)

  13. Other Issues and Future Works • procedure to release new schemata (almost agreed by WG) • user wants to markup his/her ensemble generated with a new action • write a request to the MDWG • a MDWG member (or the user) writes a schema for the new action and circulates it to the MDWG mail-list • the proposal of change is automatically accepted if there is no objection circulated within a certain period of time (e.g. 2 weeks) • Chris Maynard volunteers final editing of schemata • we put the new release somewhere at the ILDG web-page • and announce new schemata via ILDGNews blog of the ILDG web-page or other mail-lists • write glossary for marked-up actions • write human readable document or guidebook , like “How to markup QCD ensembles and configurations”

  14. Summary • New version of QCDml (QCDml1.2) was released • Dirk Pleiter volunteered to write this version. • Discussion is on-going on whether/how we extend QCDml to include observables. • feedback of your opinions/comments are welcome • Work is in progress to write schemata compatible with W3C definition • Chris Maynard is tackling this issue • Mail list qcdml@ccs.tsukuba.ac.jp is closed to members only. • write to yoshie@ccs.tsukuba.ac.jp if you want your mail address be included, or you send your opinions/comments to members. • mails are archived at http://www.ccs.tsukuba.ac.jp/ILDG/QCDML-MA/

More Related