Notes on the discussions that followed the various presentations during the Data Model session, held Thu Oct 6, at the ESAC Interop 2005. Notes by A.Micol and M.Louys Overall recap of ACTIONS and ISSUES (before actual notes): ---------------------------------------------------------- # STC: Issue: STC & STCLite, two incompatible though similar schemas -> confusion. Issue: still required a simple layer include files for coos system for VO table requirements. # Line Data Model: Action: Vizier (FO) to provide feedback to Line Data Model (MG) # Characterisation Data Model: Action: Discussion needed (Anita, Arnold, Igor,...) regarding the opportunity of describing the spectral resolution in km/s instead of wavelength (or frequency); different views regarding spectral/velocity axes. # Catalog Data Model: Action: Identify use cases for Catalog Data Model, and define which properties should be part of it. -------------------------------------------------------- In order of appearance, comments were given by the following friends: JCM - Jonathan McDowell GL - Gerard Lemson FO - Francois Ochsenbein FB - Francois Bonnarel ML - Mireille Louys AM - Alberto Micol AR - Anita Richards Arnold - Arnold Rots MNS - Maria Nieto Santisteban PO - Pedro Osuna BT - Brian Thomas MG - Matteo Guainazzi IC - Igor Chillingarian RW - Roy Williams GD - Gilles Duvert BEGINNING OF SESSION JCM WG presentation -------------------- JCM: It would be nice to have basic reusable DM components (Char, STC, Curation, SEDCoord, etc.) IVOA, EGSO, IHY ...: ideally, common components would help. ======================================================================== P.Osuna: Catalogue Data Model Status Report and Proposal -------------------------------------------------------- JCM: separate catalog model and source model, use case: queries that understand the astrophysics In current model the concept of a "column" (or "property") is missing. GL: simplest model is an "extracted source", more info needed (eg provenance) for eg measuring the redshift. FO: What's the Difference between Source and Astronomical Object? AR: We cannot describe the whole of Astronomy in one DM. AM: Many properties missing, many parameters available in Vizier; question: (how to decided) where to stop? MG: assigning a property to a source or an object (process of cross-identifying sources)? although list of properties of a source could be too long, MG prefers to keep it as it is. MNS (Maria): queries common to all catalogues; e.g. a skynode with a common set of attributes to set the same query to all services, than it is practical. AM: but the scientific answer could be wrong if eg a galaxy diameter is defined differently in different catalogues. JCM WRAP UP: identify use cases, controversies: is this the right approach? which properties should go in? ======================================================================== Mireille Louys: Characterisation of Observations ------------------------------------------------ DT: Where is the accuracy/error? Answer (FB): Resolution Sampling Coverage, to which we added AxisFrame->accuracy AW: How to go from Char to Source DM? Answer: via the Observation DM -> Provenance -> Characterisation DM AW: No, Source parameters (e.g position) depends on the e.g. accuracy given in the Char DM. Audience: How to describe the Fourier plane? AR's answer: keen to move to version 2 because Sensitivity will help there. JCM's answer: other axes can be added. Morphology in the Fourier domain? Is it a per observation model and not per source or catalogue model. What if sensitivity changes in the FOV? Answer: Version 2 with exposure maps, etc. Peter Skoda: subjective description of the precision/accuracy (e.g. by observer)? JCM's answer: User reviews, no data police FB gave a historical clarification explaining why Provenance etc are not yet available. ======================================================================== Igor Chillingarian: EURO-3D --------------------------- UTYPE: axisCharacterization[axisFrame/ucd=em]/Coverage/Location/coord/Spectral/Value Flux Sampling: 1 ADU Flux Resolution: 1 electron Char -> axisChar -> axisFrame -> coverage -> location -> Bounds -> Support -> resolution -> RefVal ... GD: Why not 2 axes for spatial? IC: Schema does not allow it. FO: If schema does not allow it, than I cannot use it. JCM +charac people: there are many dimensions for the spatial axis and reference to a coordinate system. IC: Should we add a Signal to Noise Ratio property ? # Action: Discussion needed (Anita, Arnold, Igor,...) Discussion regarding the opportunity of describing the spectral resolution in km/s instead of wavelength (or frequency). IC: Velocity is a property of the object not of the data. AR: Freq. and Veloc. are model dependent, hence are different axes. JCM: Sampling should be in wavelength units for consistency. IC: Should we put the radial velocity as a possible axis? Arnold: Radial Veloc. is a interpretation. BH: What about the FITS WCS SPECTRAL paper? JCM: there is a velocity axis. but a wavelog axis is also possible. But the FITS WCS paper does not define a standard for Resolution. There was an argument about axis and units name: km/s may be radial velocity or red-shift? How we name axis according to units? BT: follow W3C specs for XML Laszlo Dobos: Space separated method for arrays? IC: small char attached to small parts of a dataset, how to do that? AR: Model is probably ok to extract from a 3D data cube a 1D spectrum. BT: data cubes: slices are supported in Quantity AR: Long slit spectroscopy? how to describe non-orthogonal spatial axes? ======================================================================== Arnold Rots: STC V 1.22 ----------------------- STCLite is not a subset schema of STC! STCLite is Space & Time only. Incorporation of Quantity in appropriate places being pursued. JCM: No subset of STC because...? Arnold: STC specifies things with attributes and not elements, not that extensible for this. Action: circulate documentation (done: see Arnold's email to dm@ivoa.net October 6, 2005 13:24:41 CET, hoping that CET is allowed in STC :-) ) RW: only gives the coordinate frame? Arnold's answer: No it defines it AND assign a label. RW: Is there a list of available coord frames? Arnold's answer: Yes, an enumerated list. RW: Space Time frame (string) a couple of numbers and a Time thing. Arnold's answer: No. You need to provide the Space Frame (with value), same with Time Frame, plus define a reference position for both of them, with no default reference position. # Issue: STC & STCLite, two incompatible though similar schemas -> confusion. FO: STC master piece of the VO, but now we have STC and STCLite with different schemas, etc. Answer: Stylesheet to transform one to the other. BH: From STCLite not possible to go to STC-heavy. JCM: XML issue. complexity of full astronomy means a lot of nested parameters. RW: VOEvent don't need Wavelength. Answer: In STC wavelength is now optional. PO: Make all attributes optional and that will solve the problem. AR: we want to provide tools/guidelines. FO: VOTABLE to cope with all kinds of DM. And we cannot include all the DMs in VOTable. Arnold's Answer: one could use xinclude files for Coord sys. Any smart parser that knows what the xinclude is does not need to go out there and get it. still required: a simple layer include files for coos system for VO table requirements. FB: What is public and what will become private is to be discussed. Matteo Guainazzi: Line Data Model (Version 0.2 26/9/2005) ================================================================ Atomic/Molecular line databases Line -> Levels -> QuantumState/QuantumNumber (energy, lifetime, stats weight...) -> Species (describe properties of atom/molecule/compound where the trnsition generating line occurs) -> Environment -> Process (what modifies a line in astrophysical plasma, inflows, outflows, cosmological redshift,...) -> Environment (Process depends on it) e.g. temperature, density... Do we need a general "Chemistry" model to reconstruct molecule or coumpounds (formulae) forom their element particles? A full list of "acceptable" quantum numbers will be published. User cases: - radio astronomical spectroscopy of the hyperfine structure in N2H+ (Caselli et al 1995) - ISO Line Database (see Isa Barbarisi's talk in the afternoon). Proposed recommendation by next Interop meeting. # Action: Vizier (FO) to provide feedback to Line Data Model (MG) FO: hundreds of tables exist in Vizier. Answer: suggestion welcome. Question? Why is a model of molecule composition needed? goal: reconstruction of mol from components This effort is modeling the content of actual databases. IC: is it possible way to trace if lines belong to same multiplet? AR: vibrational line with substructure? If quantum number can be given, then no problem. Requirement: QuantumNumber optional when not known. END OF SESSION