/[volute]/trunk/projects/theory/snapdm/doc/note/SimDB-note.html
ViewVC logotype

Contents of /trunk/projects/theory/snapdm/doc/note/SimDB-note.html

Parent Directory Parent Directory | Revision Log Revision Log


Revision 481 - (show annotations)
Wed May 14 07:38:34 2008 UTC (12 years, 6 months ago) by bourges.laurent
File MIME type: text/html
File size: 77063 byte(s)
added css samples (display only)
1 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
2 <html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
3 <head>
4 <title>IVOA Working Group - Internal Draft</title>
5 <meta http-equiv="content-type" content="text/html; charset=iso-8859-1" />
6 <meta name="keywords" content="IVOA, International, Virtual, Observatory, Alliance" />
7 <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" />
8 <meta name="maintainedBy" content="IVOA Document Coordinator, ivoadoc@ivoa.net" />
9
10 <link rel="stylesheet" href="http://volute.googlecode.com/svn/trunk/projects/theory/snapdm/css/ivoa_wg.css" type="text/css" />
11 <link rel="stylesheet" href="http://volute.googlecode.com/svn/trunk/projects/theory/snapdm/css/simdb-note.css" type="text/css" />
12 </head>
13
14 <body>
15
16
17 <div class="customCSSDisplay">
18 <h1>Custom CSS Classes :</h1>
19
20 <div class="revision">
21 Revision marks [revision] :
22 <br/>
23
24 <br/>normal text
25 <br/>
26 <br/>
27
28 <div>
29 <span class="citation">original text [citation]</span>
30 =>
31 <span class="proposal">replacement text [proposal]</span>
32 </div>
33
34 <br/>
35 Authors :
36 <div class="gerard">Gerard's comments [gerard]</div>
37 <div class="laurent">Laurent's comments [laurent]</div>
38 <div class="rick">Rick's comments [rick]</div>
39 <div class="patrizia">Patrizia's comments [patrizia]</div>
40
41 </div>
42 </div>
43
44
45 <div class="head">
46 <a href="http://www.ivoa.net/"><img alt="IVOA" src="http://www.ivoa.net/pub/images/IVOA_wb_300.jpg" width="300" height="169"/></a>
47 <h1>Simulation Database (SimDB)<br/>
48 Version 0.x</h1>
49 <h2>IVOA Theory Interest Group <br />Internal Draft 2008 April 19 </h2>
50
51
52 <dt>This version:</dt>
53 <dd><a href="http://www.ivoa.net/Documents/...">
54 http://www.ivoa.net/Documents/...</a></dd>
55
56 <dt>Latest version:</dt>
57
58 <dd><a href="http://www.ivoa.net/Documents/latest/...">
59 http://www.ivoa.net/Documents/latest/...</a></dd>
60
61 <dt>Previous versions:</dt>
62 <dt>Interest Group:</dt>
63 <dd><a href="http://www.ivoa.net/twiki/bin/view/IVOA/IvoaTheory"> http://www.ivoa.net/twiki/bin/view/IVOA/IvoaTheory</a></dd>
64 <dt>Author(s):</dt>
65 <dd><a href="http://www.ivoa.net/twiki/bin/view/IVOA/GerardLemson">Gerard Lemson</a> (editor)<br /></dd>
66 <dd><a href="http://www.ivoa.net/twiki/bin/view/IVOA/LaurentBourges">Laurent Bourges</a><br /></dd>
67 <dd><a href="http://www.ivoa.net/twiki/bin/view/IVOA/PatriziaManzato">Patrizia Manzato</a><br /></dd>
68 <dd><a href="http://www.ivoa.net/twiki/bin/view/IVOA/RickWagner">Rick Wagner</a><br /></dd>
69 <dd>others?</dd>
70 <hr/></div>
71
72 <h2><a name="abstract" id="abstract">Abstract</a></h2>
73 <p>In this note we propose that the IVOA develop a standard protocol for discovering simulations.
74 We will call this protocol the <i>Simulation Database</i> (SimDB). Implementations of the SimDB will allow users to query for
75 results of simulations in quite some detail and will provide links to services for accessing these
76 simulations. </p>
77 <p>The results presented in this note, which form the core of the peoposed standard, are one half of a concerted effort of the theory Interest Group that originally went by the name
78 S<i>imple Numerical Access Protocol</i> (SNAP), and is now split up in two parts. The second part defines protocols
79 for accessing the simulations data products themselves. This part will be written up in a separate Note
80 (Gheller, Wagner et al, in preparation), under the name Simulation Data Access Protocol (SimDAP).
81 </p>
82 <p>The current proposal is built around a UML data model describing simulations, a representation (mapping) of this model as a relational
83 database schema and a mapping to an XML schema.
84 We propose the relational schema to be the outer facade of a SimDB-TAP implementation which is to be queried using
85 <a href="http://www.ivoa.net/internal/IVOA/IvoaVOQL/ADQL-20080415.pdf">ADQL</a> <em class="todo">.@@ TODO update the ADQL link to later versions @@</em>
86 The XML schema provides type definitions from
87 which a machine readable serialisations of the model may be constructed. The schema also defines root elements for documents
88 describing SimDB-resources. The SimDB should return such documents for identified SimDB-Resources upon request, as an
89 alternative to the tabular (VOTable) results of ADQL queries.
90 In case updates are supported by a SimDB implementation, such documents may be sent
91 </p>
92 <p>
93 This Note describes use cases and requirements and the approach we have taken to define a specification
94 that and current state of the results. We feel that the results are
95 sufficiently far evolved that they can start following the formal IVOA standardisation track.
96 To this end it could be turned over to one of the existing working groups. If that is the decisions we feel
97 that the data modelling WG is closest to its scope, but there exist very strong links to Registry, Semantics, ADQL
98 and DAL as well. One might argue that a targeted WG for this effort alone might be as appropriate.
99 We leave the decision about this to the IVOA exec.
100 </p>
101
102
103
104 <div class="status">
105 <h2><a name="status" id="status">Status of this Document</a></h2>
106 This is a Note. The first release of this document was 2008 April 19.
107 <p></p><br />
108
109 <!-- Choose one of the following (and remove the rest)-->
110 <!--Note-->
111 <p>This is an IVOA Note expressing suggestions from and opinions of the authors.<br/>
112 It is intended to share best practices, possible approaches, or other perspectives on interoperability with the Virtual Observatory.
113 It should not be referenced or otherwise interpreted as a standard specification.</p>
114
115
116 A list of <a href="http://www.ivoa.net/Documents/">current IVOA Recommendations and other technical documents</a> can be found at http://www.ivoa.net/Documents/.
117
118 </div><br />
119
120 <h2><a name="acknowledgments" id="acknowledgments">Acknowledgments</a></h2>
121 <p>We thank various persons for useful discussions in the course of this work. First the participants of the
122 <a href="http://www.ivoa.net/cgi-bin/twiki/bin/view/IVOA/CambridgeTheoryWorkshopFeb06">Feb 2006 theory
123 workshop</a> in Cambridge, UK, where this work was started. Second the participants of the
124 <a href="http://www.ivoa.net/cgi-bin/twiki/bin/view/IVOA/GarchingSNAPWorkshop200704">April 2007 SNAP workshop</a> in
125 Garching, Germany, where the design started taking shape. Then we want to thank particularly the following persons
126 for useful discussions and feedback: Jeremy Blaizot, Klaus Dolag, Ray Plante, Volker Springel. We finally want to thank
127 participants to the theory sessions in the interoperability meetings in Victoria, Moscow, Beijing and Cambridge where parts
128 of this work was discussed.
129 </p>
130 <h2><a id="contents" name="contents">Contents</a></h2>
131 <div class="head">
132 <ul class="toc">
133 <li><a href="#abstract">Abstract</a></li>
134 <li><a href="#status">Status</a></li>
135 <li><a href="#acknowledgments">Acknowledgements</a></li>
136 <li><a href="#contents">Contents</a></li>
137 <li><a href="#sec1">1. Executive Summary</a></li>
138
139 <li><a href="#sec2">2. Overview</a></li>
140 <ul class="toc">
141 <li><a href="#sec2_1">2.1 SNAP &rArr; SimDB + SimDAP</a></li>
142 <li><a href="#sec2_3">2.3 Simulation Database: structure and interface</a></li>
143 <li><a href="#sec2_3">2.3 Registration</a></li>
144 <li><a href="#sec2_4">2.4 Technology: UML, XMI, XSLT</a></li>
145 <li><a href="#sec2_5">2.5 Reference implementations</a></li>
146 </ul>
147
148
149 <li><a href="#sec3">3 Usage scenarios</a></li>
150 <ul class="toc">
151 <li><a href="#sec3_1">3.1 "20 questions"</a></li>
152 <li><a href="#sec3_2">3.2 SimDB-standard implementation</a></li>
153 <li><a href="#sec3_3">3.3 Legacy database</a></li>
154 <li><a href="#sec3_4">3.4 Meta data production pipe line</a></li>
155 <li><a href="#sec3_5">3.5 Client tools</a></li>
156 </ul>
157
158 <li><a href="#sec4">4 Analysis model</a></li>
159 <ul class="toc">
160 <li><a href="#sec4_1">4.1 Universe of Discourse</a></li>
161 <li><a href="#sec4_2">4.2 <i>Domain Model for Astronomy</i></a></li>
162 <li><a href="#sec4_3">4.3 SimDB analysis model</a></li>
163 </ul>
164
165 <li><a href="#sec5">5 Logical model</a></li>
166 <ul class="toc">
167 <li><a href="#sec5_1">5.1 Overview</a></li>
168 <li><a href="#sec5_2">5.2 Normalisation</a></li>
169 <li><a href="#sec5_3">5.3 Model content</a></li>
170 <li><a href="#sec5_3_1">5.3.1 Resource hierarchy</a></li>
171 <li><a href="#sec5_3_2">5.3.2 Object types</a></li>
172 <li><a href="#sec5_3_3">5.3.3 Target</a></li>
173 <li><a href="#sec5_3_4">5.3.4 Characterisation</a></li>
174 <li><a href="#sec5_3_5">5.3.5 Semantics</a></li>
175 <li><a href="#sec5_3_6">5.3.6 Units</a></li>
176 <li><a href="#sec5_3_7">5.3.7 Services</a></li>
177 </ul>
178
179 <li><a href="#sec6">6 Physical models</a></li>
180 <ul class="toc">
181 <li><a href="#sec6_1">6.1 Identity and referencing</a></li>
182 <li><a href="#sec6_2">6.2 RDBM Schema</a></li>
183 <li><a href="#sec6_3">6.3 XML Schema</a></li>
184 <li><a href="#sec6_4">6.4 UTYPE-s</a></li>
185 <li><a href="#sec6_5">6.5 JAVA/JPA+JAXB (non-normative)</a></li>
186 </ul>
187
188 <li><a href="#sec7">7. Query protocols</a></li>
189 <ul class="toc">
190 <li><a href="#sec7_1">7.1 ADQL</a></li>
191 <li><a href="#sec7_2">7.3 REST</a></li>
192 <li><a href="#sec7_3">7.2 TAP?</a></li>
193 </ul>
194
195 <li><a href="#sec8">8. Next steps</a></li>
196 <ul class="toc">
197 <li><a href="#sec8_1">8.1 Reference implementations</a></li>
198 <ul class="toc">
199 <li><a href="#sec8_1_1">8.1.1 France</a></li>
200 <li><a href="#sec8_1_2">8.1.2 Germany</a></li>
201 <li><a href="#sec8_1_3">8.1.3 Italy</a></li>
202 <li><a href="#sec8_1_4">8.1.4 USA</a></li>
203 </ul>
204 <li><a href="#sec8_2">8.2 SimDAP services</a></li>
205 </ul>
206 <br/>
207 <li><a href="#appA">Appendix A: Data modelling specifics</a></li>
208 <li><a href="#appB">Appendix B: XSLT pipe line</a></li>
209 <li><a href="#glossary">Glossary and Acronyms</a></li>
210
211 <li><a href="#references">References</a></li>
212 </ul>
213 </div>
214 <hr/>
215
216
217 <br/>
218 <h2><a name="sec1">1. Executive summary</a></h2>
219 <em class="todo">@@ TODO Modify this text, which was originally an email to be sent to THEORY, TCG, DM, maybe EXEC @@</em>
220 <p>
221 We propose to derive two WG projects from what was so far the
222 SNAP project of the theory interest group: SimDB and SimDAP.
223 In this note we discuss the first of these, SimDB, in some detail.
224
225 </p>
226 <h3> Simulation Database (SimDB)</h3>
227 <p>We propose to developa standard specification project, called the "Simulation Database" (SimDB).
228 It is based on the description+discovery part of the old
229 SNAP project. Its normative deliverables are
230 <ul>
231 <li> A logical data model for describing simulations.<br/>
232 Following SNAP we keep concentrating
233 on 3+1D simulations, with which we mean simulations modelling a
234 space-time sub-volume of the universe OF ANY SIZE, so not only large
235 scale structure, galaxy clusters, but everything down to asteroid collisions etc.
236 As the model <i>describes</i> simulations, it may be called a meta-data model.
237 It will be a logical model in the sense of standard data modelling approaches <em class="todo">@@TODO add some references@@</em>,
238 and is based on an analysis, or domain model which is presented but not normative.
239 The logical model is presented in fully detailed and documented UML2, serialised
240 to XMI 2.1, created using the MagicDraw 12.1 Community edition tool.
241 The data model is using a small subset of UML2 and has some UML profile
242 extensions added. Together this can be seen as a domain specific language,
243 and this can be formalised in a UML Profile. We will propose using such a profile
244 to the DM working group as a general approach for DM efforts.
245 </li>
246 <li>A query protocol based on the logical model.
247 <br />We propose this to have at least an ADQL version.
248 To this end we will provide a relational mapping.
249 This physical model is completely derived from the SimDB logical model using rules
250 implemented as a pipe-line of XSLT2 scripts working on the XMI representation of
251 the UML. The scripts will produce relational database DDL scripts defining the
252 database schema. That schema itself is not normative, instead we will define the
253 replies to TAP metadata queries. We provide implementaiton scenarios in the text below,
254 for the case of someone using the results from this project completely and for the
255 case of someone implementing a SimDB on top of a legacy database.
256 </li>
257 <li> a messaging format for sending instances of the various components
258 in the data model around.
259 <br />This format will be based on a number of XML
260 schema documents (XSDs), one of which contains the root elements defining valid SimDB resources.
261 This requires a mapping from the UML to XSD.
262 This mapping will take the form of one or more XSLT documents.
263 </li>
264 <li> An IVOA working draft document describing these components.
265 <br />This will be based on the current document.</li></ul>
266 </p>
267 <p>
268 We introduce some non-normative solutions that can be taken over for generic
269 data models (this is ofcourse also true for the UML/XMI+XSLT approach for the
270 normative standards).
271 <ul>
272 <li> The XSLT scripts we propose above do not work on the XMI itself, but on
273 an intermediate representation of the UML data model. This is an XML dialect
274 based on a schema we define and which captures the UML profile more directly.
275 XMI is very generic and rather cumbersome to work with. The representation of
276 the UML in our intermediate XML form is much more readable and XSLT based on it
277 is much simpler. It also allows easier adaptation to future modifications in UML,
278 or to tools whose XMI representation is different from the standard. We only need
279 to update the XMI->Intermediate XSLT transformation scripts. Not the more complex
280 transformations to the other official representations.
281 We will propose a similar approach to the DM WG.
282 </li>
283 <li> We will provide XMI->Java+JPA+JAXB transformation scripts in XSLT (properly, intermediate->Java).
284 These scripts generate Java classes corresponding to the types (Class, DataType, Enumeration)
285 in UML. These classes are annotated with Java Persistence Architecture (JPA)
286 and Java Architecture for XML Binding (JAXB) attributes to assist in the transformation
287 between relational database and XML representations.
288 Similar scripts can be written for C#. C# allows the same annotations as Java 5 supports
289 already for longer. For persistence we will likely use Linq, which seems similar to JPA.
290 </li>
291 <li>We propose an approach for including application specific and legacy simulation databases
292 in this framework. This approach follows the "global-as-view" approach to information
293 integration (see for example http://www.deg.byu.edu/papers/PODS.integration.pdf;
294 Leonid Kalinichenko from the RVO is an expert in this field).
295 Implementors with an existing relational database schema may be able to define database
296 views which implement the relational representatiopn of the SimDB data model,
297 and in this way provide a simple way to support querying of their database using ADQL.
298 </li></ul></p>
299 <h4>organisation</h4>
300 <p>
301 The SimDB is ready to be transferred to the DM WG.
302 <br />We propose that Gerard Lemson keeps leading this effort (as main editor), also when it is moved
303 to that WG. The DM WG's chair (Mireille Louys) will be responsible all WG-chair
304 issues associated with moving a specification through the document process.
305 The people at the bottom will be part of a "tiger team" to push the standard to RFC.
306 We may want to expand this group with an expert from each of the WGs mentioned below.
307 </p>
308 <p>
309 We have been discussing the data model for some time now.
310 Various projects (Italy, USA, France and Germany) have implementations that are similar
311 to the envisioned SimDB. We believe that by autumn 2008 it can go to RFC.
312 Patriza Manzato and Rick Wagner will have reference implementations based on existing DBs,
313 so will various projects in France (Lyon: Jeremy Blaizot and Laurent Bourges;
314 Galmer database: Igor Chillingarian) and GAVO.
315 </p>
316 <p>
317 Other relevant working groups for this process are Registry, ADQL and Semantics, possibly DAL.
318 Registry because the simulation database is similar to a registry. We can
319 learn from implementations and the registry interface. Also, we (think we) may need an
320 extension to the IVO Identifier in the implementation of references in SimDB.
321 ADQL because we propose it to be the standard (main) query interface to a SimDB implementation.
322 Semantics because our model includes usage of semantic vocabularies, maybe full ontologies
323 DAL because we our proposal for using ADQL in the query phase requirs a version of
324 the TAP protocol for defining the interface.
325 We would like to include a person from each of these WGs in the tiger team.
326 Our wishes are: Ray Plante (Registry), ? (ADQL), Norman Gray (Semantics), (?) TAP.
327 Ray and Norm have contributed to early discussions about SNAP.
328 </p>
329 <p>
330 Of these other efforts it seems TAP offers the main risk for the SimDB standard to go to
331 RFC by the Autumn. What may help us is that we do not need all the details of TAP.
332 In particular the information_schema approach allowing users to
333 query for the data model is not required as it is part of SimDB specification.
334 We mainly need a prescription for sending ADQL queries to the SimDB, and what the
335 format of results should be.
336 Since we expect meta-data databases to be relatively small (compared to
337 say an SDSS or Millennium database), we expect fewer, if any problems with
338 performance and can stick to synchronous behaviour at first.
339 </p>
340 <p>
341 We may need some explicit registry-interface like features such as returning a
342 complete XML document according to the messaging format of the SimDB data model.
343 Other issues will come up during the next phase of the discussions.
344 </p>
345
346 <h3>Simulation Data Access Protocol (SimDAP)</h3>
347 <p>
348 The second spin-off of the SNAP project we propose we rename to <i>Simulation Data Access Protocol</i> (SimDAP).
349 It deals with accessing the data after discovery by some means,
350 likely trough an implementation of a Simulation Database.
351 It should handle special services such as cut-out, projection,
352 extraction (AMR-like cut-outs, produces regular grids), but also staging etc.
353 It should also deal with data formats. Claudio Gheller (Italy) is leading
354 this effort with close help of Rick Wagner (USA).
355 </p>
356 <p>
357 This project needs more fleshing out and is hopefully ready to be transmitted
358 to a WG, likely DAL by the Autumn interop.
359 </p>
360 <h3>Connections between SimDB and SimDAP</h3>
361 <p>
362 The two projects are connected as follows:
363 The meta-data formats to be included in SimDAP messages are derived from
364 the data model of the SimDB.
365 Vice versa, the SimDB will include a component describing
366 which SimDAP services are applicable/available for a given simulation.
367 </p>
368
369 <!-- ++++++++++++++++++++++++ -->
370 <h2><a name="sec2"/> 2 Overview</h2>
371
372 <h3><a name="sec2_1"/>2.1 SNAP &rArr; SimDB + SimDAP</h3>
373 <p>This document presents a model for describing certain types of numerical computer simulations
374 and certain types of simulation post-processing products. The model was oringinally envisioned to
375 be used in the query part of the <i>Simple Numerical Access Protocol</i> (SNAP),
376 and in discovery of interesting SNAP services in the first place.
377 After investigating the application domain carefully, we have decided to leave the concept of
378 designing a DAL-like SxAP protocol for simulations. Instead we have split up the effort into
379 two separate efforts that can be used each in their own right, though their is a clear link between them.
380 This document discusses the firsts of these, which we have named the <i>Simulation Database</i>, and
381 will have the acronym <i>SimDB</i>. The second will be developed further in a separate effort amd is
382 called the <i>Simulation Data Access Protocol</i> (SimDAP, "Sim" stands for "Simulation", <i>not</i> "Simple"!).
383 </p>
384 <p>
385 Following SNAP, SimDB only explicitly considers simulations for systems that represent a space-time
386 sub-volume of the universe and (part of) its material contents. Examples of such simulations are
387 cosmological, pure dark matter N-body simulations of the large-scale structure of the universe;
388 adaptive mesh refinement (AMR) simulations following the evolution of a galaxy cluster using full hydrodynamics;
389 a simulation of the evolution of a globular cluster using a combination of tools, together simulating
390 the various types of physics <em class="todo">@@ TODO reference to MODEST-like activities</em>; or
391 simulations calculating the few seconds of a super nova explosion in full 3D.
392 </p>
393 <p>
394 In general these simulations will evolve this system forward
395 in time and are able to produce <i>snapshots</i>, representing the state of the system, a 3D volume of space,
396 at a number of discrete times (though there are alternatives: light cone simulations, individual particle orbits).
397 These direct, raw results of simulations we call Level-0 products, following
398 similar terminology for observations.
399 SimDB also covers Level-1 products, which consist of the results of certain types of post-processing
400 of simulations, namely those products that in some form create an alternative representation of
401 a spatial sub-volume of the universe. For example a density field calculated on a regular grid, derived
402 created from an N-body or an AMR simulation; a cluster catalogue derived using some group finder applied
403 to a cosmological simulaiton, or a synthetic galaxy catalogue derived from the cluster catalogue using
404 halo occupation distribution models (HODs) or semi-analytical models (SAMs).
405 </p>
406 <p>
407 We do not make any restrictions on the type of systems being simulated, or the size of the
408 simulation, or the way the system is represented in the simulation code and results. We also
409 make no restrictions on the type of "observables" produced by the simulations.
410 </p>
411 <p>
412 The SimDAP
413 specification will includes protocols for services that process level-0 or level-1 results and produce
414 other level-1 results. The allowed services deal with selecting the results in a
415 sub-volume of the complete result, sampling a regular 3-dimensional grid, etc. SimDAP also allows for
416 services, that do not produce SimDB-like, level-0 or 1 products. Examples are projections, 1D or 2D samplings.
417 But also custom services will be allowed, for example calculating statistical properties such as correlation
418 functions or power spectra in cosmological simulations. A more detailed description of SimDAP
419 is outside of the main scope of this note.
420 </p>
421 <h3><a name="sec2_2"/>2.2 Simulation Database: structure, interface and applicable services</h3>
422 <p>
423 SimDB is a specification that defines the interface to a database containing meta data describing
424 simulations. To this end it contains two main parts, one is a model for the meta data, the other
425 a protocol for interacting with the database. The model is the core of the specification.
426 It describes the structure of individual data products in the database. We have chosen UML
427 as modelling language, as prescribed by the data modelling working group in the interoperability meeting
428 in Cambridge, UK, May 2003.
429 </p>
430 <p>
431 The UML model is a logical model (see [..] <em class="todo">@@ TODO add reference @@</em>) and
432 forms the basis for physical representations of the data products in the standard
433 language that the IVOA has chosen for such purposes, XML. We derive an XML schema defining valid
434 XML documents directly from the logical model. The SimDB interface will include functions for insetting
435 SimDB data products using such documents, and for retrieving individual, identified data products.
436 </p>
437 <p>
438 The logical model also forms the basis for a physical representation supporting formulation of queries.
439 For various reasons explained below we have chosen ADQL to be the query language and accordingly we derive
440 from the model a relational schema that defines the tables and columns that can be used in ADQL queries sent
441 to a SimDB implementation. The result of ADQL queries is supposed to be a VOTable, and this will in general
442 not represent a complete SimDB data product. However it can be used to browse the database, finally identifying
443 resources and possibly requesting these from the SimDB as XML documents.
444 </p>
445 <p>
446 We make very limited assumptions on <em>how</em> a data product discovered in a SimDB can actually be accessed.
447 We only assume there is a web-based service available, identified by a base URL and tagged with a service type.
448 The range of service types will be defined by SimDAP, but it will at least include "download" and "custom".
449 The data model contains an explicit element for indicating which services are available for a given data product,
450 and users may, if they wish, retrieve this information through ADQL queries and follow the links directly.
451 SimDB implementations can and likely will eventually provide SimDAP related functionality, but this is not part
452 of this specification.
453 </p>
454 <h3><a name="sec2_3"/>2.3 Registration</h3>
455 <p>
456 It must be possible to find SimDB instances in an IVOA Resource Registry <am class="todo">@@TODO add references&&</am>.
457 This implies we need a corresponding resource type, and we have to design its structure.
458 We also assume that one may define resources in the sense of [...]
459 <em class="todo">@@ TODO add reference to Resource data model document @@</em>
460 from within the contents of a SimDB. We take this into account explicitly in the model.
461 The SimDB will have a "getIVOAResource" function, which will execute the appropriate transformation from
462 the internal representation of the SimDB data products to the Resource model's XML representation [...]
463 <em class="todo">@@ TODO link to Resource XML schema document@@</em>.
464 This will likely put more requirements on the Registry model itself, maybe requiring extensions to its schema.
465 Possibly a SimDB itself can be an extension registry. This we think can be postponed to a future version of the
466 specification.
467 </p>
468 <h3><a name="sec2_4"/>2.4 Technology: UML, XMI, XSLT</h3>
469 <p>
470 We
471 </p>
472 <h3><a name="sec2_5"/>2.5 Reference implementations</h3>
473 <!-- ++++++++++++++++++++++++ -->
474
475 <h2><a name="sec3"/>3 Usage scenarios</h2>
476 <em class="todo">@@ TODO needs severe editing @@</em>
477 We have assembled a list of explicit use cases and scenarios from which we derive
478 requirements for the current model and the SNAP protocol.
479 <h4><a name="sec3_1"/>3.1 "20 questions"</h4>
480 <p>
481 SimDB defines a common data model for simulations.
482 Following the good practice for database design initiated in [], we here provide a number of
483 scientific questions one might want to ask such a database. The data model and associated data
484 access protocol need to be sufficiently rich that they can support such questions.
485 </p>
486 <ul>
487 <li> Scientific goal: investigate baryon wiggles in the evolved density field<br/>
488 Query: Return all cosmological, pure dark matter, N-body simulations with WMAP 3 initial
489 conditions and a box size of at least 1000 Mpc comoving, containing snapshots at about
490 10 redshifts between 3 and 0.
491 </li>
492 <li> Scientific goal: investigate whether observed structures in X-ray cluster that seem to
493 indicate turbulence, can truly be that.<br> Query: return all hydro-dynamical simulations of
494 galaxy clusters of mass at least 1o<sup>14</sup> M<sub>sun</sub>,
495 that have a model for viscosity included in the simulation.
496 Moreover, return only those simulations that have associated to them an online visualisation
497 service that can produce projected temperature and pressure maps.
498 </li>
499 <li> Scientific goal: interpret the possible histories of an observed galaxy merger to calculate
500 possible star formation episodes and compare these to the observed stellar populations.<br>
501 Query: Return all simulations of galaxy mergers where the component galaxies have a particular
502 mass ratio and where there are enough snapshots to follow the evolution over a few Gyr.
503 </li>
504
505 <li> Scientific goal: compare the luminosity function of galaxies in the SDSS survey with those
506 in synthetic catalogues.<br>Query: Select all cosmological simulations that have produced as
507 secondary product synthetic galaxy catalogues on a light-cone and provide those via an SQL (ADQL?)
508 query interface.
509 </li>
510 <li> ...
511 </li>
512 </ul>
513 <p>
514 In the design of the model it is useful to think about the steps a user might go through
515 when querying a database system in various "drilling down" steps. For example the following
516 questions might be asked :
517 </p>
518 <p>
519 <ul>
520 <li>What system/object is being simulated?</li>
521 <li>What physical processes are included?</li>
522 <li>How is the system being represented in the simulation
523 (particles (Langrangian), (adaptive) mesh (Eulerian)), both, other?</li>
524 <li>Per process:<ul>
525 <li>How are the physical processes implemented ?</li>
526 <li>Characterise the numerical approximations (.e.g. resolution, softening parameter)</li></ul></li>
527 <li>What observables are available for the system/object, possibly as function of time?
528 As it is a spatial system, at least size, center-of-mass position.</li>
529 <li>What observables are available for the constituents, i.e. what is the schema of the atomic objects?</li>
530 <li>Per snapshot, per atomic object type, per variable:
531 <ul>
532 <li>Characterise the possible values</li>
533 <li>Characterise the result</li></ul></li>
534 <li>Are post-processing results available?</li>
535 <li>Are services/applications available working on the results?</li>
536 <li>Which code ran the simulation?</li>
537 <li>What were values of physical parameters?</li>
538 <li>How were initial conditions created, what parameters?</li>
539 </ul>
540 </p>
541
542 <h4><a name="sec3_2"/>3.2 SimDB-standard implementation</h4>
543 We foresee a simple implementation scenario based directly on products developed
544 in the course of the SimDB effort. We believe that from the data model to be developed
545 in this effort we should be able to derive physical representations that
546 can be used directly in implementations. We envisions that with only a little custom infrastructure code
547 it should be possible to
548 <ul>
549 <li>fill a relational database with tables and views representing the SimDB data model from
550 DDL scripts generated from the UML</li>
551 <li>create a web-based service that accept XML documents for inserting new simulation results
552 and translates these, using generated code with JAXB annotations, to in memory Java objects</li>
553 <li>flush these objects to a relational database using the Java Persistence Architecture (JPA) implementation,
554 structured using the JPA annotations generated on the Java classes.
555 It should be not too hard to support other languages as well if they provide similar simple XML binding and
556 OR-mapping capabilities. Python+Django and C#+LINQ or NHibernate come to mind.<em class="todo">
557 @@ TODO check with people knowing more about these technologies @@</em></li>
558 <li>accept ADQL queries that are translated to the appropriate vendor specific SQL
559 (using modules defined by the ADQL effort?) and return a VOTable</li>
560 <li>accept requests for identified SimDB resources (using an IVO or implementation specific identifier),
561 translate this into a JPA query to retrieve the object form the database, which is translated to
562 the appropriate XML using the JAXB layer and sent back to the user.</li>
563 </ul>
564
565 <h4><a name="sec3_3"/>3.3 Legacy database</h4>
566 Although by no means as common as similar efforts in the observational domain,
567 databases have been developed containing the meta data of simulations.
568 How could a SimDB be implemented around such a database.
569 Our ideas are inspired by (what we understand from) the "global-as-view" approach to information
570 integration. We assume the implementers have their own way of filing up their database with meta-data
571 describing simulations from their own efforts. The idea is that they write database views to provide
572 a virtual implementation of the SimDB/RDB schema. ADQL queries sent to their service can now still be
573 understood and replied to. The users should also be able to write custom code to produce the appropriate
574 XML documents based on a request for an identified resource, possibly by querying these same views.
575
576 <h4><a name="sec3_4"/>3.4 Meta data production pipe line</h4>
577 The SimDB data model is relatively comprehensive, which reflects itself in XML documents
578 of substantial size ad complexity for realistic cases.
579 For a registration scenario, i.e. one where a user is allowed to upload XML documents to a SimDB implementation,
580 one would prefer not to have to produce these documents by hand. By far the preferred manner in our opinion
581 would be for simulation and post-processing pipe-lines to produce compliant documents.
582 We have contacted authors of some of the most popular major simulation codes (Springel; Norman et al; more needed),
583 and they have agreed that this is feasible and are willing to participate in this effort.
584
585 <h4><a name="sec3_5"/>3.5 Client tools</h4>
586 One reason to produce a standard which uses ADQL on top of a standard data model is that client tools
587 can be written to query different such holdings. For example we could envision a tool such as VisIVO [..]
588 to offer some user-friendly interface for querying SimDB implementations retrieved from an IVOA Registry.
589 The user need to see any ADQL, that is all generated by VisIVO, but can be shown results and services.
590 In particular if a cut-out service is available, VisIVO could provide an interface for the user to decide
591 on the sub-volume, retrieve and visualise it. The advantage of having a standard data model
592 clearly is that the same ADQL can be sent to all SimDB services.
593 <em class="todo">@@ TODO contact VisIVO people to see whether this could be implemented @@</em>.
594
595 <!-- ++++++++++++++++++++++++ -->
596
597
598 <h2><a name="sec4"/>4 Analysis model</h2>
599 <p>
600 <em class="todo">@@TODO Gerard @@</em>
601 An <i>analysis model</i>, also called domain model, is an abstract, high-level representation of the
602 <i>universe of discourse</i> (UoD), the part of the world that our application deals with.
603 It is a UML model, with emphasis on the concepts and their exact relationships in the UoD, though details
604 such as attributes need not be completely filled in.
605 Importantly, it should not be influenced by application scenarios apart form knowledge of their UoD.
606 Here we describe the UoD and our analysis model. The model is strongly influenced by patterns
607 discovered in earlier work on a
608 <i><a href="http://www.ivoa.net/internal/IVOA/IvoaDataModel/DomainModelv0.9.1.doc">Domain model for Astronomy</a></i>,
609 co-written by one of the authors of the present note. We describe some of its main patterns below as well.
610 <em class="todo">@@ TODO or will we? @@</em>
611 </p>
612 <h4><a name="sec4.1"/>4.1 Universe of Discourse</h4>
613
614 <h4><a name="sec4.2"/>4.2 Domain Model for Astronomy</h4>
615
616 <h4><a name="sec4.3"/>4.3 SimDB analysis model</h4>
617 <em class="todo">@@TODO create a version and add it to volute@@</em>.
618
619 <!-- ++++++++++++++++++++++++ -->
620
621 <h2><a name="sec5"/>5 Logical Model: SimDB</h2>
622 <p>
623 Here we introduce the core of our proposal, the UML representaiton of our logical data model
624 for our Simulation Database. The exact representation of this model is an
625 <a href="http://volute.googlecode.com/svn/trunk/projects/theory/snapdm/input/SimDB_DM.xml">XMI file</a>,
626 which can be found in the <a href="http://volute.googlecode.com/svn/trunk/projects/theory/snapdm">snapdm section</a>
627 of the <a href="http://volute.googlecode.com/svn/">Volute subversion database</a> on Google code.
628 Other representations can be found in that same hierarchy, in particular check out the
629 <a href="http://volute.googlecode.com/svn/trunk/projects/theory/snapdm/output/html/SimDB.html">HTML documentation</a> which we generated from the XMI
630 representation with the XSLT pipeline described in <a href="#appB">Appendix B</a>. This generated documentation file contains
631 the explicit description of all of the elements in the model and forms the reference documentaiton document for the model.
632 </p>
633 <h3><a name="sec5_1"/>5.1 Overview</h3>
634 <p>
635 The logical data model is a fully detailed model of the application domain. It is to form the basis of physical
636 models, representing the model in various computational environments.
637 The logical model is represented as a set of UML diagrams, which we created using MagicDraw Community Edition 12.1 and stored as an
638 XMI file in the GoogleCode
639 SVN repository: <a href="http://volute.googlecode.com/svn/trunk/projects/theory/snapdm/input/SNAP_Simulation_DM.xml">
640 SNAP_Simulation_DM.xml</a> <em class="todo">@@TODO should change all occurrences of names with SNAP to using SimDB@@</em>
641 JPG representations of the model can be found in <a href="http://volute.googlecode.com/svn/trunk/projects/theory/snapdm/input/images/">this</a>
642 directory. <em class="todo">@@TODO find proper representation image of the complete model. Possibly color packages differently.@@</em>
643 </p>
644 <h3><a name="sec5_2"/>5.2 Normalisation</h3>
645 <p>
646 We have tried to find a balance in the level of <i>normalisation</i> of the data model.
647 </p>
648 <h3><a name="sec5_3"/>5.3 Model contents</h3>
649 <p>Here we discuss the actual contents of the model, though the detailed descritpion </p>
650 <h4><a name="sec5_3_1"/>5.3.1 Resource hierarchy</h4>
651 <p>
652 At the root of the SimDb data model is an abstract class called Resource, in the rest
653 of this document we will refere to this as SimDB/Resource.
654 It represents the different types of highest level meta-data objects to be stored in a SimDB.
655 Examples of this are represented as subclasses. First Experiment (SimDB/Experiment), which represents
656 different types of experiments that have been performed (run/executed/...) and have produced the results
657 that SimDB users may be interested in. Examples of SimDB/Experiment-s are first simulations,
658 but also the various post-processing operations transforming simulation results into other products
659 such as halo catalogues, density fields etc.
660 </p>
661 <p>
662 The second major type of SimDB/Resource is the SimDB/Protocol.
663 This concept represents a <i>formally prescribed way of doing an experiment</i>.
664 It is derived from the concept with the same name in the domain model, which itself was inspired
665 by the concept with the same name in Chapter 8.5 in <a href="#r_AnalaysisPatterns>[3]</a>.
666 In the SimDB/DM this concept has concrete representations in the computer programs that are being
667 used to run simulations and post-processing etc. As such it defines the possible input parameters,
668 possble algorithms, the kind of results that can be produced by the code. Every SimDB/Experiment must
669 indicate which SimDB/Protocol was used and for example provide values for the input parameters, indicate
670 which physics was used
671 </p>
672 <p>
673 The SimDB/Resource concept is clearly similar, but in general <i>not equivalent</i> to the Resource Registry's Resource concept.
674 In data modeling terms, it is not true that a SimDB/Resource <i>is a</i> Registry/Resource.
675 Often the reason is similar to the reasons that a single image is not a Registry/Resource, whereas a SIAP-compatible service is.
676 The granularity of a SimDB will be higher than a Registry and many simulations on their own will be too small.
677 The SimDB itself will have to be registered (see <a href="#">section ???</a> for a further discussion
678 <em class="todo">@@ TODO add propoer section and href@@</em>),
679 i.e. a SimDb service <i>is a</i> Registry/Resource. In discussion with Ray Plante (IVOA Interop May 2007, Beijing)
680 on this issue it was proposed that some part of the contents could also be registered in a Registry directly.
681 I.e. we should be able to identify Registry/Resource-s in SimDB. Considerations to decide on how to make this identification would be for example
682 that all data products resulting form a well defined (and published) scientific project could qualify.
683 To represent such a possibility for now we have introduced another subclass of SimDB/Resource: SimDB/Project.
684 This is not much more than an aggregation of experiments, with some additional atrributes describing the motivation etc.
685 The metadata of a SimDB/Project is not the same as that of a Registry/Resource, however we propose that we should be able
686 to define a transformation (possibly implemented again in XSLT) to transform a SimDB/Project and produce a Registry/XML representation.
687 Some more thoughts on this subject will be given in <a href="#">section ???</a> <em class="todo">@@ TODO add proper section and href@@</em> mentioned above.
688 </p>
689 <ul>
690 <li>Should we define explicit transformations for SimDB/Resource -> Registry/Resource ?</li>
691 </ul>
692
693 <h4><a name="sec5_3_2"/>5.3.2 Object types</h4>
694 <p>
695 One of the main differences between the SimDB data model and other data models in the IVOA so far, is that we do not
696 know in advance what types of results we can expect. The Spectrum data model describes spectra, the characterisation data model
697 characterises observational results, the model implicit (for now) in SIA deals with 2D images etc.
698 This implies that many features describing these results can be explicitly modeled: Spectra have been taken of sources on the sky,
699 during a certain time persiod, covering a certain wavelength range. And fluxes were measured (in some form).
700 This makes it possible for space/time/wavelength/flux to make explicit appearances in the corresponding models.
701 </p>
702 <p>
703 In contrast, simulations come in a great variety of types, even if we constrain ourselves to the "3+1D" kind.
704 We can not make many assumptions on the type of objects making up the result, or on the "observables" of these objects.
705 It therefore becomes necessary to add components to the model that allow publishers to describe these explicitly.
706 We do so in the ObjectType hierarchy.
707 </p>
708 <ul>
709 s<li>Should we add units to the properties and not to the characterisations; similar for InputParameter and ParameterSetting</li>
710 </ul>
711
712 <h4><a name="sec5_3_3"/>5.3.3 Target</h4>
713 <p>
714 The first question most people want to know about a simulations is: "what is being simulated?".
715 The answer should correspond to a real (astronomical) object, or collection of objects,
716 or possibly a physical process. For SimDB to answer such questions implies that publishers must be
717 able to describe these concepts in the model.
718 We have introduced the TargetObjectType and TargetProcess classes for this.... <em class="todo">@@ TODO expand @@</em>.
719 </p>
720
721 <h4><a name="sec5_3_4"/>5.3.4 Characterisation</h4>
722 <p>
723 Much of the metadata in the model concerns itself with describing how the results that are supposedly the ultimate
724 goal of users, describing the kind of objects contained in the results and the scientific content. In an implementation of
725 the model as a database, one does not expect the actual data to be stored and therefore there is no need to have model elements
726 describing these. However there is some use to getting a summary of the actual data values, both obtained, and obtainable.
727 To this end we have added the characterisation elements, in reference to the <a href="#r_Characterisation">Characterisation data model [5]</a>
728 that, as we will try to explain, performs a similar function for observations.
729 </p>
730 <p>
731 <em class="todo">@@ TODO expand @@</em>
732 </p>
733 <ul class="issue">
734 <li>We need to add characterisation to TargetObject, so users can ask for "simulations of 1e14 M<sub>sun</sub> galaxy clusters</li>
735 <li>Do we need other types of characterisation, such as accuracy etc?</li>
736 </ul>
737
738 <h4><a name="sec5_3_5"/>5.3.5 Semantics</h4>
739 <p>
740 There are many instances in the data model where we need to describe elements of the
741 SimDB/Resource-s explicitly, because we do not have implicit information based on the context.
742 Examples are the various properties of object types, the target objects and processes etc.
743 Apart from a name and a description we then frequently add
744 an attribute which is supposed to "label" the element according to an assumed standard list of terms.
745 We model this using the <pre>&lt;&lt;ontologyterm&gt;&gt;</pre> stereotype. Attributes with this stereotype
746 are assumed to take their values form such a predefined "ontology". See
747
748 </p>
749 <ul class="issue">
750 <li>We need to list the onotlogies that we create first attempts on some that do not yet exists.</li>
751 <li>We need to have proper locations of machine readable vocabularies</li>
752 <li>We need to get feedback on what kind of ontologies we want. Narrower/broader types, fully linked ontologies? etc</li>
753 </ul>
754 <h4><a name="sec5_3_6"/>5.3.6 Units</h4>
755 <p>
756 The current (May 2008 <em class="todo">@@ TODO update when necessary @@</em>) version of the model
757 allows publishers to specify numerical quantities using a real value and a unit.
758 I.e. we do not prescribe units for particular quantities.
759 Allowing this flexibility in units assignment does pose a problem for a query interface that allows user to query on
760 characterisation values and other numerical quantities. ADQL does not include units for example, but a user
761 can not assume that every publisher will use the same unit for for example the typical size of a simulation box.
762 This is even worse of course for the characterisation values of properties that have to be defined
763 in the model and can have any kind of assumed unit.
764 </p>
765 <p>
766 We believe we should treat units as a special semantic vocabulary, possibly an ontology.
767 This implies we push its development off to elsewhere for now, and assume we can
768 at some point use a standard list of units in a similar way to the other ontology references.
769 Maybe this could include a link to the physical quantity (etc, see for example the
770 <a href="http://physics.nist.gov/cuu/Units/introduction.html">NIST reference on SI</a>) to which the unit applies.
771 </p>
772 <p>
773 If this kind of link can be made, we could eventually attempt to impose a single unit to correspond to
774 all properties sharing a given <a href="http://physics.nist.gov/cuu/Units/introduction.html">quantity in the general sense</a>.
775 This may lead to very small or very large values, depending on the simulation, but at least allows simpler
776 interfaces.
777 </p>
778 <ul class="issue">
779 <li>We need input from the rest of the IVOA on how to deal with this issue</li>
780 </ul>
781 <h4><a name="sec5_3_7"/>5.3.7 Services</h4>
782 <p>
783 The goal of the SimDB specification is to define a protocol for querying interesting simulations
784 and related SimDB/Resource-s.
785 Once these have been identified the user should be able to access these simulations.
786 We assume that web services are the means to do so, and allow publishers to indicate such
787 web services as are available for a given Experiment. We assume for now that we know little of the
788 web service beyond some generic types: <i>download, cut-out, extraction, projection, custom</i>.
789 The SimDAP specification is being developed to address those aspects in detail.
790 We assume that there will be a base-URL implementing some standard DAL (VOSI?) like services
791 and leave it up to SimDB-client implementations to interact with these services in standard manners.
792 Only custom services can be directly accessed, and for now many services will necessarily be custom.
793 </p>
794 <ul class="issue">
795 <li>How do we get the complete list of service types? Predefined (as enumeration) in model?
796 </li>
797 </ul>
798 <h2><a name="sec6"/>6 Physical models</h2>
799 <p>
800 Here we describe how we create <i>physical models</i> out of the logical model.
801 A <i>physical model</i> is (see <em class="todo">@@TODO reference to some standard reference on data modelling@@</em>)
802 a representation of the logical model that is adapted to a particular software environment.
803 We present physical representations for the following contexts:
804 <ul>
805 <li>XML: we present an <a href="http://www.w3.org/XML/Schema">XML schema</a> defining valid XML documents</li>
806 <li>Relational databases: we derive a relational database schema for storing instaces of the model.</li>
807 <li>Java: we present Java classes representing the data model in a JVM. These classes are annotated with
808 <a href="http://java.sun.com/javaee/technologies/persistence.jsp">Java Persistence API (JPA)</a> and
809 <a href="http://java.sun.com/developer/technicalArticles/WebServices/jaxb/">Java Architecture for XML Binding (JAXB)</a>
810 annotations to enable easy transformations to the XML and relational contexts.</li>
811 <li>TAP: we present a representation of the model in a manner that hopefully has some similarity to the way
812 TAP will mandate meta-data about ADQL-queriable databases must be returned.</li>
813 <li>UTYPE: we present for the simple, i.e. non-structured elements in the SimDB/DM serialisations taht should
814 resemble UTYPE-s. These can be used when representing (parts of) the model in VOTable.</li>
815 <li>HTML: we present a representation of the model as a web browser (and human) readable HTML document.
816 This contains all details of the model in human readable form.</li>
817 </ul>
818 We have <em>completely automated</em> the derivation of these representations from the logical model using transformation
819 rules implemented in XSLT.
820 Our XSLT pipeline is described in more detail in <a href="#appB">Appendix B</a>.
821 annotations, provides simple means to store contents of SimDB/XML documents in
822 a SimDB relational database and retrieve them from there again.
823 </p>
824 <ul class="issue">
825 <li>Discuss adoption of this approach with DM WG</li>
826 <li>Ultimately we believe that defining these mappings is the realm of the DM WG,
827 which might come up with a kind of meta-specification.</li>
828 </ul>
829 <h3><a name="sec6_1"/>6.1 Identity and Referencing</h3>
830 <p>
831 The main elements in our data model are the object types, these embody the core concepts that we model.
832 In our approach we follow standard Object-Oriented design approaches
833 (see [<a href="r_DMApproaches">10</a>]) where object types are assumed to have an explicit <i>identity</i>.
834 Two objects (i.e instances of an object type) can have the same values for all fields, but if their identity is
835 not the same they are not the same object. Objects can be referenced by stating their identity (in whatever form this comes).
836 In contrast to this, <i>value types</i> are assumed to be identical if their value (or values, in the case of structured
837 value types) is the same.
838 In our UML model we have not defined an explicit <i>identifier</i> attribute on each object type to represent its identity,
839 its existence is assumed. There are some identifier-like attributes, but those refer to an identity the object has in abnother context,
840 generally the one of the publisher or creator of the object.
841 In most of the physical models we need to be able to represent this object identity explicitly however.
842 </p>
843 <p>
844 Related to this issue is that we need to be able to represent <a href="#uml_reference">reference</a>
845 relations between different objects.
846 Most contexts provide a natural mapping for references. For example relational databases have the concept of foreign keys,
847 XML documents allow references using ID/IDREF and other mechanisms for references to entities in the same document,
848 Java uses pointers (implicitly) to objects in the same virtual machine.
849 Problems arise when we need to leave the local contexts: references to resources not in the current database,
850 or in another XML document.
851 </p>
852 <p>
853 It is easy to imagine cases where this may occur. For example when registering a simulation run with the
854 open source Gadget [<a href="#r_Gadget">12</a>] simulation code, one needs to have a reference to the corresponding Gadget SimDB/Simulator.
855 Unless one registers the experiment in the same SimDB where Gadget is registered, one needs to use a reference
856 across SimDB-s. One obvious way is to map all references to globally unique identifiers,
857 possibly using URIs or IVOA Identifiers [<a href="#r_IVOAIdentifier">11</a>].
858 The size of such URI-s makes this a rather expensive storage mechanism for use in a relational database,
859 certainly compared to simple integer (or bigint) columns.
860 </p>
861 This issue is not yet resolved satisfactory. The following possible approaches offer themselves and need discussions:
862 <ul class="issue">
863 <li>Never allow references to objects not in the same SimDB.
864 This may require mirroring of resources not currently in the SimDB.
865 Registries have experience with similar such mechanisms, though likely for different reasons.</li>
866 <li>Use complete URIs for all references, and allow references to objects not in the same SimDB.
867 This makes it impossible to have foreign keys on these references, as the referred to object may not exist.
868 It may be relatively expensive, though this may be reduced with an extra level of indirection.
869 If referenced SimDBs are themselves registered in each SimDB, they are themselves assigned the possibly smaller
870 local ID (an integer or bigint). A reference need then not require more than two IDs, possibly one if a standardised
871 mapping is used.</li>
872 <li>Use IDs adjusted to the specific. for example use full URIs in XML documents, but resolve these to
873 smaller representations inside the database.</li>
874 </ul>
875 <em class="todo">@@ TODO this needs rewriting, too much stream of consciousness @@</em>
876
877 <h3><a name="sec6_2"/>6.2 RDBM Schema</h3>
878 The public schema, i.e. the view the outside world has of a SimDB, is a relational schema.
879 This will be formally defined using VOTables containing the appropriate TABLE definitions.
880 Our Object-Relaitonal mappingprescritpion contains the following elements:
881 <ul>
882 <li>object types are mapped to tables, one table per object type</li>
883 <li>Inheritance hierarchies: JOINED strategy as defined in JPA, i.e. each table only has columns for the attributes and references defined on the corresponding type.
884 Also an ID column that is a PK and also a FK to the ID of the base class' table. Possibly a container column (see below)</li>
885 <li>Primary key column: <tt>ID NUMERIC(18)</tt></li>
886 <li>Foreign key to container: <tt>containerId</tt><br/>plus foreign key and index declaration</li>
887 <li>References: &lt;referenceName&gt;Id<br/>plus foreign key and index declaration.</li>
888 <li>Using topological sort of object types based on (extends|container|reference) relations we generated
889 create table statements and ther indexes and foreign keys in blocks. drop table statements in opposite order.</li>
890 <li>For each class we create a view named "v_&lt;class name&gt;<br/>returns all columns for that class; uses join to base class's view.</li>
891 <li>generate a discriminator column on table for root in inheritance hierarchy, stores name of class (must be unique in inheritance hierarchy!)</li>
892 <li>attributes mapped to single column if their type is simple (i.e. primitive, or enumeration)</li>
893 <li>if attribute's type is dataType mapped to as many columns as the dataType has attributes,
894 with column names the name of the dataType's attributes, prefixed by &lt;attribute-name&gt;_</li>
895 <li>For PK columns we use the
896 </ul>
897 <em class="todo">@@ TODO add links to actual generated schemas @@</em>
898
899 <h3><a name="sec6_3"/>6.3 XML Schema</h3>
900 <p>
901 The DM WG has mandated (IVOA interoperability meeting, Cambridge, UK, May 2003) that each data model should come with
902 an XML schema that represents valid XML serialisations of the data model.
903 We foresee that this representation can be used to communicate instances of SimDB/Resource-s as XML documents.
904 Such communication can be for registering new SimDB/Resources in a SimDB, or
905 used in message to communicate instances of the SimDB Resource type.
906 Here we shortly describe some of the rules for deriving an XML schema from our logical model.
907 </p>
908 <ul>
909 <li>object and data types are mapped to comlexType. Object types inherit from a base class taht defines
910 features dealing with identity.</li>
911 <li> primitiveType-s are mapped to appropriate simpleType-s</li>
912 <li>enumerations are mapped to simpleType-s which are a restriction of xsd:string and have
913 an enumeration element for each literal.</li>
914 <li>packages are mapped to namespaces and eahc package has its own file, with dependencies translated
915 to schema imports.</li>
916 <li>A root element is generated for each concrete (i.e. non-abstract) root (i.e. not contained in other types)
917 object type.</li>
918 <li>attributes are mapped to elements (<i>not</i> attributes!) of the appropriate type.</li>
919 <li>collections are mapped to elements of the appropriate type, contained within the complexTYpe of the containing complexType</li>
920 <li>references are mapped to a elements of a special purpose base complexType, Reference.
921 The precise definition of this type is postponed until the issues about identity and referencing is resolved.
922 For now it has multiple sub-elements reflecting the different possible ways to refer to other elements.</li>
923 </ul>
924 <em class="todo">@@ TODO add links to actual generated schemas @@</em>
925
926
927 <h3><a name="sec6_4"/>6.4 UTYPE-s</h3>
928 <p>
929 It is generally the case that contents of databases may be represented in ways that do not
930 conform to one of the standard serialisations. Nothing prevents services to be developed on
931 top of SimDB that represent SimDB/Resource-s or even fragments of these in another form.
932 The standard example would be to have VOTables storing the results of a generic ADQL query of the SimDB/RDB representation.
933 VOTable first introduced the option to have a UTYPE attribute in FIELD definition tags store
934 a pointer to an element in a data model that the column represents.
935 </p>
936 <p>
937 The <a href="#r_SpectrumDatamodel">Spectrum data model</a> was the first to add explicit
938 UTYPE-s for each of the attributes in its model and the <a href="#r_CharacterisationDM">Characterisaiton data model</a>
939 has followed that example. As long as the precise usage and relation of the syntax of the underlying data model is
940 is not defined, we will follow these examples by assigning UTYPE-s explicitly to all elements in the model.
941 However, we will follow a fixed set of rules to makes this assignment and implement these in XSLT.
942 If a similar approach is at some time accepted within the IVOA, possibly in an alternative form, it will be straightforward
943 to adjust our definitions. The important point we want to make is that it is possible to simply define rules that then will
944 automatically produce the UTYPE-s for a given data model, i.e. the only discussion that is required is on the rules for doing so.
945 </p>
946 <p>
947 Our assumption is that the UTYPE should be able to uniquely represent any element in the data model, and in a manner
948 that is also easily interpreted. For now we assume that we need to point to those elements
949 that can be stored in a column in a VOTable, i.e. for now we are looking for "simple" elements.
950 We can use our relational mapping to identify all these features, they are
951 <ul>
952 <li> attributes (paying attention to attributes with non simple data types)</li>
953 <li> references (an identifier </li>
954 identifying the referenced object) and
955 <li>collections (through a pointer to the containing, parent object). </li>
956 </ul>
957 VOTable also allows arrays to be stored in single columns, so a collection can be stored as an array of identifiers of
958 child objects. There are some other features that are not explicitly modelled, but are implied.
959 Examples are the identifier (ID) assigned to all objects and the name of the object type of an object.
960 </p>
961 <p>
962 Of course we could give each of the elements a uniquely generated identifier, but we assume that UTYPE-s should hold
963 semantic information, otherwise we could use the XMI-ids generated by the UML modelling tool.
964 To identify any of these elements uniquely within the context of the IVOA,
965 we then need the following components:
966 <ul>
967 <li>name of element (possibly a path expression for structured attributes leading to a "leaf attribute")</li>
968 <li>name of containing object type</li>
969 <li>a path expression for the package(s) containing the object type</li>
970 <li>unique identifier of the model, possibly its name if that is to be unique in the IVOA DM efforts</li>
971 <li>some indication of the context, unless this can be implicit.</li>
972 </ul>
973 NB this assumes that we do not have a uniqueness rule on the names of object types within a model, something we do actually
974 assume in the mapping of SimDB/RDB above. In that case we could leave out the package path.
975 </p>
976 <p>
977 One could argue one could also give nice, unique names to each of the elements, but to find out what the actual element in
978 the model and in other representations one would still need to perform a look up. Such a uniqe name would likely include some of
979 the elements above anyhow. So we believe it would be a waste of efforts to do so and instead propose a simple convention
980 for deriving the UTYPE-s form the model based on this hiherarchy.
981 We have done so using these rules (in BNF-like notation)
982 <dl>
983 <dt>attribute</dt>
984 <dd>
985 <pre>
986 &lt;model-name&gt; ":" &lt;package-name&gt;[ "/" &lt;package-name&gt;]* "/" &lt;objecttype-name&gt; "." &lt;attribute-name&gt; [ "." &lt;attribute-name&gt;]*
987 </pre>
988 </dd>
989 <dt>reference</dt>
990 <dd>
991 <pre>
992 &lt;model-name&gt; ":" &lt;package-name&gt;[ "/" &lt;package-name&gt;]* "/" &lt;objecttype-name&gt; "." &lt;reference-name&gt;
993 </pre>
994 </dd>
995 <dt>collection (as array of p0inters to child objects)</dt>
996 <dd>
997 <pre>
998 &lt;model-name&gt; ":" &lt;package-name&gt;[ "/" &lt;package-name&gt;]* "/" &lt;objecttype-name&gt; "." &lt;collection-name&gt;
999 </pre>
1000 </dd>
1001 <dt>container</dt>
1002 <dd>
1003 <pre>
1004 &lt;model-name&gt; ":" &lt;package-name&gt;[ "/" &lt;package-name&gt;]* "/" &lt;objecttype-name&gt; "." "CONTAINER";
1005 </pre>
1006 </dd>
1007 <dt>ID</dt>
1008 <dd>
1009 <pre>
1010 &lt;model-name&gt; ":" &lt;package-name&gt;[ "/" &lt;package-name&gt;]* "/" &lt;objecttype-name&gt; "." "ID";
1011 </pre>
1012 </dd>
1013 <dt>object type name</dt>
1014 <dd>
1015 <pre>
1016 &lt;model-name&gt; ":" &lt;package-name&gt;[ "/" &lt;package-name&gt;]* "/" &lt;objecttype-name&gt; "." "DTYPE";
1017 </pre>
1018 </dd>
1019
1020 </dl>
1021 The HTML documentation generated from the logical model contains UTYPE-s for these features, generated according to these rules.
1022 It will be obvious how to accommodate changes in the precise UTYPE specification, <em>as long as similar rules are upheld</em>.
1023 </p>
1024 <em class="todo">@@ TODO add links to actual generated schemas @@</em>
1025
1026 <h3><a name="sec6_5"/>6.5 Java/JPA+JAXB (non normative)</h3>
1027
1028 <h2><a name="sec7"/>7 Query Protocols</h2>
1029 <p>
1030 The previous chapter has defined a number of physical representations of the logical simulation data model.
1031 Using these we can implement a database that can store instances of SimDB/Resources.
1032 This could be done using an XML database, or using a relational database management system such as
1033 Postgres, MySQL or any of the commercial versions. The data model is rather complex,
1034 and more hierarchical than most other data models so far defined in the IVOA.
1035 Querying such a data model requires a rich query language and we propose to use
1036 ADQL working on the relational representation. ADQL produces tabular results, whose structure is completely
1037 governed by the query itself. We also assume it possible, once appropriate information is available, to
1038 retrieve complete SimDB/Resource-s as XML documents and propose a simple REST-like query interface for that.
1039 Such an XML based interface will likely also be used to upload new resources to SimDB implementations taht support
1040 that functionality.
1041 </p>
1042 <h3><a name="sec7_1"/>7.1 ADQL + TAP</h3>
1043 <p>
1044 We expect no problems in formulating ADQL queries based on the relational representation of the data model
1045 described in the previous chapter. We need to require an appropriate protocol for sending these queries to
1046 a SimDB service though. In DAL work has started on the Table Access Protocol (TAP) and clearly some version
1047 of that seems to be applicable to our situation. However there are some simplifying features.
1048 Foremost is that we pre-define the relational schema, so a generic TAP "getMetadata" service seems not necessary.
1049 There are likely going to be other standard DAL service features that we need to support (getCapabilities?),
1050 but as meta data databases are expected to be relatively small we may again not require the full richness of
1051 asynchronous querying, staging, VOSpace and what not.
1052 </p>
1053 Issues that need discussion:
1054 <ul class="issue">
1055 <li>(How) does TAP deal with units?</li>
1056 <li>In TAP, does a table column containing values always have a single UCD and a single Unit?</li>
1057 <li>Is TAP suited for this kind of meta data databases?</li>
1058 </ul>
1059
1060 <h3><a name="sec7_2"/>7.2 REST</h3>
1061 <p>
1062 Under this heading we mean a protocol whereby data products can be retrieved through
1063 HTTP GET requests. Possibly also they can be POST-ed, or PUT.
1064 This needs to be discussed further, but maybe can be punted until a future release.
1065 The GET will always only be able to get a complete SimDB resource, serialised to SimDB/XML,
1066 similar to the IVOA Resource Registry interface <em class="todo">@@ TODO is this actually a correct statement?@@</em>.
1067 </p>
1068
1069
1070
1071 <h2><a name="sec8"/>8 Next Steps</h2>
1072 <h3><a name="sec8_1"/>8.1 Reference implementations</h3>
1073 <h4><a name="sec8_1_1"/>8.1.1 France</h4>
1074 <em class="todo">@@ TODO Laurent @@</em>
1075 <h4><a name="sec8_1_2"/>8.1.2 Germany</h4>
1076 <em class="todo">@@ TODO Gerard @@</em>
1077 <h4><a name="sec8_1_3"/>8.1.3 Italy</h4>
1078 <em class="todo">@@ TODO Patrizia @@</em>
1079 <h4><a name="sec8_1_4"/>8.1.4 USA</h4>
1080 <em class="todo">@@ TODO Rick @@</em>
1081 s
1082 <h3><a name="sec8_2"/>8.2 Generating SimDB/XML documents from simulation pipe lines</h3>
1083 <p>
1084 Assigning meta-data to describe simulations etc is quite a lot of work if this is to be done aftre the fact.
1085 It seems more fruitful to see if simulation codes could make the production of the appropriate
1086 documents part of their pipe-line. It is our goal to contact the writers of some of the major
1087 simulation packages and see whether they are willing to do so.
1088 First contacts with Volker Springel (Gadget), the group in San Diego (Enzo) give us hope that this could
1089 be achieved. The TIG should see it as its task to contact more authors of such codes and promote this idea further.
1090 </p>
1091 <h3><a name="sec8_3"/>8.3 SimDAP services</h3>
1092 <p>
1093 Together with SimDB implementations we need to urge scientists to develop online services for accessing
1094 their published simulations. Until the SimDAP specification is further developed these can be custom services,
1095 but it is important that services are available asap. This outreach is a task for the TIG.
1096 </p>
1097 <h2><a name="appA"/>Appendix A: Data modelling specifics</h2>
1098 Here we describe various aspects of UML modelling as we applied it to the current
1099 problem area.
1100 <p>
1101 UML allows communities to create a domain specific modelling language through its Profiling capabilities
1102 <em class="todo">@@ TODO is this the proper term ?@@</em>.
1103
1104 We have an initial implementation of a UML profile as created by MagicDraw available under
1105 <a href="http://volute.googlecode.com/svn/trunk/projects/theory/snapdm/input/IVOA%20UML%20Profile%20v-2.xml">this link</a>.
1106 Here we list the main elements and give a a short motivation for their inclusion in the model/.
1107 It is our opinion that the DM working group should be ultimately responsible for a profile such as this,
1108 defining a domain specific language for all IVOA data modelling efforts.
1109 </p>
1110 <p>
1111 As first step in our simulation pipeline we generate an XML document that represents the data model in a form
1112 that is more easily interpreted, both by human readers and by XSLT scripts, than the XMI representation.
1113 This document itself is structured according to an XML schema that
1114 represents the UML profile rather directly and that we here shortly describe.
1115 </p>
1116 This schema is located in
1117 <a href="http://volute.googlecode.com//svn/trunk/projects/theory/snapdm/input/intermediateModel.xsd">
1118 http://volute.googlecode.com//svn/trunk/projects/theory/snapdm/input/intermediateModel.xsd</a>.
1119
1120
1121 We introduce our own XML format, defined by the XML schema in
1122 <a href="http://volute.googlecode.com/svn/trunk/projects/theory/snapdm/res/intermediateModel.xsd">intermediateModel.xsd</a>,
1123 for representing the logical model. For the time being we call this the <i>intermediate representation</i>.
1124 The first step in the generation pipeline is a translation of the XMI to an XML document following this format.
1125 This transformation is implemented in the
1126 <a href="http://volute.googlecode.com/svn/trunk/projects/theory/snapdm/res/xmi2intermediate.xsl">xmi2intermediate.xsl</a>
1127 XSLT script. The latest version of the intermediate representation for the SimDB data model can be found in
1128 <a href="http://volute.googlecode.com/svn/trunk/projects/theory/snapdm/output/SNAP_Simulation_DM_INTERMEDIATE.xml">this location</a>.
1129 All other generation scripts work on this intermediate representation, not on the XMI document.
1130 Variations in tool-generated XMI, or different versions of XMI can now be supported by an appropriately adjusted
1131 XSLT script.
1132 One reasons why this may be useful is that are different tools may produce different versions or different
1133 dialects of XMI. Another reason for this representation is that XMI is a rather complex representation of a UML
1134 model. Since we are using a rather restricted <a href="#profile">profile</a> we do not need this generality, and
1135 this allows us to represent the model using XML documents that are much easier to handle with XSLT.
1136
1137
1138 <p>
1139 We illustrate out UML profile using an example data model
1140 derived form the SimDB/DM, shown in the following diagram:<br/>
1141 <img src="img/example.jpg"/>
1142 <br/>
1143 We now describe the individual elements.
1144 some of these are standard, some of these are domain specific extensions following
1145 standard UML profile <i>stereotype</i> extension elements and associated <i>tag definition</i>.
1146
1147 <dl>
1148 <dt><a name="uml_model"/>Model (no visual counterpart)<dt>
1149 <dd>
1150 <ul>
1151 <li> &lt;&lt;model&gt;&gt; </li>
1152 <ul>
1153 <li>TagDefinition: author</li>
1154 <li>TagDefinition: title</li>
1155 </ul>
1156 </ul>
1157 </dd>
1158
1159 <dt> Package <br/><img src="img/package.jpg" /></dt>
1160 <dd>
1161 <ul>
1162 <li>package containment</li>
1163 <li>package dependency</li>
1164 </ul>
1165 </dd>
1166 <dt> Class <br/><img src="img/class.jpg" /></dt>
1167 <dd>
1168 <ul>
1169 <li>isAbstract<br/>
1170 Indicated by <i>italicised</i> name of the object. Implies that no instances can be made of the class,
1171 one needs sub classes for that.
1172 </li>
1173 </ul>
1174 </dd>
1175 <dt> DataType <br/><img src="img/datatype.jpg" /></dt>
1176 <dd></dd>
1177 <dt> Enumeration <br/><img src="img/enumeration.jpg" /></dt>
1178 <dd></dd>
1179 <dt> Property: attribute<br/><img src="img/attribute.jpg" /></dt>
1180 <dd>
1181 <ul><li>&lt;&lt;attribute&gt;&gt; </li>
1182 <ul>
1183 <li>TagDefinition: minLength<br/>
1184 </li>
1185 <li>TagDefinition: maxLength<br/>
1186 </li>
1187 </ul>
1188 <li> &lt;&lt;ontologyterm&gt;&gt; <br/>
1189 There are many instances in the data model where we need to describe elements of the
1190 SimDB/Resource-s explicitly, because we do not have implicit information based on the context.
1191 Examples are the various properties of object types, the target objects and processes etc.
1192 Apart from a name and a description we then frequently add
1193 an attribute which is supposed to "label" the element according to an assumed standard list of terms.
1194 We model this using the <pre>&lt;&lt;ontologyterm&gt;&gt;</pre> stereotype. Attributes with this stereotype
1195 are assumed to take their values form such a predefined "ontology".
1196 </li>
1197 <ul>
1198 <li>TagDefinition: ontologyURI<br/>
1199 A URL locating a standard (RDF|SKOS|OWL|???) document containing
1200 a list of terms from which the value for this attribute may be obtained.
1201 It is our opinion that the Semantics working group should be responsible for the
1202 definition of relevant ontologies (or semantic vocabularies, or thesauri, or ...)
1203 required for a given application domain, though the contents should be decided in
1204 cooperation with domain experts.
1205 </li>
1206 </ul>
1207 </ul>
1208 </dd>
1209 <dt>Inheritance
1210 <br/><img src="img/inheritance.jpg" /></dt>
1211 <dd>
1212 Indicates the typical <i>is a</i> relation between the sub-class and its base-class (the one pointed at).
1213 In this profile we do not support multiple inheritance. <em class="todo">@@ TODO explain? @@</em>.
1214 </dd>
1215 <dt>Binary association end: collection
1216 <br/><img src="img/collection.jpg" /></dt>
1217 <dd>
1218 This relation indicates a <i>composition</i> relation between one, parent object and 0 or more child objects.
1219 The life cycles of the child objects are governed by that of the parent.
1220 </dd>
1221 <dt><a name="uml_reference"/>Binary association end: reference
1222 <br/><img src="img/reference.jpg" /></dt>
1223 <dd>
1224 This is a relation that indicates a kind of <i>usage</i>, or <i>dependency</i> of one object on another.
1225 It is in general shared, i.e. many objects may reference a single other object. Accordingly the referenced
1226 object is independent of the "referee". In our model the cardinality can not be &gt; 1.
1227 </dd>
1228 <dt>Binary association end: subsets
1229 <br/><img src="img/subsets.jpg" /></dt>
1230 <dd>
1231 This indicates that a relation overrides a relation defined on a base class.
1232 It does so by specifying that the class at the end point of the relation should be a subclass of the
1233 class at the enpoint of the original, subsetted relation.
1234 </dd>
1235 </dl>
1236 </p>
1237
1238
1239 <h2><a name="appB"/>Appendix B: XSLT pipe line</h2>
1240 <em class="todo">@@ TODO Laurent @@</em>
1241
1242 <h2><a name="glossary"/>Glossary and Acronyms</h2>
1243 <dl>
1244 <dt><a name="g_SimDB">SimDB</a></dt>
1245 <dd>Acronym for <i>Simulation Database</i>, the standard that we propose to define in this Note.
1246 Implementations of SimDB offer a query interface for discovering simulations (and related entities)
1247 using ADQL, based on a prescribed (i.e.normative) relational data model and for describing simulations
1248 via XML documents following prescribed XML (i.e. normative) schema.</dd>
1249 <dt><a name="g_SimDAP"/>SimDAP</dt>
1250 <dd>Acronym for <i>Simulation Data Access Protocol</i>, a related standard to SimDB,
1251 which will define services for accessing simulations discovered using SimDB.</dd>
1252 <dt><a name="g_SimDB/DM"/>SimDB/DM</dt>
1253 <dd>The logical data model defining the structure of <a href="#g_SimDB">SimDB</a>.</dd>
1254 <dt><a name="g_SimDB/RDB"/>SimDB/RDB</dt>
1255 <dd>The representation of the SimDB/DM as a relational data base schema.
1256 This implies a parti</dd>
1257 <dt><a name="g_SimDB/RDB"/>SimDB/Views</dt>
1258 <dd>The representation of the SimDB/DM as a collection of database view definitions. Each View directly represents
1259 a complete DM class as a relational table, this in contrast to the underlying SimDB/RDB representation in tables,
1260 at least in the JOINED object-relational mapping strategy.</dd>
1261 <dt><a name="g_SimDB/XML"/>SimDB/XML</dt>
1262 <dd>The XML representation of the SimDB/DM</dd>
1263 <dt><a name="g_SimDB/Resource"/>SimDB/Resource</dt>
1264 <dd>A top-level data product stored in a SimDB.
1265 A SimDB/Resource can be described in a SimDB/XML document, but none of its constituents can.</dd>
1266 <dt><a name="g_SimDB/TAP"/>SimDB/TAP</dt>
1267 <dd>The TAP(-like) metadata representation of the SimDB/DM.
1268 This is currently (May 2008 <em class="todo">@@ TODO update once the TAP specification is out @@</em>
1269 a representation of the <a href="#g_SimDB/Views">SimDB/Views</a> as a VOTable document.
1270 </dd>
1271 </dl>
1272
1273 <h2><a name="references">References</a></h2>
1274
1275 <p><a name="r_UML">[1] ???, <i>UML standard</i></a>
1276 <br/><a href="http://">http://</a>
1277 </p>
1278 <p><a name="r_XMI">[2] ???, <i>XMI standard</i></a>
1279 <br/><a href="http://">http://</a>
1280 </p>
1281 <p><a name="r_AnalaysisPatterns">[3] Martin Fowler, <i>Analysis Patterns</i>, 1997, Addison Wesley.</a>
1282 <br/><a href="http://">http://</a>
1283 </p>
1284 <p><a name="r_TheoryinVO">[4] Lemson & Colberg, <i>Theory in the virtual observatory</i></a>
1285 <br/><a href="http://">http://</a>
1286 </p>
1287
1288 <p><a name="r_Characterisation">[5] ???, <i>Characterisation DM</i></a>
1289 <br/><a href="http://">http://</a>
1290 </p>
1291
1292 <p><a name="r_informatonIntegration">[6] <em class="todo">@@ TODO @@</em>references on global-as-view and information integration</a>
1293 <br/><a href="http://">http://</a>
1294 </p>
1295
1296 <p><a name="r_visivo">[7] <em class="todo">@@ TODO @@</em>reference to VisIVO</a>
1297 <br/><a href="http://">http://</a>
1298 </p>
1299
1300 <p><a name="r_SpectrumDatamodel">[8] <em class="todo">@@ TODO @@</em>reference to Spectrum data model</a>
1301 <br/><a href="http://">http://</a>
1302 </p>
1303
1304 <p><a name="r_Normalisation"/>[9], <i>Some links to pages on data model normalisation</i><br/>
1305 <a href="http://www.datamodel.org/NormalizationRules.html">http://www.datamodel.org/NormalizationRules.html</a><br/>
1306 <a href="http://en.wikipedia.org/wiki/Database_normalization">http://en.wikipedia.org/wiki/Database_normalization</a><br/>
1307 </p>
1308
1309 <p><a name="r_DMApproaches"/>[10], some data model references<br/>
1310 <a href="http://www.agiledata.org/essays/dataModeling101.html">http://www.agiledata.org/essays/dataModeling101.html</a><br/>
1311 Meyer, B. <i>Object Oriented Software Construction, 2<sup>nd</sup> edition, Prentice Hall, 1997</i><br/>
1312 On object identity: <a href="http://en.wikipedia.org/wiki/Identity_(object-oriented_programming)">http://en.wikipedia.org/wiki/Identity_(object-oriented_programming)</a><br/>
1313 </p>
1314
1315 <p><a name="r_IVOIdentifiers">[11] <em class="todo">@@ TODO @@</em>reference to IVOA Identifiers ...</a>
1316 <br/><a href="http://">http://</a>
1317 </p>
1318
1319 <p><a name="r_Gadget">[12] <em class="todo">@@ TODO @@</em>reference to Gadget ...</a>
1320 <br/><a href="http://">http://</a>
1321 </p>
1322
1323
1324 </body>
1325 </html>

msdemlei@ari.uni-heidelberg.de
ViewVC Help
Powered by ViewVC 1.1.26