|
|
Theory in the Virtual Observatory
One of the core objective of the GAVO project is the publication of simulation results in a VO compatible manner. What it means to "publish simulations" in a VO context is not obvious, but one of the main requirements of the Theoretical VO is to be a bridge between observational and simulation archives by enabling access to either in a uniform manner. This is often referred to as the Theory-Observational Interface.The comparison to the publication of observational archives, for which a lot of work has been done already is not straightforward. One reason is that most current VO standards for accessing observational archives such as the Simple Cone Search (SCS) and the Simple Image Access Protocol (SIAP) are based on positional searches. The reason that positional queries play a key role in the "observational VO" is that most correlations between observations in different wavelengths regimes start with matching by position. An important task people hope to achieve is the identification of sources in different catalogues as being different observations of the same object.
Positional queries of this type seem hardly relevant to querying simulations. Only when different simulations are run from identical initial conditions or when simulations are constrained to reproduce a particular observation can positional queries play a role. Such simulations are not the norm, though an interesting example of the latter is formed by the Constrained Realizations available in the simulations archive at MPA.
The theory-observational interface must therefore be based on correlations of a different nature. They will most likely be based on common identification of the class of an object and its physical parameters. An example of this might be the mass and temperature of a cluster simulated using a hydro-code, or the relative orientation of a pair of galaxies involved in a merger. Another possibility for both simulation-simulation and simulation-observation correlations is offered by statistical quantities. Examples of these are luminosity functions in a cosmological simulation.
Data modeling for simulations
During the IVOA Interoperability meeting in Cambridge, UK, May 2003, it was decided that the data modeling of simulations deserved special attention. A member of the GAVO team was asked to lead this effort. So far, this effort has not been high on the list of priorities, both in GAVO and the IVOA, but recently progress has been made in the context of an application aimed at publishing data obtained from large scale cosmological simulations run at the MPA. To store the meta-data describing these results we will use a relational database. The (relational) data model used for that database is obtained by an object-relational mapping from the VO data model. The actual model that we use consists of two parts. One is an actual model for the data that will be stored. Apart from the data itself we will also store the meta-data describing the data products, how they came to be etc. That part of the model is what can hopefully be generalized to the larger VO community, whereas the details of the data themselves are likely to remain specialized to the specific archive for quite some time. This model is a slightly elaborated subset of the model presented in Lemson et al. 2004. This prototype application will be the first time this model will be implemented in source code and database tables. A description of the details of the mapping to the relational model and the techniques of creating representations of the model in code are described in the webapplication itself, http://www.g-vo/org/mpasims.Theory in the VO whitepaper
During ADASS XIII in Strasbourg, in 2003, several VO members discussed issues related to theory in the VO. It was felt that most VO efforts so far had concentrated on observational data products and that there were various issues that needed to be addressed possibly again when also theoretical data products are to be supported. It was decided that a whitepaper would be written listing these issues and proposing actions for taking these into account. This whitepaper was then to be circulated and proposed to the IVOA executive committee for further action.The result of this is the whitepaper titled Theory in the VO. Some of the issues addressed in the whitepaper are the following:
- Why publish simulations in the VO?
- How are data published in the VO?
- Why do we need a common data model to describe data?
- What kinds of theoretical data products can be published?
- What kinds of theory related services should be offered?
- What is the theory/observational interface?
- How do we ensure IVOA compliance?
- Why does theory require special attention in the IVOA?
- What concrete tasks can we define?
