Session of the workshop on "Theory in the Virtual Observatory"

Introduction

In this session we have asked the invited speakers to give an overview of the Virtual Observatory. This includes an overview of the ongoing world-wide efforts of which the Euro-VO project is an example, of science done within the VObs context and an introduction of the ideas concerning the introduction of theory in these efforts.

Simulations

There are likely many ways of characterising simulations, something which will be discussed during the meeting. We have chosen to separate simulations in two classes, which we could define as large and small scale simulations, and we devote separate sessions on each. Such a demarcation will necessarily be somewhat ill defined "near the boundaries", but we feel we have contributions in each of these sessions that clearly belong there.

3+1D Simulations (Large)

In this session we want to address issues related to the incorporation of large scale simulations in the VObs. The type of simulations we are thinking of are N-body, AMR simulations modelling gravity, hydrodynamics, MHD etc. The subject of these simulations can range form cosmology and large scale structure of the universe down to planet formation. The main characterising feature of them is that a volume of 3D space is explicitly modeled and evolved forward in time.

The theory interest group of the IVOA is currently focusing on this type of simulations in a first effort to come up with standard protocols for describing, discovering and retrieving them. Their generally large size means that their retrieval will often not be feasible and ideas are being explored on special purpose online services that could be implemented by the publisher. Examples are services for cutting out sub volumes or for projecting or visualising the results.

We have invited speakers who are experts in producing and analysing this type of simulations, and have also been involved in attempts to publish the results to the larger community. We would like to hear from them about typical scientific projects and their experiences with usage of the results. Interesting subjects are possible standardisation of data formats, explicit use of meta-data descriptors and possible advanced services for accessing the results of their work. We would like to hear from them how they see this work could be extended into a general VObs setting.

Micro-Simulations (Small)

In this session we want to address issues related to the incorporation of small scale simulations in the virtual observatory. We are aiming at models of astrophysical processes that do not necessary involve space coordinates in the VObs. Examples are stellar atmosphere models (both in terms of structure of the atmosphere and emited fluxes), stellar evolutionary tracks and isochrones, synthesis models of stellar populations, photoionization models, chemical processes in interstellar clouds like cloud cooling and fragmentation, interstellar medium, global galaxy models and chemical evolution models.

An important feature distinguishing this class of simulations from the other class we discuss in the workshop is the size and type of products. In general the results of these simulations are much smaller and more easily handled, and because of this in general come in large grids, resulting from the study of an often high-dimensional parameter space Consequently this puts different constraints and requirements on their publication to the general community, and offers different opportunities. For example online simulations are more easily enabled.

We have invited speakers that produce this type of models, and speakers that have worked in the characterization of data retrieved from these high-dimensional parameter space. We would like hear from them about their scientific projects and experiences with usage of this kind of models. In particular, about the description of what is produced by the models and what is needed to describe the models result univocally, including explicit and implicit assumptions. Interesting subjects are possible standarisation of the data description, services for accessing these results and what requiriments they think that are needed to include this kind of data in a general VObs setting.

Interoperability

Interoperability means that multiple online resources (i.e. data and/or web-based applications/services) can be accessed and combined programatically. Interoperability is a main goal of most VObs efforts. To achieve it the IVOA is developing standards by which resources should be published and which allow one to write general purpose tools that can be agnostic of the implementation of these standards in the individual cases. We have devoted two sessions on this subject, one dealing with the interoperability between purely theoretical resources, the other with interoperability betwen observation and theory. In this workshop on "theory in the VObs" we leave out observation-only interoperability.

Theory-Theory Interoperability

In this session we want to explore how the Virtual Observatory (VObs) could support "interoperability" between theory resources. So far efforts at achieving interoperability in the VObs have been restricted to observational resources. An important reason for this is that scientific use cases for such interoperability are easy to imagine: when looking at the same part of the sky with different telescopes and in different wavelengths one may expect to obtain information about the same objects. If the corresponding data products are published online in a standardised fashion, it will be straightforward to write applications that can provide a multiwavelength view of these sources.

There are some key points which facilitate interoperability in the observational context. First is the assumption(fact) that we all see the same sky, with the same objects in it no matter which instrument we point at it. Second, the resulting data products are very similar, which is because the relevant observables are very limited and the same: space, time, wavelength and luminous intensity. Third is that we can build on existing standards for represent these data products, for example FITS for images. Important here is that FITS is not only a data format standard, but especially that it contains standards for describing the meta-data.

It is much less straightforward to imagine use cases for interoperability in a purely theoretical setting. Consequently it is more difficult to define appropriate standards to support interoperability of theoretical resources. The reasons can be seen from noticing that the points which facilitate this effort in the observational context do in general not hold for simulations: There is no common world to observe (simulate), the objects under investigation come with a number of observables only limited by the simulator's imagination and capabilities of implementation, and standardisation of data products is consequently less well evolved, if existent at all.

We felt therefore that when addressing the issue of interoperability for theory, it was first and foremost important to discuss possible use cases. To this end we have invited speakers whose work has consisted of bringing together results from various sources, and of combining these with their own to add value and/or produce new results.

We want to hear from them about their projects, about the type of scientific goals they were persuing and about the challenges they needed to overcome to work with the results of other scientists. We would also like to hear their thoughts on whether similar projects could benefit from the VO approach towards interoperability. Do they think that they might have benefitted from standardisation of protocols and data formats. Could they imagine that their work might have been performed using some kind of online workflow processor, or that their work might even made into an online service in some such manner? Or if not, what do they think are the main obstacles. Is it data size in combination with limited network bandwidth, is it even unfeasible to define standards for web service access, messaging and data formats that would enable this?

Theory-Observational Interface

In this session we will discuss how observers can make use of theory resources (simulation data, online applications, ...) and what role the VObs might play in facilitating this. This is one of the potentially very promising results of theory VObs projects, that it might narrow if not bridge the gap between observers and theorists. How to do so, where each discipline has its own and generally very different technical requirements, is still an open question, though some proposals have been made and efforts are under way to implement them. On example is the concept of the "virtual telescope". Theorists might attempt to provide services that allow users to observe their simulations and produce synthetic images or spectra that can be directly compared to their observational counterparts. Other examples are services that use theoretical models to interpret observations, such as tools to "invert" galaxy spectra to predict their stellar populations.

To evaluate these and other ideas we have invited a number of speakers to present work they did that was similar in spirit to typical use cases. We would like to hear about the reasons for using simulation data, which type of simulation data were used and how. We would like to hear about possible problems they encounter when trying to use, or maybe find relevant relevant results. We would like to hear some ideas how this interface between observers and theorists could be improved.

Computational infrastructure

In this session we will discuss what computational resources are available and/or needed to assist the introduction and use of theory resources in the VObs. A prime goal of the VObs is to move beyond data downloads as the prime way of publishing one's results. Instead more sophisticated services are envisioned and these consequently need more sophisticated techniques to implement. Many of these techniques are relatively unfamiliar to the scientists and it is a goal of many VO projects to shield them as much as possible from such implementation details.

That said, there are developments in this area that are of interest to the scientist. In this session we will from scientists and developers who participate in such efforts. Interesting questions to discuss are what benefits the scientists get from these, what is their learning curve, are they actually useful, or what requirements do scientists have for the VO projects to support them in these effort, or devise tools to shield them from this.

Discussions

-- GerardLemson - 18 Mar 2008