tehoretical SEDs in VO applications

[Jesus Salgado]

Hi Miguel,

On Fri, 2006-05-05 at 15:37 +0200, Miguel Cerviño wrote:
> Thanks a lot Jesus,
> 
> I completely agree with you points, but
> only for "increasing" the discussion:
> 
> About metadata in the VO table, there is two particular issues
> specifically related with theory, and any analysis tool:
> Suppouse that you overplot real data and several theoretical models.
> And you obtain a good fit of a particular model. If the application  
> is not able to show the
> model parameters (that should be included somewhere in the VOTable  
> metadata), how
> can you recover them for, for example, a publication? I would agree  
> that an
> link to a generated page would be fine, but it would only works for  
> catalogs,
> and it would not work if the teoretical data is generated by the use  
> of diferent
> aplications (example a virtual telescope machine). Since model  
> parameters should
> be included in the VOtable, I think is more easy, and safer, that the  
> applications
> would be able to show the information that is yet included in the  
> VOTable
> they manage instead to refer to an external link. Additionally, an  
> external link
> will make not possible to work with downloaded data without internet  
> conexion....
> 

In this particular case, I would put the metadata to identify the
spectrum inside the spectrum itself, e.g., inside the header in case of
a fits or inside the votable spectrum in case of a votable spectrum. 
In this case, the file downloaded would be self-consistent and the model
provider will be sure that all the information to properly "understand"
the spectrum is there after saving the file. If I remember correctly,
all the applications have a save function for the file itself, so this
approach would work for all the VO applications. (in the case of VOSpec,
clicking twice on one spectrum, a window appears with a Save button)

Probably, there is room to more specific metadata in the VOTable
response (in particular publications links should be understood better
in the clients) but, except for the VOTable editors, many of the
metadata will be ignored by the VO clients. 
On the other hand, and probably because of the same reason, the VO
clients do not save the VOTable responses but only the data link files.

There is a problem added to save VOTable responses. For the client, the
result is a combination of several VOTables from different servers, so
the only two ways that I see to save the metadata is either saving every
single VOTable or to create a VOTable as a combination of several
heterogeneous VOTables (really a mess if we want to maintain this result
VOTable as something usable as input for future e.g. VOSpec sessions)

> (An specific comment about that is that the providers of synthesis  
> models
> would not be very happy and may be will not join actively the VO  
> iniciative if
> they do not obtain credits for the use of their models, and metadata  
> is the
> only place where such a credit can be written explicitelly, including  
> references to
> the models papers etc...)
> 
> And about TSAP, yes, I know (I had been also working) it ;)
> However, most of VOtools applications are not designed to
> support a TSAP-like protocol: Obtain a parameter VOTable and
> process it before access the data.... Maybe is time that other
> applications also will think about this kind of implementation, isn't  
> it?
> 
I agree with you as the implementation of TSAP for a VO SED client
application is quite easy and can be done for other applications.

During the presentation in Kyoto, it was agreed to include TSAP as a
particular case of SSAP. The only thing to be changed in the SSAP was to
demote the position compulsory input to optional, as "position" is
meaningless for most of the model servers. 
The call to the format=metadata was something standard to create
specific queries to servers and it should be already implemented in the
clients, but most of the VO applications create "blind" queries using
only the compulsory parameters (what it represents a problem for the
TSAP implementations)

In any case, the creation of on-the-fly query forms using the
format=metadata VOTables (or other more complex response in the future,
in case of evolution) should not be a major problem.


> cheers
> 
> 	miguel
> 
> 
Best Regards,
-- 
Jesus J. Salgado


ESA-VO team
e-mail: Jesus.Salgado at sciops.esa.int
Tel + 34 91 8131271

European Space Agency/European Space Astronomy Centre
VILLAFRANCA Satellites Tracking Station
P.O. Box 50727
E-28080 Villafranca del Castillo
MADRID - SPAIN