DataLink issues

[François Bonnarel]

Hi Petr,
all
Le 27/09/2012 11:43, Petr Skoda a écrit :
>
> Hi Pat !
>
>> My sense from the splinter sessions in Urbana was that everyone 
>> agreed with
>> this idea: DataLink does not itself take much in terms of parameters 
>> *but* it
>> does have to be able to describe both links to static data products data
>> products and links to services that perform some processing such as 
>> you and MD
>> have described.
>
> Yes link to more specific services
>
>
>> In addition to "big data" that cannot be feasibly downloaded and 
>> cutouts in
>> regular data to make client work straightforward, the Theory people 
>> need this
>> as well since they access to "data products" -- maybe exclusively -- 
>> that
>> require cutouts of some form and often along custom axes.
>  I think its OK - DL will give you "link" to another service  and that 
> service will allow you to give specific parameters of given protocol
>
> So lets say you are browsink the reply of datalinks to given object - 
> you will see - ha there is a spectrum - so we can launch the 
> predefined tool (e.g. SPLAT) who understands SSA and has a fields to 
> enter parameters to make the cutout of given line.
>
> The key issue is the client - it would need to understand all possible 
> parameters of all protocols and build the web form - like query fields 
> on the fly .....
>
> I foresee two different modus  operandi  of astronomers:
>
> I am searching for a bunch of spectra of my objects having the halpha 
> line (or some x-ray line) and they will go directly with specific 
> request to the SSA
>
> Or they are interested what is available in VO about my objects?
> So use TAP to find all observations (images, spectra etc ......, raw 
> data only ...) and find - ha - they have raw data - I can run ds9 to 
> see them using DL - and they have as well reduced spectra - I can run 
> SPLAT - and they have images - so I will (automatically) launch 
> Aladin)...
>
> So it is the real discovery of VO, while the specific protocols will 
> provide the more advanced and focused results (including the 
> postprocessing).
>
> Concerning big data cube - we have discussed this with Felix Stoehr 
> and my solar colleagues due to ALMA.
>
> The only feasible solution is the discovery of datacube, getting it 
> whole and then apply various non-trivial algorithms to get images, 
> intensity maps, spectral lines etc .... and its very hard to foresee 
> the wishes of user. So The postprocessing like simple collapsing of 
> the cube in one dimension does not have a scientific sense without 
> more specific control which would require a lot of parameters and 
> visual interaction.
>
> it terms of DL - you would see there is an ALMA observation on this 
> spot in given time and you would get the link to the file. If someone 
> will do the postprocessing tou would see e.g. the spectra in integral 
> beam or list of spectra in different spots on the sky - and you could 
> run SPLAT to see the spectrum but its difficult to imagine to do the 
> processing on the fly for you - perhaps using UWS to run CASA remotely 
> .....
>
I think this is typically a question for data access and SIAV2 issues, 
not for dataLink in itself.
I think we are facing two very different use cases there.
               - What you were talking about with Felix is actual full 
cube processing. Of course it must be done with specific tools on your 
desktop
               - What SIAV2 accesData is dealing with is cube processing 
in the context of fine-tuned data discovery. Give excerpts or 
transformations of cubes to help the user to figure out what kind of 
dataset is on the other side of the interface.
> I am sorry to the radioastronomers for intruding their business ;-)
> And sorry to the Francois for extinguishing his enthusiasm about DL 
> capabilities -
The definition of DataLink i am supporting is very simple... This has 
not to be confused with the capabilities of the services which are 
linked to the dataset.
(eg SIA or SSA.)
> After sad experience with SSA potential and reality I am getting more 
> realistic ......
>
This is actually an SSA or SIA discussion I think
Cheers
François
> Petr
>
> *************************************************************************
> *  Petr Skoda                         Phone : +420-323-649201, ext. 361 *
> *  Stellar Department                         +420-323-620361           *
> *  Astronomical Institute AS CR       Fax   : +420-323-620250           *
> *  251 65 Ondrejov                    e-mail: skoda at sunstel.asu.cas.cz  *
> *  Czech Republic                                                       *
> *************************************************************************