S3/SimDB

[Miguel Cerviño]

Dear all,

After reading Franck and Dave mails, I think that there are several  
points raised in the previous e-mails and I would like to comment them  
explicitly:

a) S3/SimDB have different objectives: I agree on that: S3 goal is to  
access simulations that are "assumed" you know that are there, like a  
Zsolar 1Gyr-old isochrone or a 1Mo evolutionary track (I do not know  
how many
track/isochrone developers really compute a 1Mo evolutionary track,  
but I assume that it is computed). SimDB in this case is to present  
simulations and people can "explore" them, it means: You do not know a  
priori what is there. In this case,

S3 covers the cases where "on the go" process work, like "take a set  
of theoretical spectra
for posterior fit with observational data", that can be used for work  
flow process, and SimDB looks to me
more related for discovered (and database-exploration). Or in other  
words;  SimDB would allow to explore the database  in a more flexible  
way, and in S3 (with its PARAM-QUERY based structure) looks more  
restrictive
for exploration and more direct for directly retrieve results (if the  
user/application knows a priori what they want, like atmosphere  
libraries/tracks/isochrone/synthesis models cases).


b) The need (or help) of tools to develop theoretical services (Volute  
for SimDB and S3Wizard for S3)


c) Data-models/description: SimDB includes a generic DM and S3 would  
need specific DM depending on the
service. In both cases, it is needed a scientist to describe the  
fields. Here there is a subtle problem, some times there is not a  
clear consensus about what is/should be the result of particular type  
of simulations (In my case, synthesis models compute a mean value of a  
distribution, but a mean does not define a distribution, so it is not  
enough, for others, photoionization only means emission-line spectrum  
and not ionization structure etc...)
So it is needed a lot of flexibility in DB description/definitions,  
since they are in the very base of research, and
they are not necessarily "fixed-well-defined" descriptions.


In the middle of all this, there are people waiting to "at least" have  
a recipes to put their models in the VO
(and VO people are waiting for them to develop analysis tools ). And  
we still have no idea/suggestion about
how to registry the services (last notice I remember was a parallel  
registry for SimDAP, but I not sure, and I prefer
that standard applications would browse any theoretical service. So,  
how to proceed?

cheers

	Miguel



P.D.: Franck, the spectral implementation of SSA for theoretical  
models is quite similar to any S3 implementation. Differences are that  
SSA assumes an observed spectrum as data model, with space-time  
coordinates even in theoretical spectra, and S3 does not assume a  
specific data-model.