Re: Geant4

[David Lawrence <davidl@jlab.org>]

Hi Richard,

	I certainly agree that it would be foolish for anyone nowadays to
write their own XML parser. Presumably, they would use something like
xerces to do it for them in a way similar to HDGeant (though they might
consider integrating xerces with the Geant4 simulation directly so
geometry changes wouldn't require re-compilation).

	I would strongly second your suggestion that we make a standard
output format a requirement. I have not got as far as looking at hddm yet,
but that appears to be the choice at present. It appears clear from
conversations I've had recently that an standard analysis framework of
some type is one of the next things that needs to be developed. It would
make zero sense, IMHO, to do start diversifying output formats as opposed
to converging them.

-Dave

------------------------------------------------------------------------
  David Lawrence Ph.D.
  Staff Scientist                 Office: (757)269-5567   [[[  [   [ [       
  Jefferson Lab                   Pager:  (757)584-5567   [  [ [ [ [ [   
  http://www.jlab.org/~davidl     davidl@jlab.org         [[[  [[ [[ [[[
------------------------------------------------------------------------


On Thu, 29 Apr 2004, Richard Jones wrote:

> David,
> 
> If the goal is to make a plug-replaceable version of the present Geant3-based HDGeant in Geant4 then I have no objections to outsourcing the project (might even win us some points with the Administration in Washington).  But we don't want to end up with two different structures for the hits information.  Just as important as interfacing with hdds geometry is to get the hits structure information from hddm.  The hddm library is used to give the simulation access to generator information and to let the simulation push the results in structures defined in a xml-based data model that is shared by all of the simulation and analysis packages.
> 
> HDFast is not going to be a suitable model for this because it does not have a fully developed hits package like Geant does, and no one has yet thought it important enough to manually integrate hddm into HDFast.  Eventually we expect to rely on HDGeant and its successors for our simulation needs.
> 
> Question: Given the incomplete documentation and evolving nature of these interfaces, is it realistic to expect something useful to come out of this?  It relies on a person figuring out how these pieces all interact and understanding the idea of encapsulation.  For example, if we look into the new Geant4 code and see that he is opening the xml database file directly and hunting for some keyword on some given line number, that is a no-no.  I guess it depends on you knowing how good the student is!
> 
> Richard Jones
> 
> 
> 
> > Hi All,
> >
> > 	I just signed up for this list today and this is my first time
> > posting to it. I am currently familiarizing myself with the two Hall-D
> > Monte Carlo programs. I'm sure as I get deeper into it, I'll have other
> > questions/comments to post here.
> >
> > 	This message, though, is to solicite feedback from the Hall-D MC
> > group on GEANT4 development. I have a colleauge in Thailand (Burin
> > Asavaphibop which some of you may know) at Chulalongkorn University.
> > He has a student who is looking for a project. I was thinking he could
> > maybe work on a GEANT4 simulation of Hall-D. It would, of course, use
> > the HDDS files like HDFast and HDGeant. I was wondering if anyone has
> > put any serious effort towards this yet, or has a concern over them
> > working on it.
> >
> > 	All comments are welcome.
> >
> > -Dave
> 
> 
>