Hi Jacek,
Independently if you are in a hurry or not, TGeant3, TGeant4, TFluka
is an excellent investment for the future. I really would like to encourage
you to continue the exercise in collaboration with Ivana and Isidro.
The new geometry modeller will fit nicely in this picture.
As I said in my previous mail, we will release the specs and also the first
version in a few weeks. Andrei and Michaela Gheata are currently doing
a great job in implementing the package. We strongly hope that a working
version will be part of the final release of ROOT 3.03 in June.
This package will:
- be an alternative to the mini geometry classes currently in g3d.
- provide detector geometry modelling capabilities supporting the full
Geant3 functionality and much more.
- provide the complete graphics 2-d and 3-D visualisation
- provide functions like
-where am I?
-distance to surface
Before releasing more info, we want to run benchmarks comparing the capabilities
and the performance to systems such as Geant3 and Geant4. Currently we can
model the full versions of the 4 LHC detectors as simulated with Geant3.
The visualisation/browsing in these 4 detectors is working. Also the logic
for "Where am I?" is working for the 4 detectors.
Once this package is running, it is our intention to collaborate to an effort
aiming at replacing the existing geometry packages by this new package.
This will take more time.
If people currently using Geant3 to simulate their detectors are interested
to see the performance of the new ROOT package, please contact us. We will let
you know how to proceed.
Rene Brun
Jacek M. Holeczek wrote:
>
> Dear Ivana, Rene and Others,
> My primary goal was to create a standalone "virtual" interface to TGeant3
> and TGeant4 using the existing code from AliSoft.
> I'm not an AliRoot developer, so I have no influence on how the software
> is organized nor maintained. Thus I needed to make sure that any future
> AliSoft version is to be easily imported by any user. Then my strategic
> decision was to extract all important parts AS THEY ARE, and for the rest
> a set of "dummy" classes should be provided. That means that the
> "TGeant3" and "TGeant4" subdirectories, for example, are expected to be
> taken AS THEY ARE, WITHOUT ANY MODIFICATIONS. I'm sorry that these
> "abstract" VirtualMonteCarlo interfaces contain Ali* specific files, but I
> cannot change it. I cannot also change the situation that different parts
> of AliSoft cross-reference each others' classes. This, in turn, makes the
> problem that one needs to provide a "minimal" set of classes/files which
> substitute real Ali* classes (these are the "dummy" classes like AliRun,
> AliModule, ...).
> I decided to offer it to the whole TUniverse, because maybe there are
> others who would like to use the results my work as it is now.
> I explicitly stated that "I did NOT test that this whole suite is
> usable", however everyone is free to try it, modify it, improve it, ... .
> Everyone is welcome ... (the grandma, too).
> Note also that my purpose was NOT to run aliroot/aligeant4, nor I intend
> to do it - my purpose was to get a STANDALONE "virtual" interface.
> Finally, the Rene's idea of a new geometry modeler sounds very well.
> What is the time plan, however ?
> In a "few weeks" there will be the first release. Then some time is needed
> to test it, fix bugs, ... . Then finally a possible approach to Geant3,
> Geant4, Fluka will be developed. This means ... one year from now ?
> I'm in a hurry. I need to provide a "glue" between geant and root NOW.
> I have exploited a couple of scenarios. Two most promising are :
> "extraction of TGeant3 and TGeant4 (and possibly TFlugg in the
> future) from the AliSoft and Isidro's G4UIRoot (no TGeant3, no Fluka).
> I don't know yet which of them I will choose, for the moment, however,
> G4UIRoot seems to be easier to work with ...
> Best regards,
> Jacek.
This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:50:46 MET