halld-mc: minutes-2000-02-25

[Randy MacLeod <rwm@jlab.org>]

On Friday Feb 25 we had a teleconference meeting.

Attendees:
  Randy MacLeod, Elliott Wolin, Paul Eugenio, John Cummings, 
Richard Jones, Dennis Weygand, Carlos Salgado

Guest:  Mikhail  Kossov.

Agenda:
 1) Plan for the Hall-D Monte Carlo.
 2) Status report on Geant 4 - Mikhail Kossov.

Overview:
=========

Discussion of Geant 4. We concluded that is unlikely that we
can use G4 for simulations contributing to the next design report.

We will perpare a plan for simulations ASAP.



MINUTES:
========

(These are from my notes and 1 week old memories ;-) )

We outlined the plans for HD MC for Mikhail.  We need to optimize the
Detector design and performance for the next version of the Hall-D design report
due in August, 2000. 

G4 is a possible replacement for Geant 3. Geant 3 is no longer 
being developed by CERN.

G4 does not do reconstruction - obviously, each user will have to provide this.

Hadronization. This was poor in 1999 but Mikhail has improved this
for physics below the 10 GeV scale. Is this important for Hall-D?
Perhaps for pi0 -> gg, g + A -> A' + n but for rough simulations this
is not important.

Dennis Weygand: Consider Hall-D detector with pi0 into the calorimeter,
how does G4 compare to G3? M.Kossov: G4 just as good as G3.

M.K. There is some work going on in Japan to accelerate the simulation
of EM showers via parameterization.


Mikhail cautioned us that using G4 would require significant training.
There are about 500 classes in G4 and a user must be familiar with
say >10% of these. 


*****************************************************************************
--->> Mikhail thought that to just a Hall-D MC working (input all the
detector geometry and materials, etc) would take 6 months of solid
effort. Why? The geometry is complicated, there are physical and
read-out geometries, it is C++ - powerful but complex.
We all agreed that to get results in 6 month for the next Design Report,
we should use a combination of G3 and MCFast. Nevertheless, G4 will be
needed eventually.

Perhaps a first stab at using G4 would be to simulated 9 GeV photon 
into LH2 target in B field and put in simple detector elements out to
say 50 cm.

*****************************************************************************

There is some work going on at FNAL in using Root for persistancy instead
of Objectivity (a commercial object oriented database that cost >$1000.00)
Persistancy means being able to read and write objects to disk .
(See http://www.cyberdyne-object-sys.com/oofaq/oo-faq-S-3.5.html#S-3.5 for
more info)
Do we need this? Probably not. I'll write a note to the halld-mc list
about geometry storage.

Richard Jones pointed out that there are muons and neutrons produced
when the beam is collimated. Richard also has a program that translates
the MCFast geometry to Geant 3 format.

Paul Eugenio mentioned using a geometry database & material library.
Also MCFast is still useful.


To do:
======

RWM & working group:
   Prepare an outline of the work that needs to be done to contribute
   to the next Design Report in August, 2000.

??? - Ken Hicks?
   Confirm Mikhail's estimates of how much effort is required to do
   full simulation in G4. Someone said that Ken Hicks @ Ohio U has
   experience using G4 for simulations of something at Spring-8.

PE: 
   Look at geometry DB.
   Think about using the E852 shower library for Geant 3 sims of Hall-D.

RWM, EW:
   Check out G4 examples. Work towards a simulation of backgrounds.