TA Meeting University of Utah

1
TA Meeting University of Utah

• Fabrice Cohen
• ICRR University of Tokyo, Japan

15-16 July 2006
2
Air Shower Simulation

• Database in more details

3
How to build database ?
Which Software ? mysql, postgres, db4o

• Mysql very popular, accessible by many
  languages (C,C, java)

• Db4o Object oriented database and data format,
  
• developed for Java (can be used by C)

Originally, structure of db was thought in case
of mySQL (different kind of tables to access to
data) but TA Simulation code (SD and FD) is in
Java, if we can have only one kind of database,
it is better !
Recently, we re-defined the structure in case of
DB4o.
4
Simulation Code
Software Cosmos (v7.20) http//cosmos.n.kanaga
wa-u.ac.jp/cosmosHome/Cosmos/cossrcindex.html
Hadronic model available DPMJet III QGSJet 2
database will be built first with DMPJet III
Specificities Full MC Hybrid
mode Thinning Parallel calculation
5
How to do database ?
Build a Model shower

• 2 Parts
• Longitudinal profile Nparticles Vs Depth
  (g.cm-2)
• Particles at ground for a given depth,
• Lat. dist., nrj spec, arr. time

6
Longitudinal profile
Simulation 1000 showers / energy / zenith
angle Energy 1e17, 1e18, 1e19, 1e20, 1e21eV
(intermediary nrj added) Cos?1.0 ? 0.5 (step
0.025) Primary particle proton Thinning method
used (1000 showers in 10h with 25 CPU)
Nb of part. e, ?, µ, h
depth
7
Particles at ground Full MC
Simulation with full MC of proton primary one
shower at 1018 and one shower at 1019
eV, Partially full MC at 1020 eV for cos ?1.0 ?
0.5 (0.025)
Divide the profile in layer of 25 g/cm2
(horizontal plane)
For each layer, we record information of
particles
8
Particles at ground (2)
For each observation level, we build a spider

• Bins are defined in r and f
• r 42 bins (0.01 130 Moliere unit)
• F 12 bins (30 degrees each)

9
In one bin
Srf surface of the bin NTOT total nb of
part. falling in the bin NREC nb of part.
recorded with detailed information (E, part type,
r, ?, time) Maximum nb of particles recorded in
a bin, for each type 7500 (to have enough
statistics)
10
DB4o Structure
Preliminary
LDD
FDD
Object ShowerLong Header E0,
prim. Type, cos ?, nb of shower
simulated, Had. Model etc...
VectorltProfileLonggt ------- ProfileLong
Z0, depth max (??)
Profile -------- Profile depth, age, Ne, Nµ,
N?, Nh, dedx
Object bin Header E0,
prim. Type, cos ?, had. Model threshold
of part., depth, age, value of moliere
radius, r (meter), f(º) Real
nb of particles in the bin. Nb of part.
Recorded in the bin. Total nb of particles
at ground (all r, allf) (for each type e,
?, µ, h) VectorltParticlesgt -------
Particles part. type, r , t, E , ?
11
How use it ?
Eo, cos?, prim. type
It will give you one object ShowerLong (with
1000 diff. profiles)
Get randomly one Profile ( N Vs depth )
define Observation level
Information you have Depth, age, Ne, Ng, Nm,
Nh, dedx (at ground)
Search in database, the closest value to age (for
this cos ?)
The result is one spider (all bins at one obs
lev, all r, all f)
Fix r and f
One bin
12
Information in the bin
Srf area of the bin
Detector, area Sd
Values given by database NTOT total nb
of part. falling in the bin (each type) NREC
nb of part. recorded with detailed
information (each type) NGROUND nb of part. at
obs. lev.
NGROUND nb of part. at obs. lev. (profile)
Density in the bin ? NTOT / Srf
NTOT total nb part. in (r,f) bin
NTOT (Nground / NGROUND ) N TOT
Nb part. falling on detector nD ? SD
poissonian fluct.
Random sampling of nD particles from NREC (if lt
Nb max recorded)
13
Summary

• Almost all full simulation are done.
• Some routines will used to make some calculation
• ( interpolation between bins )
• Interface with SD simulation (which imput for SD
  simulation )
• We have to fill database (in java db40, c
  db40)
• (around 1-2 month to finish simulation and fill
  bd (?) )
• after the db filled, make routines for
  calculations.
• Next plan
• Simulation with photon primary, heavy nuclei..
  (?)
• Simulation with QGSJetII
• (same db for Corsika for comparison (?))