Simulation Status/Plans Malcolm Ellis Sci Fi Tracker Meeting Imperial College, 10th September 2004

Download Report

Transcript Simulation Status/Plans Malcolm Ellis Sci Fi Tracker Meeting Imperial College, 10th September 2004

Simulation Status/Plans
Malcolm Ellis
Sci Fi Tracker Meeting
Imperial College, 10th September 2004
1
Outline
• Status at Osaka meeting
• Unresolved issues:
– PT, Pz resolution
– Emittance calculation
– Correct beam and
Background
• Aims by RAL meeting:
– Correct beam
(G4BeamLine)
– New RF background
(Rikard)
– New track model?
– Resolutions
– Emittance calculation
2
Status at Osaka Meeting
• Pattern recognition and reconstruction had been
improved in a number of areas.
• In particular:
– Improved dE/dx model used in track parameter
propagation.
– Better seed parameters to track fit.
– Moving towards removal of MINUIT fit.
– Increased efficiency in CPU and memory use per
event (important for high RF background rates).
3
Unresolved Issues
• Some concerns over PT resolution compared to
expectations based on “back of the envelope” type
calculations.
• Pz resolution (RMS of PZ resolution not less than 10% of
the RMS of the true distribution) and hence t’.
• Still hadn’t made an actual emittance calculation, so
couldn’t show emittance resolution.
• Default G4MICE beam isn’t terribly realistic, and the RF
background simulation has since been improved.
4
Aims by RAL Meeting
• Use a more correct beam – G4BeamLine
June04
• Use a better RF background simulation
(Rikard)
• New track model?
• Final resolutions (hopefully show better
than 10% resolutions in all parameters)
• Demonstrate calculation of emittance, and
of course the resolution!
5
G4BeamLine Beam
• During August, Kenny made a run of
G4BeamLine using the June04 setup, and
produced approximately 15k muons going
through the channel.
• The output of G4BL can be read into G4MICE.
• In order to get sufficient statistics, we are going
to need to use the distribution from G4BL as the
input for a generator in G4MICE (rather than
attempting to produce 1,000,000+ muons in
G4BL!) – not sure yet exactly how to achieve
this...
6
Pz Resolution (G4MICE)
• At the Osaka
meeting,
resolution shown
was 3 MeV/c
• The RMS of the
PZ distribution
was 5.3 MeV/c –
“resolution” is
therefore ~60% not very good...
7
PZ from G4BeamLine
• RMS of the G4BL
PZ distribution is
43.1 MeV/c
• The “resolution” is
therefore now
7.0%!
• PZ resolution is
now acceptable
without changing
the reconstruction
code...
8
t’ Resolution
• Because t’ = E/PZ,
the same effect as
seen with the PZ
resolution happens
with t’
• At Osaka, the
“resolution” was
5.62E-3 / 2.45E-2 =
23%.
• The RMS of the t’
distribution has
increased now to
1.58E-1, therefore
the “resolution” is
now 3.6%
9
RF Background Simulation
• At NuFact/Osaka MICE meeting, Rikard
presented the latest RF background simulation.
• Currently awaiting the resolution of a few
technical issues, then it will be commited to the
G4MICE repository.
• Talk to be given at next Video Conference. Once
simulation is “approved”, it will be used for all
results for the RAL meeting.
• There is also a better description of the material
in the cooling channel coming from Yagmur.
10
New Track Model
• At Osaka, discussion about the pattern
recognition efficiency as a function of PT resulted
in the conclusion that a PT-insensitive track
model is needed in order to have high efficiency
at ultra-low PT.
• Ken has defined such a model, and done the
maths, and produced an Excel/VB based
implementation to test it.
• Plan is to make a C++ version under the current
G4MICE framework and use it in the pattern
recognition.
• However, is it really needed?
11
PT Distribution from G4BL
• 10 tracks out of
32k with very low
PT
• Transition from
“curved” to
“straight”
happens at about
PT = 0.5 MeV/c
• Do we care about
0.03% loss of
tracks in a 0.1%
experiment?
12
Emittance Calculation
• Chris has developed emittance calculation code
(see his talk), which will be used to calculate the
true (i.e. Monte Carlo) and reconstructed
emittances, and make a study of emittance
resolution.
• We will be unable to generate many 1M event
runs (each of which in theory produce one
measurement of true and reconstructed
emittance), so will look at different sized subsets
of the data that we can produce, and study the
emittance resolution as a function of statistics
and extrapolate to the nominal MICE case.
13
Conclusions
•
•
We are almost there.
The two principle objections at Osaka
should be answered by RAL:
1. t’ resolution – broader beam from G4BL
fixes that for us
2. calculation of emittance and use of better
beam and RF background simulation are the
remaining hurdles.
14