K group activitiesK group activities
K+K- retracking features
New vs Old : resolution
future planning.
Conclusion and outlook
K+K- retracking features
New vs Old : resolution
future planning.
Conclusion and outlook
KPM ReconstructionKPM Reconstruction
KPM Selection KPM retracking DST production
Done and working
Finalized, under test
Not yet addressed
The KPM reconstruction has been updated/ridesigned, in particular with respect of: Track merging, T0 finding, K± TOF correction, Multiple scattering treatment optimization, ECLO and Event selection interaction
In Otranto In Otranto
Done andworking
Since August Under debugging
now
KPM retracking:new & updated modules
KPM retracking:new & updated modules
Kmerge
T0_find
atfmod
Merge tracks and udpdate pattern candidates. Orders and updates K id from different ECL algos in ECLO
Pattern Banks & ECLO
Finds the KPM t0 global T0GL
New K TOF implementation, specific K m.s. treatment, correct hit times for T0
Fit Banks
module New features Updated bank
K+K- reconstruction: merging
K+K- reconstruction: merging
Due to K large TOF, wrong T0 and large m.s. the “standard” rec. produces a large fraction of broken tracks (at the pattern level!). No recovery possible by the Fit
The Kmerge module merge all the possible tracks (not only K+K- ) on the basis of their angle, distance and momenta and rebuild the KPM ECLO banks. After merging and T0 correction, the tracking/pass=2 (usually) splits only the correct kinks
Kaon track
Kaon track
K reconstruction: T0 findingK reconstruction: T0 finding
o Applied to tagged events with K decay products with ECAL cluster associated.
o From K decay vertex position, cluster times, momentum of K and charged daughter at vertex is possible to trace back the T0 of the event.
t+-t- (ns)
2001 data
Kresolution/1Kresolution/1
In Otranto we had switched off multiple scattering. dE/dx implementation did not distinguish in between pion and kaon.
Pools (errors) were wrong for kaons!
Sigma = 3.3 MeV Sigma = 3.4 MeV
Sigma = 2.0 MeV Sigma = 2.0 MeVNew
Old
K pullsK pulls
Pools are now ok!
The dE/dx for a kaon is subtracted while tracking and multiple scattering is taken into account in the error matrix
Pull = 1.1 Pull = 1.15 New
Old
K retracking AC pathK retracking AC path
use kpm_restore dcdbini raw2dtce t0meno dconvr dcprar merge atfmod/par=1 vtxfin/par=1 trclmod/par=1 eclmod/par=3 ecl_kpmfilt t0_find/par=4 clufixtin/par=2 kmerge atfmod/par=2 vtxfin/par=2 trclmod/par=2
The correct merging of the broken tracks must be done at level of pattern candidates DTCE1 and DHRE1 banks are needed. The minimal AC path for K retracking is reported. The red modules have been updated/written for this stream
The KPM retracking is very CPU consuming. The production rate on fibm12 has been estimated to be ~ 1 pb-
1/day/CPU
Now M.S. is not switched off for K here !!!But it was switched off in Otranto.
New vs Old recostructionNew vs Old recostruction
We have tuned the retracking on MC. This work has been done
on events selected by the NEW selection algorithms, reconstructed
with the standard tracking (no dE/dx for K+/- and no MS corrections).
The problem is a multivariate one.. We have analized one
refinement at the time.
The efficiencies have been studied in a subsample of tagged
events: given a tag we look at what happen in the opposite
hemisphere.
New vs Old : K momentum resolution/2
New vs Old : K momentum resolution/2
The distorsion induced in the OLD rec. gets more evident as far the track gets longer
Kaon Prec – Ptrue distribution at the decay vertex versus the Kaon decay path
Effect on dataEffect on data
-- old recostruction-- new recostruction
The new recostructionslightly improves resolution
New vs Old: vertex resolution
New vs Old: vertex resolution
New
Old
The vertex resolutionis slightly improved. In particular tails are reduced
New vs Old : broken tracks New vs Old : broken tracks
Distribution of total length vs momentum of splitted kaons of the entire MC sample (1.7x105 ev).The merging get rid of most of the broken track better decay length distribution !
OLD
NEW
New vs Old : tracking eff. vs DC hitsNew vs Old : tracking eff. vs DC hits
On the selected sample the tracking efficiency are similar.
K tracking efficiency
New vs Old : tracking eff. vs cos K
New vs Old : tracking eff. vs cos K
Due to the request for at least a TAG to compute any eff, large events statistic is needed.
Rec.
Rec.
New vs Old : tagging eff. vs DC hits
New vs Old : tagging eff. vs DC hits
The new reconstruction seems to behaviour better than the standard one. The tagging efficiency is sensitive to the combination of tracking and vertex eff iciency
N.B. is the eff of a very clean sub-sample!
Tagged/tracked eff.
Future planning:Future planning:
The cross section: (next talk) memo by end of 2002.
Br (K+ Br ( K+ Maxim PhD thesis.
Absolute Br(K+) by summer 2003 (2 body, semileptonic).
Br (K+ Andrea Ventura PHD thesis by summer 2003.
Br (K+ e+Andrea Ventura PHD thesis by summer 2003.
ConclusionsConclusions
• The retracking code of the KPM stream is in its final shape, yet has to be tested on high statistics
• Sizeable (but not fantastic) improvement with respect to the standard reconstruction at first glance
• We now have good pulls. • DST production is starting (it should last 1
month).