1. Jet Studies 1. Changes in seed finding for cone jets reconstruction
Emmanuel Busato
1. Changes in seed finding for cone jets reconstruction
2. t42 and jets

2. 1. Changes in seed finding for cone jets reconstruction
- technical description
- effect on cone R=0.5 jets reconstruction
- what's the best value for the new parameter « n »
- conclusion

3. ? Technical description List of towers (pT ordered)
Take the highest pT tower in the list « T »
PT > seedET = 500MeV
Current preclustering
(Simple Cone)
yes no
Go to clustering around T
(with R=coneSize=0.3).
Remove progressively towers
from the list.
Stop preclustering

4. ? List of towers (pT ordered) Modified preclustering Yes no yes no
Take the highest pT tower in the list « T » ?
pT > seedET =500MeV
Yes
Take highest ET cell in T : « C »
if C is CH or ECMG
and if C has E < n * sigma
 pT = pT  ETC no
pT > seedET =500MeV ?
yes
Stop preclustering no
Go to clustering around T
(with R=coneSize=0.3).
Remove progressively towers
from the list.

5. Effect on cone R=0.5 jets reconstruction
Data sample :
 Run # (30 Dec) and # (16 feb)
Reconstrution version : p
 Quality cuts :
0.05<emf<0.95
chf<0.4
hotf<10
n90>1
 No trigger selection
Event by event comparison
Number of jets/event
 5.1% of events have more jets with old seeding
1.4% of events have more jets with new seeding

6. Jets not found with new seeding (R>0.5)
mean = 0.43
mean = 2.22
Jets found in both cases Jets not found with new seeding
Jet not found with new seeding :
- mostly low pT jets
- in ICR
- big number of merge

7. New Jets found with new seeding (R>0.5)
- First, an example :
Old preclustering :
1 “normal” seed + 1 CH seed
 2 low pT clusters
reconstruted with cone algo,
individually above threshold
New preclustering :
1 “normal” seed
 1 low pT cluster reconstruted with cone algo, above threshod
__________________________ Before split/merge __________________________
Splitting :
Assign shared cells to nearest
“protojets”
 clusters below threshold
(or fail quality cut)
No jets left
One jet left
__________________________ After split/merge __________________________
No split/merge

8. - What are those jets ?
 Look like low pT good jets

9. What's the best value for the new parameter « n » ?
nb1
nb2
nb1
nb2

10. Conclusion - Jets not found with new seeding look like noise jets
- New jets found with new seeding look like low pT good jets
Looking at f90 distribution as a function of n, we decided to remove all CH seed (n=1000sigma)
This new seeding scheme is in p14 with n=1000sigma

11. 2. t42 and jets
- jets properties (emf, chf, f90 ...) without and with t42
- jets reconstruction without and with t42

12. Jets properties (emf, chf, f90 ...) without and with t42
Top group’s alljets skim (reco version p ) :
* Passes the 4JT10 trigger.
* At least 4 jets (JCCB).
* HT > 100 GeV (just plain sum of uncorrected JCCB jets).
  events
Jets are JES corrected.
No quality cuts applied because jet id distributions change significantly.
Color code : t no t42

13. Color code : t42 no t42 New seeding not applied, its effect
is small once t42 is applied

14. Warm zone around phi ~ 0.8
disappears with new seeding !

15. Summary - id distributions change significantly with t42
- t42 reduces the number of jets by ~ 11%
- mostly in -2 < eta < 2
- mostly low pt jets
- mostly high f90 jets
- merge > 0
- t42 doesn’t affect jets with 0 merge

16. Jets reconstruction without and with t42
There are jets, close to a track jet (R<0.5) ( probably good), not found with t42 !
 Part of these jets can be explained simply
• low pT jets : - threshold effect (the same jet is less energetic with t42)
• high pT jets : there is still 0.9 of jets with not found with t42
- threshold effect up to 25 GeV
- unphysical cases :
- split/merge effect :
pT same jet
R(cal jet, track jet) < 0.5
pT> 25

17. JETS
PT=64 GeV
merge=2
split=0
PT=42 GeV
merge=1
split=1
PT=31 GeV
merge=0
Without t42
With t42
This particular example is not isolated, approximately 40 of high pT jets close
a track jet not found with t42 can be explained by this effect.

18.  Another simple reason could be that we remove good jets
 Look at MC sample without noise :
- no calorimeter noise
- no minimum bias
Single top events – schannel production mode
JETS
without t42 : Number of jets = 3568
with t42 : Number of jets = 3540
 Reduction of  0.8

19. t42 not42 Same number of Ws ! ELECTRONS W selection :
Hmx8 < 20
isolation < 15
EMF>0.9
no track match
PT (electron)>20 Gev and MET>20 GeV
not42
t42
Same number of Ws !

20. Conclusion and to do
 Looking at single top events without noise reconstructed without and
with t42, we find that the number of jets is reduced by  0.8.
 look what happens on a QCD sample without noise
 understand lost jets
Number of Ws not reduced with t42
 In data, good jets (and all objects) are polluted with noise, we saw that split/merge effects changes jets multiplicity.
 make sure that it goes in the good direction