1. Low-p T Multijet Cross Sections John Krane Iowa State University MC Workshop Oct. 4 2002, Fermilab Part I: Data vs MC, interpreted as physics Part II: Data vs MC, interpreted as a tuning problem

2. 2 John Krane -- Iowa State University Motivation High-p T inclusive jet spectra appear to be well described by NLO QCD Possible exceptions include k T algorithm analysis, possibly also the ratio of cross sections at 630/1800 GeV, large-  dijets (BFKL). But only at 1 or 2 , not actual disagreement Originally, this was a search for BFKL effects, which could produce extra jets in low-Q 2 events

3. 3 John Krane -- Iowa State University The Analysis Jets with E T > 20 GeV, usual jet and event cuts, efficiencies applied but no unsmearing Study inclusive samples of events having at least: 1-jet, 2-jets, 3-jets, 4-jets Compare to normalized Pythia + GEANT and Herwig + GEANT, tune if necessary The usual sample

4. 4 John Krane -- Iowa State University Data and Pythia Inclusive xsec looks fine Multijet xsecs exhibit deviations from Pythia Let’s pretend it’s physics points=data, histo=Pythia

5. 5 John Krane -- Iowa State University (D-T)/T Solid lines: energy scale  lum uncertainty Dash: smearing uncertainty Dotted: total error in ratio

6. 6 John Krane -- Iowa State University Data and Herwig Started generating jets at 0.5 GeV Multijet xsecs exhibit similar deviations points=data, histo=Herwig

7. 7 John Krane -- Iowa State University Vector sum p T Define The more jets in the event, the more imbalance in energy Could this be ISR, with p T lost down the Beampipe? Events > 250 are the excess in 3+ jet events (Data and Pythia) >150 in 4+

8. 8 John Krane -- Iowa State University Angles in 3-jet events Find which jet is “the third one” by isolating the two jets with minimal  p T Many back-to-back in the data Usually, third jet is near one of the first two, but more so in Pythia (Data and Pythia)

9. 9 John Krane -- Iowa State University Angles in 3-jet events Find which jet is “the third one” by isolating the two jets with minimal  p T Min  p T not bad… Third jet is often at 90 degrees, often composed of underlying event E (Data and Herwig)

10. 10 John Krane -- Iowa State University Signs point to initial state radiation effects in data – DGLAP style? – BFKL style? …or a need to tune the MC Early impressions of these results

11. 11 John Krane -- Iowa State University “Try tuning Pythia, also compare to Herwig; see what works…” Herwig defaults also did poorly Many iterations required Only compared to distributions shown today (and a few other very similar ones…) Pythia works if PARP(83) = 0.32 (from 0.5) Fraction of matter in the proton “core” Herwig works if PTMIN = 3.7 GeV p T generation threshold A Multiple parton scattering parameter Does this change underlying event in some way? Both higher and lower values do worse! Didn’t try Jimmy…

12. 12 John Krane -- Iowa State University Data and Tuned MC Points = Data Solid = Pythia Dash = Herwig

13. 13 John Krane -- Iowa State University (D-T)/T No remaining deviations from data Is this because there were no ISR effects? Answer lies in the validity of our tuning

14. 14 John Krane -- Iowa State University Vector sum p T Tuned MC reproduced the small “shoulder” in addition to the 3+ and 4+

15. 15 John Krane -- Iowa State University Angles with Pythia (and Jetrad) Points = Data Solid = Pythia (error bands and…) Dash-dot = Jetrad

16. 16 John Krane -- Iowa State University Angles with Herwig Points = Data Solid = Herwig Dot = Herwig with cut on merged jets Peak (from “underlying event jets”) becomes enormous if ptmin>3.7 GeV Pythia’s CKIN(3) shows no such sensitivity… Cross section shapes very strange if ptmin<3.7 GeV

17. 17 John Krane -- Iowa State University Conclusions Results not entirely satisfying – Would like to make definitive statements about ISR – …or provide solid tuning suggestions Instead, we found sensitivity to several params. – Think the multiple parton scattering is constrained by other data, we provide a new handle – We don’t understand the Herwig tuning at all Our decision: publish the data, leave tuning to experts with a more global view – Tuning isn’t really our forté – If we do it, we probably want a second paper out of it!

18. 18 John Krane -- Iowa State University Backup Slides

19. 19 John Krane -- Iowa State University Cone Algorithm Details (Run I) Draw a cone around a “seed” Calc sum E T, and E T -weighted position Draw new cone here and recalculate sum E T, position Reiterate until stable

20. 20 John Krane -- Iowa State University Energy Scale Correction back to “the particle level” Remove noise, underlying event,extra pp interactions Correct for detector response Undo misassignment of particle energies to jets calorimeter jet Time parton jet particle jet hadrons  CH FH EM  K