Delayed Photons in the exclusive g +Met final state Dave Toback for the Delayed Photon Group *

1. Delayed Photons in the exclusive g+Met final state Dave Toback for the Delayed Photon Group* Texas A&M University *Apologies for not showing this to all members before showing it…

2. g+Met excess • The delayed photon excess is very exciting • Lots of things for the authors to do before we can publish this • Need to rule out prosaic explanations for excess at high times • Try to get more evidence that we understand production • Try to explain why excess doesn’t look asymmetric in CESZ • Need to explain why there is no excess in g+jet+met • Same with sideband

3. Wen  gfake+Met Hypothesis An ensemble of Photon candidates that are more likely to fly across the detector will have a different timing distribution than photons coming from the usual Z and h distribution Picking “wrong vertex” produces Z=100cm

4. The Wrong Vertex Assumption • Wrong distribution created from reconstructed Wen events (good track) • Good model for physics that is produced in the same way Wen events are produced • Wengfake+Met has a different production • Photon fake rate is Z and h dependent • Different timing distribution (under study with MC) • Sensitive to true Z distribution Right vertex “Wrong” vertex

5. Geometric effects in real data • Lots of examples of this type of effect • Look at real electrons and pick ones where we get the wrong vertex well separated from the true vertex • Can calculate Et from right vertex and wrong vertex • Case where Et correctly calculated>40 GeV, but incorrectly calculated (from wrong vertex)<40 GeV • Shifted by 700 ps • Huge asymmetry in events in [2,7] compared to [-7,-2]

6. Beam profile • We know the MC’s don’t have the correct beam profile implemented • We don’t have any good studies about the beam profile (in Z or time) at |Z|>>80 cm. • Are things linear out there still? Satellites certainly aren’t linear-type behavior…. Mean Vertex time vs. Z also makes a difference

7. eg acceptance A Toy MC with a simple Z distribution with a Gaussian shape, a mean of 100 cm and an RMS produces a timing distribution centered at 2 ns

8. Putting it all together Toy MC Simulation with admixture of fake eg where production is different the normal Z=0 beta* Looks about the same to me… Allowing smaller Z would just put more events under the right-vertex distribution so that’s ok too… Real MC in progress

9. More studies needed • Fake rate of eg as a function of Z and h • Understand beam outside |Z|>60 cm • Work on both is in progress

10. Another Hypothesis: g+jetg+met • g+jet where the jet went down the beam pipe which is high eta production • Low eta wouldn’t have Met • Biased towards high Z again Photon Z=100cm Jet that went down the beam pipe Note: Photon always at large Z, opposite side to vertex Sideband more likely to produce vertex? Would need to study Study acceptance vs. vert Z

11. CES and CPR to get impact parameter and Z Vert Direct Evidence of Large Zprod? Would like a direct handle on the photon to see if there is any evidence that these “photons” are not coming from the beam line or the usual Z distribution • CES-CPR pointing capabilities • CPR pads are 12.5cm in Z by 5 degrees in phi • Full MC studies nearly finished • Can test with data W’s • Started thinking about LSHR

12. Plug Timing Timing in the Plug calorimeter to see if there is any evidence this is beam-gas interactions that produced fake tracks • PEM and PHA

13. Not sure what to make of this

14. The To-Do list • Finish testing the Wengfake+met hypothesis: W events that produce fake photons are biased towards Large Zvert and large h (not well modeled by wrong vertex estimate)  shifted timing distribution and excess • Photon fake rate as a function of h and Z • Better beam understanding MC study in progress to see if small systematic uncertainties in Z distribution could produce this effect since W has a large cross section • Zeg selected using kinematic quanties only. NO tracking information on either leg: eg mass only • Direct evidence that these photons come from Large Z • CES-CPR • LSHR • Test g+jet  g+Met hypothesis • Check timing in plug for non-collision sources

15. Conclusion • Clearly the excess is an interesting thing and statistically significant • Need to understand if this is something real or if it’s some forgotten but prosaic background before going public in any way • “Extraordinary claims require extraordinary evidence” – Carl Sagan • Lots of tests to be done, • Work on things mentioned is in progress, but just starting… • “Roma non fu fatta in un giorno”

16. Back up slides

17. 6: EMTiming behavior at High Lum • Since we’re now in high luminosity we’ve been seeing funny behavior in the EMTiming system at very low levels. • Admem’s often read out zero • Fake rates go up • Overlaps are more likely when lots of min-bias collisions occurring at the same time • HADTDC showed noise at high rates

18. Collision or Not collision? • Unlikely to be non-collision based •  should have seen in events without a vertex • New source of physics? Anomalous c10 pair-production in GMSB SUSY (Toback & Wagner PRD70, 114032 (2004)) •  If it were real then ~25% should have a jet and we would see this in g+met+jet • Rate W+Jet/Rate W = 0.25 • Prosaic background? Obvious guess is that this is some SM collision that is poorly measured (and when it’s well measured it looks like something else)

19. Type of effect Seen before? Yes • This effect dominates the PEM calibrations • See CDFNote 7928 (2005) & Peter Wagner’s Thesis • CDF Note 8932, Appendix D (page 202) • Need “ring corrections” to get the PEM calibrations right. When you look at W’s in the plug you get a mean offset by 0.5 ns because what “produced” the calibration sample wasn’t centered at vertZ=0 cm

20. Other places this shows up • Opposite side makes a difference in wrong-vertex timing distribution vs. Z • Would needs to be very large Z CDFNote 7960

21. Useful things to be working on… • MC study of g+jet to get rates • CES/CPR to see if there is evidence that the photons in the “excess region” point to non-beam line events • Use what Z pointing we have

22. Work? • A real MC would be needed, but if it’s g+jet the collisions with small vertZ would show up as g+jet and/or be symmetric in vertZ • Wrong-vertex timing distribution • If the photons were on the same side as the CESZ then high angle production would hit PEM and not be part of our sample

23. Gamma Met excess Low Z production  both go down the beam pipe? PEM Photon Z=0cm Jet that went down the beam pipe So… is either g+jet (low Z vert and low h), gpem+jet (low vert and high h) or gcentral CEM or PEM+met (high vert and high h). Wouldn’t give g+met+jet I would guess, and g+jet would get swamped in SM background. Still need to think about this…

24. Study 3: MC acceptance • Simple acceptance turn-on of an error function at 100 cm with a width of 10 cm would produce a timing distribution with a mean of 2ns. • Can’t use current MC’s for rate and timing… next slides g+jet g+jet or g+jet+ met? g+jet  g+Met

25. Toy MC • Ran a toy MC to see if this could produce the effect • EMTsim Resolution 0.6 ns • Vertex Res 0.2 ns • Take into account time vs. VertZ variation • Does a nice job of reproducing expected right and wrong vertex distributions

26. What do we need to produce? • The excess looks (to my eye) to be an Gaussian with a mean of 2 ns and an RMS of about 2 ns • (at least on the positive side) • What Z distribution would produce this? • Take CESZ to be flat except required vertex on opposite side • Take into account t vs. Z and resolution

27. Do something simple • A simple Z distribution with a Gaussian shape, a mean of 100 cm and an RMS of 10 cm does the trick