1. 1

2. 2 Old Efficiency Curve This is not an Apples to Apples comparison: ● SM PYTHIA includes off-shell Z, also allows inclusive decay of second Z

3. 3 New Efficiency Curve New Efficiency Curve addresses these facts: ● Requires PYTHIA events to be generated in the same window and channel as Baur events (Z window, e/mu and quark). ● Error curves probably still need tweaking.

4. 4 Baur vs. PYTHIA There remain two large differences between our fully simulated Baur and PYTHIA SM Monte Carlo: The ZZ (Center of Mass) pT distribution, due to an accidental rescaling of Q 2 to 1.0 and the Z pT distribution, showing differences between the Z pT for Baur and PYTHIA

5. 5 ZZ pT Distribution. The easiest way to correct the difference in Center of Mass pT would be to re-weight the distribution so that the Baur MC had the same distribution as the PYTHIA. This is impossible because the Baur distribution has no events on the tail. We can, though, reverse the process to look at the difference. ● Divide the Baur ZZ pT distribution by the PYTHIA one to create a “Weighting Function” ● Weight the PYTHIA MC by this function so that PYTHIA matches Baur. ● Use the shift in Z pT (our efficiency variable) to correct for this effect. Can't weight Baur to look like PYTHIA in this region

6. 6 ZZ pT Correction Plot of Z pT for normal PYTHIA/events weighted to look like Baur ● This could be applied at the same stage that we apply efficiencies. ● Note the direction of the trend.

7. 7 Z pT Distribution Baur has more events in the high-pT tails than PYTHIA. This would normally be very straight- forward to correct. ● Divide PYTHIA Z pT distribution by Baur distribution to obtain a correction function as a function of Z pT. ● Apply this correction function to Baur to normalize it with PYTHIA.

8. 8 Z pT Correction Maps best as a linear function in the signal region ● May need a bit of tweaking to satisfy ● Note the direction.

9. 9 Uh-Oh Correcting in both dimensions simultaneously results in trouble: Two corrections that are opposite to each other. Solutions: A) Panic B) Ignore one of them Let's explore B for a moment.

10. 10 Correction Proposal Why not only apply the Z pT correction? The Z pT correction acts to normalize the Baur results to the PYTHIA expectation. If we correct for ZZ pT, increasing the number of events in the tails of the Z pT distribution, we will have to re- normalize to account for this. I propose the following: ● Correct the Baur MC by a function that would normalize the Standard Model ZZ sample to its PYTHIA expectation. ● Treat this as a correction, and not a systematic. ● Create this function by dividing the raw generator Z pT spectrum from PYTHIA by the one for Baur. Apply no other corrections that would then have to be counter-corrected for. However, everything I've just shown you is totally wrong.

11. 11 The Problem This plot is of reconstructed pT divided by generator pT. BUT, I've been getting generator pT from the PYTHIA+BAUR samples. In other words, THIS is my denominator. I then apply this to generator level Baur MC, with no PYTHIA, that does not have this distribution.

12. 12 The Solution But all this can be corrected for by using the Baur only (No PYTHIA, No LesHouches format, no nothing) Z pT curve as the source of the Z pT correction. By normalizing the raw generator output to the PYTHIA SM expectation, we simultaneously correct for: ● Differences between Baur and PYTHIA in the Z pT distribution. ● The effects of proper gluon radiation. ● Any other generator level effects modeled by PYTHIA. This should provide us with one function that will allow us to correct Baur MC for all the hadron collision physics modeled by PYTHIA.

13. 13 Impact Previous expected limits for f 4 Z : (-0.1,0.1) Expected limits for f 4 Z with inclusion of the “bad” correction (not accounting for gluon radiation): (-0.12,0.12) This uses the maximum systematic method.

14. 14 What Next? Assuming this meets with approval, I have some more steps to complete: ● Do some more Generator MC: (I'm missing some points now that the limits have moved) ● Recalculate all the limits for fZ and fgamma (with and without systematics) ● Present this at EWK As I remember it, the main issues last time were: ● Comparison between Baur and PYTHIA for SM (now addressed) ● The looseness of the Efficiency Curve error bars (now addressed) ● The lack of interference in fully simulated MC (now addressed) ● After checking our final limits, we should have successfully handled all the large questions that were open at the end of last summer.