> 1) In page 2 you mention about solving the double counting. > > Do you mean that the double counting is due to including the > Z+gamma MC sample, where you have already the Z+njets MC. > > So does the Z+njets MC sample include the ISR and FSR of the > photon ? > > Yes, the Pythia sample includes the ISR and FSR of the photons. $\Delta R$ between a lepton and a photon looks the same in Pythia Drell Yang, Z+Photon Madgraph, and data. The rate of Z+ Fake photon is about 10% of the observed events and it's reproduced by Pythia well enough. We decided to remove the exclusive Z+Gamma MC from the stacked plot, since this contribution is present in the Z+jets sample. > 2) In Page 5 you have slightly better agreement between data and > MC for M(ee)~70 GeV. Previously the difference is about 30% > and now it is about 20%. I suppose this improvement is due to > the removal of the Z+gamma MC. Is this right ? > > You might want to take a look at Giulia Manca's talk : > > http://www-cdf.fnal.gov/internal/WebTalks/Archive/0607/ > 060727_exotic_physics_group/ > 05_060727_exotic_physics_group_Giulia_Manca_1_manca_prebless06.pdf > > There she has one bin (~70-75 GeV) where MC is higher than data. > > The ratio became 20% because of removal of the additional Z+Gamma MC. I compared the ztop0i-like samples and zewk6d ( the one Giulia uses) and I see about the same M(ee) distribution. The 20% disagreement does not go away. The problem is that the expectation in that region is ~20% higher then data. Adding additional contributions (like bbar faking e+e-) will just increase the disagreement slightly. Henry told me that this is a well-know issue. > 3) How much is the systematic uncertainty is from the jet energy > correction ? > > This depends on the signature you are looking at. Basically, in Ht and mEt signatures it's on the order of 5-10%. The contribution from jet energy systematics to mEt+ lepton is tiny. > 4) For the channel with photon in the final state, how do you > estimate the backgrounds from faking photons (jet->gamma, e->gamma) ? > Do you rely on the MC prediction, or do you estimate it from data > (applying measured fake rate onto jets in data) ? > It is not explain in your CDF note. > > We rely on the MC prediction. However, I compared the fake rate from Pythia with the expected from taking # of jets multiplied by the probability of a jet faking a photon. I got about the same answer (~1 event) in both cases. Also I looked at the paper where they measure Z+Gamma cross section. Their rate of Z + Fake Photon is about 10% of the observed events. So, all the three sources agree. (To estimate the photon fake rate given by MC I matched photons at HEPG level with the reconstructed ones). > While I was writing these questions to you, I went to talk to Ray > to get his expert knowledge on photons. As he explained to me that > if your Z+Njets MC sample is just inclusive Z MC from Pythia, then > it should have ISR+FSR photon radiation simulated. So it is correct > that you should not include Z+gamma MC sample. > > You could look at Henry's paper about Generator settings for Pythia and Madgraph. The meaning of the default settings is explained their. So the default Pythia is set to radiate ISR and FSR photons. I can find the CDF note number if you wish. > Furthermore I asked him for the photon candidates (Et>25 GeV), what > fraction will it be from jets faking photon. His opinion is that > most of these photons will be from ISR+FSR of the Z production, and > only small fraction from jet->photon. Therefore if the MC is not getting > the jet->photon fake rate correct (say off by factor of two) it should > not have too much of effect in your case. > > Yeah, you are absolutely right. The good thing is that the MC predicts the rate of the Z+fake photons well enough. > 5) Why did you have to fix the CES Z cut ? What is the effect after fixing this cut ? Any change to your results ? > > The cut was at 217 instead of 230, as I remember. The only difference is that we observe an additional photon in Z->mu mu sample. > 6) Since you removed the Z+gamma MC sample, this also means that all your background estimation is affected. > How do all the distribution plots (for blessing) changed ? > From the Z Pt plot it looks like there is very small change > in the distribution. Have you checked how big is the effect > on the limit that you are setting ? > > The limits are the same. Z+Gamma does not affect Pt(Z) distribution at all. We can see it only in the M(ee) distribution only because it peaks below the Z mass. I've updated the CDF note. The figures does not contain the separate Z+Gamma MC. > 7) In your Full Status talk (June 15th) you did not include > Z+gamma MC sample to your total SM background estimation. > Why did you include it in your preblessing talk ? I was asked to include the Z + gamma background to the expectations. I'm sorry for confusing you so much by including and excluding the Z+Gamma evens. My conclusion is that PYTHIA gives correct ISR and FSR estimates and adding an additional Z+Gamma sample leads to double counting. > > 8) For the category Z+electron+X , you estimate the jet faking > electron from the MC. Since in the jet multiplicity plot one > can see that the MC is underestimating at the higher jet > multiplicity, then it is likely that the fake electrons will > be under estimated. Usually what other analysis do is that they > estimate this from the data. There are two components which are significant for the fake rate estimate: a. number of jets. b. rate for a jet to fake a tight central electron The number of jets disagree only in high jet multiplicity bins which leads to a tiny disagreement in total number of jets. The biggest contribution to the number of jets comes from Z+1jet events where we observe an excellent agreement. I looked at CDF6742 and CDF6592. The rates of jets faking electrons were compared for Data and MC. The rates are found to be in a good agreement. So, PYTHIA should predict the rate of Z+electron events reasonably well.