2.5.x Hadronic Checks (click on the for the plots)

First, the problems I saw with the new version of the H1JetCalibration; I just want to explain this because of the problems with the VC today (no camera... grrrr... ) The event selection is the full high Q2 inclusive selection used in the hadronicChecks package.


Here are the 4 main plots, uncalibrated, using h1oo 2.5.8.


Here are the same 4 plots, calibrated, using h1oo 2.5.8, and the version of H1JetCalibration in the release - ie this contains the original (1 step) method, and the original constants. Here, the calibration is seen to work best at lower values of gammah (say < 110^o) and large Ptda (say > 20 GeV).


Now the same 4 plots, calibrated, using the cvs HEAD of H1JetCalibration. At first I thought it might be a problem with the 1st step of the new "2 step" calibration. As was explained to me, this part is supposed to do nothing for 99-00, so I checked if this was the case - and it is. See the 4-vector output of an event here - the names are clear if you look at the cvs HEAD of H1JetCalibration/H1JetCalibration.C, lines 665-666. The new part is not affecting the calculation at all. Then I rechecked the 2.5.8 release version, and found this output - the vectors are different - which led me to discover that new constants had been derived for 99-00. Because of the brouhaha of todays meeting I wasn't sure what these constants are with respect to forward tracks or whatever... Anyhow, the plots look worse now, with 4% over the full kinematic range a reasonable estimate. Can a clear statement be made from the Orsay guys about what these constants are and (near) future plans for 99-00? This would be much appreciated.

The second issue was with timing, which again suffered because of the conditions in the VC. Benjamin asked me to do some basic benchmark tests, which I did with these 2.5.8 files. The data files contain 321,012 events; the django 196,102 events. My tests were done on 2 samples: the simple NC selection used in the H1AnalysisExample and once again the full high Q2 inclusive selection used by the hadronicChecks package. The results of these tests are shown in this table, where the numbers are time taken in seconds. "HadrooII 2.5.8" is used where the uncalibrated hadronic 4-vector is passed in as the sum of the HAT vectors: LarComb, SpaComb and TraComb, which is identically equal to H1PartCand->IsHFS(). This was done to save a little time as then in this mode only the mODS branch H1PartJetArray is needed by the calibration. Additionally, we can now calibrate this 4-vector and then remove electron cones, add in the Iron etc (see the cvs HEAD of H1Calculator/H1CalcHad.C). However, the new method ("HadrooII cvs HEAD") also requires the H1PartCandArray, so the time saved by passing in the HAT vectors doesn't help. The result of applying the hadronic calibration (for people accessing data via the H1Calculator, at least) is for a simple selection at least a 4-fold increase in time taken by the analysis job and even for a detailed analysis a factor of 1.5. I appreciate how this is not so important with respect to 2003-04 analyses, but just wanted to quantify the increase in time taken. Also, I don't see an immediate solution if this is the way that the calibration must be applied. One last thing: I know this may all seem a little technical and possibly more relevant to the OO meeting itself, but at least Oliver, Bob and Andrey are using this method of accessing the data and they will see the increase in time taken.

All of the control plots as one file

The four main plots: pth/ptda vs ptda, pth/ptda vs gammah^e and the same dependence for yh/yda

Fits Gammah plots
Fits Gammahe plots
Fits Ptda plots

Ratio Gammah plots
Ratio Gammahe plots
Ratio Ptda plots

All of the control plots as one file

The four main plots: pth/ptda vs ptda, pth/ptda vs gammah^e and the same dependence for yh/yda

TProfiles of kinematic balances as a function of gammah
TProfiles of kinematic balances as a function of ptda
TProfiles of kinematic balances as a function of pte
TProfiles of kinematic balances as a function of gammah^e

Data Pth/Pte in bins of gammah
Data Pth/Ptda in bins of gammah
Data Yh/Yda in bins of gammah

Django Pth/Pte in bins of gammah
Django Pth/Ptda in bins of gammah
Django Yh/Yda in bins of gammah

Data Pth/Pte in bins of gammah^e
Data Pth/Ptda in bins of gammah^e
Data Yh/Yda in bins of gammah^e

Django Pth/Pte in bins of gammah^e
Django Pth/Ptda in bins of gammah^e
Django Yh/Yda in bins of gammah^e

Data Pth/Pte in bins of ptda
Data Pth/Ptda in bins of ptda
Data Yh/Yda in bins of ptda

Django Pth/Pte in bins of ptda
Django Pth/Ptda in bins of ptda
Django Yh/Yda in bins of ptda

Mean of Pth/Pte fit result in bins of gammah
Mean of Pth/Ptda fit result in bins of gammah
Mean of Yh/Yda fit result in bins of gammah

Mean of Pth/Pte fit result in bins of gammah^e
Mean of Pth/Ptda fit result in bins of gammah^e
Mean of Yh/Yda fit result in bins of gammah^e

Mean of Pth/Pte fit result in bins of ptda
Mean of Pth/Ptda fit result in bins of ptda
Mean of Yh/Yda fit result in bins of ptda

(Pth/Pte)_DATA/(Pth/Pte)_MC in bins of gammah
(Pth/Ptda)_DATA/(Pth/Ptda)_MC in bins of gammah
(Yh/Yda)_DATA/(Yh/Yda)_MC in bins of gammah

(Pth/Pte)_DATA/(Pth/Pte)_MC in bins of gammah^e
(Pth/Ptda)_DATA/(Pth/Ptda)_MC in bins of gammah^e
(Yh/Yda)_DATA/(Yh/Yda)_MC in bins of gammah^e

(Pth/Pte)_DATA/(Pth/Pte)_MC in bins of ptda
(Pth/Ptda)_DATA/(Pth/Ptda)_MC in bins of ptda
(Yh/Yda)_DATA/(Yh/Yda)_MC in bins of ptda