Minutes of the EWK Group Meeting Thursday,
15 June 2006, 4-6 PM, CDF Pump Room
(the minutes also include questions which conveners
were not able to answer for themselves going through the
presentations after the meeting and felt important)
======================================================================
Jiyeon Han : dsigmaZ/dy (PRE-BLESSING)
(1) Please incorporate the standard 1.019 luminosity scale to the data base
luminosity. Also, please update to the full 1.1/fb by recovering the
missing data files.
: done
updated result with 1fb^-1 data
(2) We recommend using the v13 run list. The v10 list only goes up to September
of 2005 (run 203799). What do you do about runs 203800-212133?
:We are using v13 good run list. We'll notify this in CDF Note.
(3) Do you have the full zewkad sample? The MC corresonding to the bhel0i data
was only recently completed. Also, you call the -1/6 X0 copper sample the
'default' sample -- can you explain which sample is used for what (zewkad
vs 'default')? In particular, what is used for acceptances, efficiencies, etc.
: We checked the run number list with zewkad sample.
The run number is up to 212133 which is the last run number of goorun v13 list(1001).
For material study, the default sample is different from zewkad sample.
Jedong generated three different mc sample with different material component.
And we set one mc sample set with standard central and -1/6X0 plug material
as a default sample. Then, we compared the acceptance between default and extra
material sample.
(4) Assuming you use zewk6d and zewkad for your acceptances, do you appropriately
reweight the samples according to luminosity?
: No, we didn't consider the luminosity weighting. But, zewk6d sample size is 3.2M
and zewkad sample is 6.9M events(6.9/3.2 = 2.1). And gen5 data sample size is 371.5pb^-1 and gen 6 data
sample size is 726.2pb^-1(726.2/371.5 = 1.95). So, the data size difference in mc sample covers the luminosity
weighting.
(5) Can you show that there is no eta dependence for the central electron trigger?
: The following plot shows the trigger efficiency of the central leg in eta after applying Et>20 GeV cut.
Trigger efficiency in eta of central electron.
(6) Can you plot the ET of plug electrons separately for CP and PP events, to
demonstrate that the structure in Figure 4 is due to kinematics?
: The following plot shows the Et distribution of electron from Zcc, Zcp and Zpp each region
Et distribution after energy scale
(7) Can you separate the energy scale into east and west for CP and PP, to deterimine
if separate scales should be applied?
:
(8) What is the effect of applying a smearing correction to the MC to match the
data resolution?
:
(9) Do you incorporate a systematic uncertainty due to energy scale and resolution?
If not, can you determine how large the uncertainty would be?
:
(10) For the acceptance calculation, you apply the generated z and mass cuts only
in the denominator, correct? The numerator cuts use only reconstructed quantities
for these?
: Yes, we only apply the generated z and mass cut in the denominator
and only applied the reconstructed quantities for the numerator. So, the generated z and mass
cut isn't applied in the numerator.
(11) Electron Efficiencies:
(i) Can you produce MC efficiencies for central & plug electrons, and for
CC, CP, PP Z's, using the same methods as data? In particular, tables
10 and 11 for MC. Can you separate the data and MC efficiencies into
0d and 0h+0i? If the data/MC ratios are different in the two data sets,
you may have to include that correction in your sample reweighting.
:
electron efficiency with MC sample
------------------------------------------------------------------------------------------------------------
el efficiency total east west scale factor for total(data/MC)
------------------------------------------------------------------------------------------------------------
tight el 0.84016+-0.00047 0.83972+-0.00067 0.84048+-0.00067 0.98964+-0.00205
loose el 0.95965+-0.00025 0.95927+-0.00036 0.96003+-0.00036 1.00177+-0.00093
plug with Zcp 0.87360+-0.00040 0.87411+-0.00057 0.87309+-0.00057 0.96300+-0.00201
plug with Zpp 0.86819+-0.00079 0.86704+-0.00111 0.86939+-0.00113 0.94259+-0.00381
------------------------------------------------------------------------------------------------------------
Z rec efficiency
------------------------------------------------------------------------------------------------------------
Zcc 0.90664+-0.00054 0.90591+-0.00077 0.90736+-0.00077
Zcp 0.73392+-0.00053 0.73401+-0.00076 0.73382+-0.00075
Zpp 0.75375+-0.00138 0.75175+-0.00193 0.75583+-0.00196
------------------------------------------------------------------------------------------------------------
(ii) Please incorporate systematic uncertainties from background into your
efficiency measurements. How do you get the uncertainties and why is
is the PP Z efficiency so large?
: Yes, we'll include the uncertainties to the efficiency measurements.
And for the uncertainties for Z(PP), it was mistypo. I corrected this in the cdfnote.
(iii) What would the background in the efficiency measurement be if you used a
same-sign method to estimate it for CC?
: with 1fb data sample,
same-sign method : tt bkg = 0.00236+-0.00033, ttl bkg = 0.01670+-0.00156, ttb bkg = 0.06759+-0.00269
iso method : tt bkg = 0.00223+-0.00081, ttl bkg = 0.01250+-0.00195, ttb bkg = 0.07891+-0.00327
-----------------------------------------------------------------------------------
with ss method with iso method
-----------------------------------------------------------------------------------
tight el efficiency 0.84027+-0.00163 0.83146+-0.00165
loose el efficiency 0.97100+-0.00075 0.96135+-0.00085
-----------------------------------------------------------------------------------
(iv) A possible check on your background estimate would be to repeat your
efficiency measurement with a tighter mass window, and compare the
resulting data/MC ratio to the wider window.
The ratio of data and MC with these two different mass window is consistent each other :
-------------------------------------------------------------------------------------
scale factor(data/MC) 66 < M < 116 82 < M < 102
-------------------------------------------------------------------------------------
tight el 0.98964+-0.00205 0.99114+-0.00210
loose el 1.00177+-0.00093 1.00151+-0.00091
plug with Zcp 0.96300+-0.00201 0.96623+-0.00207
plug with Zpp 0.94259+-0.00381 0.95712+-0.00388
-------------------------------------------------------------------------------------
(v) Which plug efficiency do you use in equations 10 and 11?
: For equation 10(efficiency of Zcp), we used the plug efficiency with using Zcp evnets.
And for equation 11(efficiency of Zpp), we used the plug efficiency with using Zpp events.
(vi) Can you show the eta dependence of the efficiencies in data?
: We assumed the background is flat in eta.
For plug electron efficiency, we need to measure the eta dependence background.
the eta dependence of efficiencies
: Top(tight electron efficiency), middle(loose electron efficiency), bottom(plug electron efficiency)
(12) How much does your phoenix Z efficiency change if you measure the pheonix electron
efficiency as a function of detector eta rather than physics eta? Similarly,
do the trigger efficiency curves change at all if you use detector eta bins rather
than physics eta bins?
: eta bin for the phoenix electron efficiency or trigger efficiency is
Pes eta of the plug electron. So, it's deteta, not physics eta.
(13) Can you separate your plug and phoenix efficiencies into east and west?
: We already applied the plug electron efficiency on east and west
separately. And also the phoenix efficiency is applied as a function of eta.
So, I think the plug and phoenix efficiencies are applied on the east and west separately.
(14) Please investigate the discrepancy in dsigmaZ/dy from PP compared to CP in
|eta| 1 to 1.5.
:
(15) Backgrounds:
(i) How would your result change if you used MC for your signal shape?
: MC shape didn't discribe the data, specially the low isolation region(iso: 2~6 GeV).
So, the fitting with signal shape from MC didn't work well.
background rate
------------------------------------------------------------------------
with data for signal with mc for signal
------------------------------------------------------------------------
Zcc 0.00380+-0.00078 0.00547+-0.00083
Zcp(cen) 0.00652+-0.00082 0.02000+-0.00084
Zcp(plg) 0.00832+-0.00143 0.03168+-0.00200
Zpp(pnx) 0.02623+-0.00223 0.05296+-0.00322
Zpp(plg) 0.02050+-0.00234 0.03824+-0.00312
------------------------------------------------------------------------
comparison of the signal isolation distribution : Red: data, Blue: MC (mass window : 66~116 Gev)
comparison of the signal isolation distribution : Red: data, Blue: MC (mass window : 82~102 GeV)
comparison of the signal shape
central leg of Zcc
central leg of Zcp
plug leg of Zcp
pnx leg of Zpp
plug leg of Zpp
signal shape from mc : central leg of Zcc
signal shape from mc : central leg of Zcp
signal shape from mc : plug leg of Zcp
signal shape from mc : pnx leg of Zpp
signal shape from mc : plug leg of Zpp
(ii) Does your signal shape change if you apply a MET < 20 cut?
:The signal shape doesn't change with applying MET < 20 GeV cut.
signal shape : Black: no met cut, Red: Met < 20 GeV.
signal shape normalized by area : Black: no met cut, Red: Met < 20 GeV.
(iii) Can you use a background template from [tight electron (minus iso) + jet]
events, rather than [tight electron (minus iso) + loose electrons
(plus anti-cuts)] events? For this you could perhaps use the inclusive
electron triggers (at least for the central).
:
(iv) How would the plots in Figure 10 change if you corrected for Z contamination?
: The blue solid line is the isolation shape after subtracting electron from W and Z.
background shape
: black: no correction, red: after subtracting electron from W+jet/gamma, blue: after subtractig electron from W+jet/gamma and Z.
(v) For the W+jet/gamma correction to the background template, how would that
change if you used a dedicated W gamma MC sample for that part of the
correction?
:
(vi) How do you determine the systematic uncertainties that are shown in Figure 19?
The dashed lines represent the uncertainty bands?
: Yes, the dashed line represents the statistics uncertainties band.
For the systematic uncertainties, we took the biggest difference value from the standard background
value(black solid line) which is the background rate without any change on the parameter.
(vii) Can you do your fits in a few eta bins to compare to the distribution
in Figure 20?
:
(viii) Can you test your PP background prediction using opposite-side PP events?
: Well, the opposite-side PP events locate only around rapidity zero.
So, it might be hard to compare the background prediction with using opposite-side PP events.
(ix) For your final fits, please correct for W+jet/gamma contribution.
:
(16) Do you understand why there is a different +/- rapidity asymmetry in the
background distribution for CP than for PP (Figure 20)?
:
(17) How significant are the largest variations in the material uncertainty plot
(Figure 21(b))? It seems these could be due to statistical fluctuations --
if that is the case, then it would perhaps be better to combine several bins
and use the same uncertainty for all of the combined bins.
: The largest variation is 1.11 pb in rapidity=0.75 ( cross section in this bin is 64.76+-1.39(stat.) ).
The RMS in Y axis of Figure 21(b) for the sum of both extra material is 1.73181.
And this doesn't seem to be due to the statistical fluctuations. The following plot is the acceptance with default sample.
The error shows the statistical error. But, the statistics is pretty enough in the all rapidity region.
the acceptance distribution in rapidity
(18) Can you produce a single table or plot showing all systematic uncertainties?
:
the table for all systematic uncertainties
(19) Can you normalize the + and - rapidity distributions to each other and overlay
them? Can you do a KS test of the two?
:
(20) Can you produce inclusive cross sections separately for 0d and 0h+0i? We
recommend comparing your result to other results (e.g., the WW cross section
result).
: working on it
