summary of efficiencies on ppg measurement
DESCRIPTION
Summary of efficiencies on ppg measurement. TRG, TV, TCA+LIK : single track efficiency found with data efficiencies combined using MC at KINE level TRK : single track efficiency found by MC and corrected with data efficiency combined using MC at KINE level VTX , MTRK : - PowerPoint PPT PresentationTRANSCRIPT
Summary of efficiencies on measurement
Marco Incagli – 23/5/03
1. TRG, TV, TCA+LIK :• single track efficiency found with data• efficiencies combined using MC at KINE level
2. TRK :• single track efficiency found by MC and corrected with data • efficiency combined using MC at KINE level
3. VTX , MTRK :• efficiency from MC, checked with data
4. FILFO :• efficiency from data
5. geometrical acceptances (angular and Pt,Pz cuts):• efficiency from MC
1.1 – Trigger + Trigger Veto
• The single particle TRG (and TV) efficiency is first obtained, then the information are combined using MCP0±(P1) = probability that the ± fires 0(1) trigger sectors
P0R(P1R) = prob. that the Rest of the event fires 0(1) trigger sectors
PTRG = 1 P0P0P0R P1P0P0R (-) (R)
• To evaluate the single particle TRG efficiency events with TV on (with the hardware prescale factor 5) + 1/5th of the events with TV off to be independent from the TV
• TV efficiency is evaluated on triggered events, so what we find is (TRG)•(TV|TRG)
TRG + TV single track efficiency
• To get the single particle efficiency – the track is associated to one or more clusters
– the clusters are associated to trigger (TV) sectors
• The first step is performed by extrapolating the TRK to the ECAL using the newextratom procedure (developed by T.S. and C.G.) and associating to TRK all clusters within a radius R=60cm
• The second step is performed using the CTRG bank
• A check of the dependence of the procedure from the R value has been performed
1 – TR+TV efficiency
• The average efficiency starts to saturate at R60cm
• The systematic error is at the 0.2 % level
• A systematic error which is function of Q2 can be used
TRK efficiency
• Two data samples are selected: : 2 prompt photons with 0 mass + a track which extrapolates
back to IP connected to a cluster which satisfies the pion likelihood
: 1 or 2 ‘prompt’ clusters ; one is associated with a track of p=490±5 MeV which satisfies the pion likelihood
• Some cuts are applied to clean up the second sample. The following categories are selected:– monotracks
– two tracks of the same sign
– proton stars
Monotracks
Monotracks
• They are characterized by a deposit of energy in one or two cells
• Monotracks associated to clusters with 1 or 2 cells are removed from the sample
Good events
monotracks
Two tracks of Same Sign
Two tracks of Same Sign
• They are removed if the minimum distance of LH-FH is larger than 100cm
• (this is done to keep inefficiencies in which the tagging track is broken into two pieces)
Proton Stars
Proton Stars
• They are removed by cutting on the variable QTOT/Ntrk
Closing kinematically the event
• The momentum of the candidate track is evaluated using the -boost from Bhabhas, the photons after imposing the 0 mass and the tagging track extrapolated at IP
• When a vertex exists in the event, then it is possible to check the goodness of the above procedure
• The plots show that the error is symmetric and has MeV
• I take bins of p=25MeV which seems to be safe
Candidate track (TRK2) assignement
• Once the event is selected the momentum components of TRK2 are evaluated (pxe,pye,pze)
• All tracks of the events satisfying the cuts reported in the next transparency are compared with the tagging track
• The track which minimizes the 2 defined below is the candidate track
• A cut at 2 <15 (it was 2 <10) is applied
2
2222
10
)pzepz()pzepz()pxepx(
2
1
2 distribution
Definition of candidate track
A candidate track must satisfy the following cuts:
1. Charge must be opposite wrt tagging track
2. First hit must have cm
3. The point of closest approach (PCA) of backward track extrapolation must have PCAcm and |zPCA|cm
4. 2 condition must be fulfilled
TRK efficiency data() vs MC() – 5slices in btwn 40o and 90o
DATA
MC
DATA/MC
TRK efficiency
• Since the ratio data/MC is rather flat I use the MC track efficiency spectrum correcting it for the following percentage:
(98.59+99.27)/2 = 98.93%
• The systematic error of this procedure is estimated as half the maximum difference btwn the data/MC ratio:
(99.27-98.59)/2=0.34%
TCA+Likelihood eff
• The ratio data/MC is not flat , therefore MC cannot be used to measure TCA. This effect is expected, since TCA efficiency requires a detailed description of hadronic showers at low energy.
• The TCA+LIK efficiency has been obtained by B.Valeriani by tagging the event with the and looking at the and viceversa.
TCA+LIK efficiency
40o<<50o
40o<<50o
50o<<90o
50o<<90o
The single track efficiency isat the level of 98%, except forthe lowest bin which is in the intersection between BARand ECA
The OR of the likelihood hasan efficiency of ~100%, whilefor the AND the correct combination of efficiencies must be done
VTXEFF a module to select events
• VTXEFF– A prompt photon having:
• 29 < L/t < 32 cm/ns
• E>20MeV ; >100cm
– Two tracks with:
• opposite charge
• FH<50cm
• PCA<8cm
• zPCA<7cm
– Track 1 associated with a cluster which satisfies the pion likelihood
• The selected sample has a large bck from , therefore the following cuts are applied:Mass()<110MeV , >160MeV (if a second prompt photon exists)
cos()>0.9 (angle btwn photon and 2 system)
|E|<20MeV
M2 (GeV2)
Num
ber
of e
vent
s
Vertex efficiency - LA
MC
Data
vtx efficiency vs Q2 (GeV2)
vtx efficiency : (data-MC)/MC
• From the comparison data-MC at Large Angle, the systematics error on VTX is of the order of 1-2%
• Note that LA spectrum essentially dies off at Q2=0.4GeV2 (MC), while data have a sizable fraction of
• More data could improve the significance of the comparison
+2%
-2%
Vertex efficiency - SA vs LA
• Small angle events are back to back in and they are on the same side in
• This causes the different VTX efficiency for the two categories
• MC is used for SA eff.; LA used as benchmark
• Systematics ~2%
Track Mass
•The data track mass distribution has been compared with MC summing the signal + the two backgrounds and ;
•The regions above and below Q2=0.5GeV2 have been fitted separately
•.AND. of the likelihood, to suppress Bhabhas
•E(prompt)<10MeV because of
the cut in RPI stream
Before the sum the following corrections have been applied:
In the low Q2 region the following corrections have been applied:
This procedure provides also the fraction of background events wrt signal; this value is used to scale the MC shape and to estimate the number of background events
The effect of the smearing+shifting on the track mass efficiency in the region of interest (>0.35) is at the per mille level
TRKMASS final efficiency
effi
cien
cy
M2(GeV2)
FILFO efficiency
Q2 (GeV2)
• If the QQ shape does not depend upon bclde, then:
)Q()bclde,Q(k)bclde,Q( 22i
2i
bclde (MeV)
Summary of efficiencies
Systematics - preliminary
• TRG + TV : 0.2 %
• TRK : 0.34 %
• VTX : < 2 %
• FILFO : < 1%
• MTRK : 0.2 % (?)
Systematic error dominated by VTX