PrimEx p0 radiative width extraction
Eric Clinton
University of Massachusetts Amherst
July 19, 2007
1
Outline




Data Source and cuts
Event selection
Hybrid Mass Signal enhancement
Yields




Yields over entire HyCal acceptance presented for
information on Incoherent photo-pion production only
Systematic effects from yield extraction
Simulation
Results

Sytematic Error Analysis
2
Data Source and Cuts


mysql -h primexdb -u
primex_user book_keeping b --execute="select run
from run_list where
radiator='A' and
target='carbon' and
type='pi0' and
production='good';" >
run_list.example
mysql -h primexdb -u
primex_user book_keeping b --execute="select run
from run_list where
radiator='B' and
target='carbon' and
type='pi0' and
production='good';" >
run_list.example












1.) Two or more clusters/event.
2.) Minimum three (3) (PbWO4 or lead
glass) detectors to define a “cluster”.
3.) 50 MeV or greater central (PbWO4 or
lead glass) crystal detector energy in
cluster.
4.) 10 MeV or greater minimum
deposited energy in (PbWO4 or lead
glass) detector.
5.) Max cluster energy 8 GeV.
6.) gg invariant mass greater than 0.085
GeV in at least one of the cluster pairs.
7.) Elasticity (cluster pair energy
sum/tagger energy) greater than 0.70.
8.) Cluster energy greater than 0.5 GeV.
9.) Cluster X or Y position must be
greater than 4.1 cm.
10.) Cluster pair energy sum between 3.5
and 6.5 GeV -- additional software cut
not imposed on the skim, but imposed
later:
11.) Timing cut of -15 ns to +5ns.
“pi0gains” used as calibration
3
Event selection
Eliminate Tagger and HyCal combinatorics

Likelihood

Event entries have invariant
mass, elasticity, and timing



Fit invariant mass, elasticity,
timing signal and background

Fitted signal lineshape as
probability density function
(PDF)




Which entry to choose in a
mutli-entry event?
Which is "most likely"?
Evaluate the PDF for each
parameter for each entry.
Three individual likelihoods.
PDFInvariant mass, PDFElasticity,
PDFTiming
Total likelihood = PDFInvariant
mass × PDFElasticity ×PDFTiming
Entry with highest total
likelihood "wins".
4
Misidentification –
any systematics?
No. MisID is random, and event selection
tends to pick smaller production angle pions.
5
Rotation of 2-D data onto 1-D
Try to enhance signal to noise

Original 2-D data

Elasticity vs.Invariant Mass

New 1-D signal

AKA “Hybrid Mass”
6
Additional “Diagonal”
Data Cut
Warning—departure from analysis note
7
Apply Additional Cut and Veto
Warning—departure from analysis note

Result



Greatly
improved
signal to noise
Removes 3rd
order curvature
from
background
Requires well
understood
veto

Veto
systematic
error small in
comparison to
fit error and
other
systematic
effect
improvements
8
Plateau Elastic Pion Yields
Additional minimization of signal to noise

Timing


Integration Range


Elastic p0 as a function of the timing cut
Elastic p0 as a function of the integration range
Fitting Range

(left, below)
(right, below)
Elastic p0 as a function of the fitting range
9
Original timing cut/data source
Timing cut vs. pion yield plateau
Timing cut set to ±5 ns
10
Integration range plateau
11
Fitting range plateau
12
Selected Hybrid Mass Fits
13
p0 yields as a
function of
production angle.
These yields are extracted
from a data set where the
“diagonal” and veto cuts are
applied.
Final radiative width MUST
correct for veto
Photon Misidentification.
14
Yield extraction for various signal
and background models
15
Simulation Work

Thrown with E-Channel Photon flux weighting


Energy correction added



Proper shower development
Resolution and centroid tuned


Energy lost out back of HyCal, out of cluster mask
Added back about 10% of energy
Tracking threshold tuned


Primakoff (with FSI), Coherent (Cornell with FSI), Incoherent
(Glauber)
Get invariant mass right to proper mock physics
Vet the Simulation

Push 4 vectors from experiment thru sim


See how p0 candidate spectrum look, look for losses
Turn off detectors, see how acceptance behaves
16
Putting physics
events thru the
Simulation
Around 99.2% fidelity
17
Turning off glass detectors
HyCal Tungstate Acceptance Only
18
Efficiencies
as a function of the photo-pion process, HyCal Tungstate acceptance
19
Geometric efficiency
and reconstruction
(cut) efficiency
HyCal tungstate
acceptance
Turning off cluster energy
and invariant mass cuts
20
Fit to Data, and Extracted Width
HyCal Tungstate Acceptance
Extracted width – 8.166 eV ± 0.133 eV (1.63%)
21
Acceptance Corrected Cross Sections
HyCal Tungstate Acceptance
22
Systematic error sources?

Extracted yields
over the entire
pion angle range
must be stable as
these parameters
are varied.
23
Systematic Effects from Yield Extraction
HyCal Tungstate acceptance

Nominal
8.166
NA

Cluster Position Finding Method




Method 0:
Method 1:
Method 2:
Method 4:
8.044
8.202
8.156
8.195
-1.50
+0.43
-0.13
+0.35
Fit Range (nominal = ±0.029 HMU’s)


* (+)
* (-)





Lineshape (degrees of freedom)***



DG3Po:
TG2Po:
8.173
8.188
+0.09
+0.26

Integration range (nom. = ±0.013 HMU’s)







±
±
±
±
±
±
±
0.010
0.011
0.012
0.013
0.014
0.016
0.018

8.102
8.155
8.148
8.166
8.170
8.206
8.242

* (-)
* (+)
-0.79 %
-0.13 %
-0.21 %
NA
+0.04 %
+0.48 %
+0.923 %


±
±
±
±
±
±
±
±
±
0.026
0.027
0.028
0.029
0.030
0.031
0.032
0.033
0.034
8.105
8.123
8.152
8.166
8.192
8.130
8.141
8.120
8.134
-0.74 %
-0.52 % * (-)
-0.18 %
NA
+0.31% * (+)
-0.44 %
-0.31 %
-0.57 %
-0.39 %
Total Error will be “asymmetric” since
many of “systematic effects” tend to go
in only one direction.


All positive contribution will be added in
quadrature for the total “positive
systematic error”
Vice versa for the negative contributions
No systemtatic effect will be claimed. Fit errors go
up faster than any shifts above, and yield plateau
for smaller pion angles is present.
***Nominal = Double gaussians with 2nd order polynominal
DG3Po = Double gaussians with 3rd order polynomial
TG2Po = Triple gaussians with 2rd order polynominal
24
Error Accounting
HyCal Tungstate acceptance
Source
% Error
Error from Fit
+1.63 %, -1.63 %
Photon flux
+1.10 %, -1.10 %
Cluster Position Reconstruction
+0.43 %, -0.13 %
Fitting Range
+0.31 %, -0.52 %
Signal Lineshape
+0.26 %, -0.00 %
Background Model
+0.08 %, -0.00 %
Dalitz Decay
+0.03 %, -0.03 %
Target Thickness
+0.04 %, -0.04 %
Veto Counter Inefficiency (if used)
+0.06 %, -0.06 %
Total (w/ Veto, w/out Error from Fit)
+1.25 %, - 1.23 %
Total (with Veto)
+ 2.06 %, -2.04 %
25
Result



HyCal Tungstate Acceptance
Gp0 = 8.166 eV ± 0.133 eV +0.102 eV – 0.100 eV
Gp0 = 8.166 eV ± 1.63 % +1.25% - 1.23%
26
Future work

Work another nuclear
incoherent generator




Latest from Tulio in hand
Evolve cross sections to
the weighted mean
photon energy
“Conjoined” analysis
Lead Target Data?
27
Extra slides
28
The Veto—how it changes the
angular spectrums
29
Extracting a Photon Misidentification Efficiency
PME = 0.80 ± 0.057% (HYCALCLUSTER veto flag == 4)
PME = 2.20 ± 0.16% (HYCALCLUSTER veto flag == 3)
PME = 2.80 ± 0.21% (HYCALCLUSTER veto flag == 2)
PME = 3.20 ± 0.23% (HYCALCLUSTER veto flag == 1) ** ongoing work
30
Photon flux
31
Poor Elasticity
32
Energy Correction
Across entire HyCal acceptance
33
Tracking Threshold, resolution, and centroid tuning
34
Turning off glass detectors
Entire HyCal Acceptance
35
Turning off tungstate detectors
Entire HyCal Acceptance
36
Efficiencies
as a function of the photo-pion process, entire HyCal acceptance
37
Geometric efficiency
and reconstruction
(cut) efficiency.
Entire HyCal
Acceptance
38