EM shower reconstruction and located neutrino event analysis

Ciro Pistillo (Bern LHEP)

on behalf of the Swiss OPERA groups

Summary

• Part I:

EM shower reconstruction, e/π separation and energy measurement, gamma conversion search (F.Juget, F Meisel)

• Part II

Analysis of the “NC” event 183545620 (I.Kreslo, C.Pistillo, L Scotto Lavina)

Part I

EM shower reconstruction, e/π separation and energy measurement,

gamma conversion search

(F.Juget, F Meisel)

Algorithm for EMshower reconstruction

• Principle • Collect base-tracks inside a defined cone

(20 mrad opening angle, maximum radius 800 m)

• Used connection criteria to select base-tracks– Slope and position

• To remove background: only keep BT with– tanx and tany <0.4

• Proceed in « back-propagation » method

• Propagate back each BT contained in the cone, keep only the one satisfying the following criteria:– r < 150 microns– < 150 mrad

(criteria optimised to have ~ 5% background)

Back-propagation

Primary track

1 2 3 4

Shower reconstruction

• Real data– Brick exposed to 6 GeV electron beam

Electron 6 GeV (real data)in 20 emulsion films: ~ 3.3 X0

XZ projection

YZ projection

e/ separationand energy measurement

• Using the reconstructed shower– 2 algorithms have been developed based on a

Neural Network method to perform:• e/ separation• shower energy measurement

– The Neural Net. is fed with shower information• Several variables from Longitudinal Profile• 1 variable Number of basetracks• 4 input variables from criteria connections

– Mean and RMS of r distribution– Mean and RMS of distribution

Neural Network Input variables

• Red: Electron • Blue: Pion

• Red: Electron • Blue: Pion

• Red: Electron • Blue: Pion

e/ separation results

• 90% efficiency is obtained at E ≥2 GeV with cont. less 1%

• 80% efficiency is obtained at 1 GeV for 1% cont.

95% efficiency at E ≥ 2 GeV gives ~ 1-2% of cont.

• 90% efficiency at 1 GeV gives ~ 5% cont.

(published in JINST 2 P02001 february 2007)

Shower energy measurementresults

PRELIMINARY

Energy resolution:

)(E40-50 %

~GeV

Vertex and showerAn algorithm has been developed in order to perform gamma conversion search in a large scanned volume.A blind search was performed using data from the event 178969961 located in Salerno (brick 22530)

PRELIMINARY

Gamma showerpointing to the vertex

Some gamma activity was found at ~ 1.7 cm from

a CC event with 6 prongs.

Gamma should com from decay produced at the vertex

Vertex and shower

PRELIMINARY

Impact parameters:

- All tracks have good IP

- Gamma Shower: Dist. From vertex 1.7 cm (13 films)

IP = 42.5 m

The gamma shower is well attached to the vertex

• Good e/separation is obtained (up to 95%)

• Good energy resolution can be obtained

• Data analysis from last OPERA run is under study (gamma conversion search, vertex assignment…)

Conclusion (I)

)(E40-50 %

~ GeVE

Part IIAnalysis of the “NC”

event 183545620located in Bern

(I.Kreslo, C.Pistillo,L.Scotto Lavina)

Event 183545620 NC

Selected brick: 47344 W23,R39,C22 (35.0%)

Other bricks 47340 W23,R39,C21 (22.6%)

47316 W23,R40,C22 (22.1%)

Predictions from SS

6 double base-tracks located in the CS at the Gran Sasso scanning station

VT1: 2 prong

IP = 2.9 m

VT2: 5 prong

<IP> = 52.1 m

The scan-back of the 6 candidates stops immediately after a few plates (low momentum)

DZ= 12.5 mm

Online scanning

GS of a 1x1cm2 area

on 25 plates (34-57)

Two vertices found

c) Angle 0.040 -0.014 (pβ = 0.27 GeV/c) (0.17/0.69 90%CL)

Other 4 short tracks recontructed (low pβ)

PRIMARY VERTEX

a) angle 0.031 0.017 (pβ > 8 GeV/c)

b) angle 0.009 0.006 (low pβ)

a)

b)

c)

momentum measured with the last release of M.Besnier’s algorithm

Event analysis

reconstrucred tracks

RMS X = 2.2 mrad and Y = 2.1 mrad comparable with the microscope resolution (1.8 mrad) the MCS routine says p > 8 GeV/c

Identification of track a)

Residual of base-track angles wrt the track angle

We then consider a sample of 1000 NC DIS simulated events

P > 8 Gev/c

Identification of track a)

P

slope

The number of events where one or more particles with p > 8 GeV/c are present is ~12%

Focusing on p>8 Gev/c and subdividing the samples according to the particle type

According to MC particle a) is likely a pion

In 1000 NC simulated DIS events we found only 28 times a

particle with momentum > 8 GeV interacting inside the brick

Adding the condition that particle a) interacts inside the brick only 28/1000 events survive (29 particles)

Summarizing our doubts…• We see a high momentum track from the neutrino interaction vertex interacting inside the brick 12 mm downstream. Within 1000 NC DIS events such a behavior (p>8 GeV + secondary interaction) occurs only 28 times (~3%).

• According to MC, particle a) is a hadron, most likely a pion but... The secondary vertex is reconstructed with <IP> ~ 52 m. It shows a small opening angle and the 5 daughter tracks have low momentum.

• To our best knowledge track a) resembles to an electron showering after a few plates… could be then the interpretation of the event as eCC possible?

• The CNGS beam has a e (+antineutrinos) contamination of ~1%

A shower is reconstructed

Response of the NN: electron (100%)

Estimated energy: 62 GeV

Associated base-tracks: 80

gamma conversion?

Application of the shower reconstruction algorithm to our event

e/e CC hypothesis NC hypothesis

Beam 0.0087 1

/ CC 1 0.31

Cut on momentum and interaction inside the brick

P>8 P>10 P>12 P>8 P>10 P>12

0.71 0.66 0.63 0.028 0.015 0.010

Total (10-3) 6.2 5.7 5.5 8.7 4.7 3.1

Not included: cross section for QE+RES,DIS

the particle a) is an electron

P(e/eCC) / P(NC)

p > 8 GeV/c 0.71

p > 10 GeV/c 1.23

p > 12 GeV/c 1.77

Event 183545620 shows an interesting topology. The interpretation as NC doesn’t seem to be the only possible. The probability of a eCC interaction is indeed of the same order of magnitude. We are going to perform a more analitic study. Further investigation of the primary vertex will be exploited as well.

Conclusion (II)