1. Status of Pixel Offline Software
ID week CERN, October 28th 2013
Status of Pixel Offline Software
F. Djama, Ph. Grenier, S. Miglioranzi

2. Introduction LS1 activities
Be ready for commissioning and data taking with IBL, nSQP and DBM.
Consolidate and improve present software.
Include radiation effects in simulation and reconstruction.
Outline
Simulation:
Geometry
Digitization
Reconstruction
Clusterization
Performance
Tracking
B-tagging
Software

3. Geometry
S. Elles
Last months dedicated efforts to deliver a more realistic updated geometry description for the detector and services.
New geometry tag (ATLAS-IBL(3D25) ) includes:
last IBL geometry updates.
IPT, IST, ISIS (Inner Positioning and Support tubes).
nSQP (new Service Quarter Panel).
new beam pipe.
DBM (Diamond Beam Monitor).
current geometry is very close to the final one.
Some mechanical parts description might slightly change
(endblocks, stave rings, ..).
Short term plans:
Validate in the next days this new description.
 freeze it for DC14
ISIS (IST support)
DBM
IST
nSQP

4. Geometry
S. Elles
New material budget estimations for the latest geometry available:
detector-only (as in TDR version)
including also Beam Pipe + structures + services

5. Digitization
S. Miglioranzi, M. Klein, M. Rudolph, F. Djama
IBL requires different digitization models (sensors technology)
New PixelDigitization pkg in place:
Code has been optimized.
Different classes per charge model (SiPl, Si3D,DBM).
DBM model exist need to be imported in head tag.
Extension of DB and Services to IBL:
SW already delivered.
Need to prepare last IBL-aware DB folders.
DBM not included yet  needs latest geo tags.
W.-M. Yao

6. Digitization of 3D Sensors at high incidence
S. Terzo
First data/MC comparison for high incidence on 3D sensors.
120 GeV pion at an incidence of θ = 80 (η =2.4 in ATLAS).
Digitization: inductive model.
Agreement should be better if we check incidence in real data: perform the sensor alignment.
Model already in the IBL cache version of the ATLAS software

7. Digitization with Irradiation Effects
C. Bertsche, M. Bomben, Ph. Grenier, G. Piacquadio
A lot of progress in implementing charge models for planar sensors, testing and debuging:
Electric field and Ramo potential mapping from TCAD.
TCAD reproduces well the chosen charge model (V. Chiochia et al., Nuclear Science, IEEE Transactions on , vol.52, no.4, pp , Aug. 2005)
Charge models validated by old results up to neq/ cm2 .
Holes become important in case of trapping !
Next step: validate charge models after irradiation with test beams of irradiated sensors.
A similar activity started for 3D sensors
M. Povoli

8. Ramo potential: TCAD simulations
Ramo potential: one slice of a 3D simulation (X-Z plane)
3D simulation of ATLAS pixels: ¼ of a 3x3 matrix
Ramo potential: profile along the bulk
M. Bomben - TCAD simulations of planar sensors - ATLAS Pixel week - CERN, 13/06/2013

9. NN-Clusterization for IBL
F. De Lorenzi
NN clustering is default pixel clustering algorithm  need to be re-trained for IBL.
In dense cores of high-pt jets, charges deposited by multiple particles can merge into a single cluster  degraded cluster resolution.
NN clustering splits merged clusters and computes positions and errors of subclusters.
NN training exercise of October 2012 failed (performances not as expected).
Some bugs identified in training code, old MC samples had geo and track problems
New training has now started (100K JZ4W from latest IBL+ATLAS prod used)  first results are promising…
Histograms contain hits from all 4 pix barrel layers; Residual: NN-computed position - G4 position

10. Analog Clusterization for IBL
M. Klein
What is used at the moment when running with IBL-geometry is digital clustering.
Need to make the code more flexible to steer different clusterization algorithms for different layers.
Develop an analog clustering for IBL.
Possible optimization for the future:
use NN for cluster splitting.
analog position for the subclusters.
Work started few days ago…

11. Tracking Performance with IBL
W.-M. Yao, M. Kwok, Y. Wu, Y. Fang, M. Garcia-Sciveres
Tracking performances vs PU (40,60,80,100).
At high pileup loss of efficiency & fakes increasing rate have been observed.
Track fake rates vs pt
LF track eff vs pt
LF track eff vs η

12. Tracking Performance with IBL
W.-M. Yao, M. Kwok, Y. Wu, Y. Fang, M. Garcia-Sciveres
Effects of dead modules on tracking quantities have also been studied:
Disabling 5,10,15% of modules (including IBL)
The degradation due to disabled modules up to 15% is modest at few % level

13. B-tagging with IBL Schorlemmer M. Kwok B. Di Girolamo W.-M. Yao
Y. Fang
M. Garcia-Sciveres
Schorlemmer
B. Di Girolamo
G. Piacquadio
S. Pires
B-tagging with IBL
0 IBL disabled Pix modules (with repairs)
5 IBL disabled Pix modules (with repairs)
5 IBL disabled Pix modules (no repairs)
5 IBL disabled Pix modules + readout ineff (no repairs)
IBL 14TeV
stdATLAS 14TeV
WH-> l nu bbar
MV1
μ=40
ttbar
IP3D+SV1
High Pt performance will improve with NN-clustering.
A note in preparation:
ttbar samples, PU = 0, 60.
Realistic dead module scenario (after nSQP connection at SR1).
Optimization with b-tag performance group started.
ttbar
IP3D+SV1

14. Other Topics L. Franconi J. Black, T. Nelson W.-M. Yao
Connectivity and ByteStream coding for IBL :
Realistic IBL Cabling Map  joint effort of detector/DAQ/DCS group.
Proposal of FE-I4 ByteStream Cnv (P. Morettini) currently coded (with help of S. Roe).
Workaround for HLT and FTK: Dummy cabling Map (F. Djama) and FE-I3 inspired ByteStream Cnv (D. Emilyanov).
Luminosity-induced saturation effects :
At high lumi, hits will be lost due to buffer overflows and busy electronics.
Simulating FE-I3 readout process:
Bunches scheme.
Realistic ToT.
Goal: an inefficiency table (eta, lumi) to be used by digitization.
FE-I4 model also needed.
Next step: simulate MCCs.
Condition Data Base :
Folders and Services coded.
Ready to be implemented in a release.
L. Franconi
J. Black, T. Nelson
W.-M. Yao

15. Releases and core sw: S. Miglioranzi DC14 releases plans: step
IBL developments and validation studies are currently performed in rel 17.3.X(.Y).
Validated caches used for IBL+ATLAS MC Production so far : Z, IBLProd.
IBL+ATLAS MC Prod twiki
Last validated cache was (contains Supercell info).
Plan to release a full build in the next days (this week?) including new geo-related sw.
will be “in sync” with :
 differs for: 64bit, supercell and TrigMC tags.
DC14 releases plans:
Atlas Prod
18.1.0 T0
upgrade
step
Releases
evgen
Y (?)
simulation Y
digitization + reconstruction Y
Atlas Prod

16. Software Improvements
S. Miglioranzi
Large and important activity to improve our software:
Identifiers
Hard coded numbers
Prepare easy ITK implementation
Code cleaning …
Coordinated by S. Miglioranzi.
PixelDigitization package: M. Klein
Geometry: S. Elles
Conditions: W.-M. Yao
We receive valuable help from J. Chapman, N. Styles, R.D. Schaffer…

17. Summary
2012/2013 have been crucial years for the development and finalization of the IBL software  but still tight schedule ahead !
Current geometry simulation is realistic:
Preliminary validation is ongoing and PhysVal should start soon.
Plans to use it for DC14.
Only minor changes foreseen in the future.
Digitization is in place:
Digitization models available for all the different technologies.
Reconstruction:
Existing reconstruction software robust  already working for IBL.
Optimization can be done and it has already started.
Clusterization is a hot topic  more people involved on that now.
Involvement of other CP groups has started  vital to keep close. connection and feedback to get the best performance out of IBL.

18. Schedule Forseen schedule in april 2013:
Hopefully only few weeks delay…

19. Backup

20. Geometry – DBM & DBM services
Included in the new ATLAS-IBL(3D25)-04-yy-zz tags
ISIS (IST support)
& IST
nSQP
Tuesday, October 15

22. Geometry – Run1 Pixel Detector
Pixel current geometry description: cooling fluid modeling:
Studies of hadronic interactions in cooling fluid.
In pixels show large data/MC discrepancy
 better modeling needed.
Data MC
hadronic interactions in cooling vs Z
PP0 region required optimization/improvement:
Inner PP0 material overestimated in current
GeoModel  2.3<|η|<2.5
SCT extension eff studies sensitive to PP0 region
New improved description by looking at construction
files and directly to the dismounted SQP in SR1
SCT extension efficiency
ATLAS-GEO
Tuesday, October 15

23. Tuesday, October 15

24. Tracking Activities ID Alignment:
Code developments/adaptation to include IBL already started.
Alignment of the IBL/Pixels during cosmic runs.
Further understanding of Silicon deformation.
Improving run-by-run alignment, reduce track parameters systematics…
Vertex Reconstruction:
Improve the power of the vertex reconstruction as in-time pile-up estimator in the next run.
Work is ongoing on the seeding algorithm.
New Tracking +IBL re-optimization is starting now…