1. CLAS12 Analysis Readiness
R. De Vita
for the CLAS Collaboration
Jefferson Lab Software Review
November 10, 2016

2. CLAS12 Analysis Readiness
Outline
Reconstruction & Calibration validation using pseudo-data
GEMC: Realistic detector simulations
Reconstruction validation and usage
Calibration development and testing
Data analysis workflow & preparation:
Analysis workflow
Event generators and background simulations
Physics analysis example
Analysis organization
Timeline
ADD ROADMAP FOR THE CHARGE
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

3. Realistic Detector Simulations
Digitization Routines
Geometry
CCDB
Scintillators attenuation lengths
Drift Chamber cell inefficiencies, Residuals
Resolution parameters
Same constants as real data: actual calibration constants
Realistic geometry description
Active and passive materials
CAD drawing conversion for most complicated elements
Same geometry for reconstruction
FTOF response to MIPs in GEMC
Counts
MAKE PLOT DARKER
Expected position from real data
FTOF in GEMC
ADC Counts
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

4. CLAS12 Analysis Readiness
Event Reconstruction
π0 reconstruction from forward EM calorimeters
Two-photon invariant mass
Two-photon opening angle
M2π0 = 2Eγ1Eγ2[1− cosθ12]
Charged particle identification from TOF detectors
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

5. Calibration & Monitoring
Development of calibration and monitoring applications in an advanced stage for both baseline and ancillary CLAS12 subsystems:
CVT monitoring GUI
Calibration and monitoring software is based on COATJAVA
Algorithm development supervised by the CLAS12 calibration & commissioning group (CALCOM)
Implementation supervised by the software group
Tests on both cosmic ray and simulated data
Preparations for first Calibration Challenge (Dec. 2016) in progress
FTOF calibration GUI
Status:
Now: ready for Feb 2017 commissioning run
June 2017: ready for physics in Fall 2017 engineering run
EC/PCAL calibration GUI
EC-Pcal (UVA/Jlab, Ohio)
FTOF (Glasgow, Iowa, MSU, JLab)
LTCC (Temple, JLab)
DC (Mississippi, ISU, Julich, JLab)
HTCC (FIU, UConn, JLab)
MM (CEA-Saclay, JLab)
SVT (Jlab, Temple)
CTOF (Glasgow, Jlab)
CND (Orsay, Glasgow)
FT (INFN, Edinburgh)
RICH (INFN, Duquesne)
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

6. CLAS12 Analysis Readiness
EC/PCAL Calibration
Attenuation (cm)
Gain
Validation of MIP Calibration Algorithm U W V
EC Inner
Fits to simulated data used to evaluate cuts and thresholds.
Estimate time needed to accumulate sufficient statistics.
Estimate accuracy of gain and attenuation extraction.
Simulated Attenuation
Fit to GEMC simulation
Expected MIP
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

7. Alignment of the SVT using Millepede
Misalignment results obtained from Millepede
tested on MC
Reconstruction of cosmic events with millepede misalignments incorporated
Type 1 track
shift barrel R1 by 20 microns
MC validations
Real Data validations
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

8. Calibration Challenge
Test of the full calibration procedure:
Generate pseudo-data with “wrong” calibration constants
Run calibrations for all systems in appropriate sequence
Extract calibration constants and save them to DB
Evaluate calibration quality by
looking at monitoring plots
comparing reconstruction output with extracted and original constants
Who:
Analysis coordinator
Calibrator team
DB manager
“Chef” for data processing
When:
December (1 week time)
How:
Generate pseudo data with Pythia and full luminosity background
1 shift (8 h) worth of data
Daily meetings and milestones for coordination and progress tracking
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

9. CLAS12 Analysis Readiness
Data Analysis Scheme
CCDB
RAW DATA
Simulation
Calibration
MC DATA
RECONSTRUCTED DATA
Reconstruction
DSTs
DSTs
Event selection
CHECK COLORS
COATJAVA/CLARA
Physics analysis
PHYSICS OBSERVABLE
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

10. Event Selection Tool CLAS
Collection of algorithms and procedures to select events from reconstruction output and provide DSTs for specific final states:
Select golden runs and files
Select events for specific final state
Apply:
kinematic corrections
fiducial cuts …
Output DST files with:
fully corrected 4-vectors for physics analysis
detector related info for refinement of PID and signal selection
luminosity and helicity related info
Implement file tagging for easy data handling and distribution inspired by the CLAS data mining project
CLAS/CLARA Data Mining
CLAS
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

11. Physics Event Generators
~1min at 1035cm-2sec-1
PYTHIA
Generated events available for calibration and reconstruction tests, and physics studies:
INCLUSIVE ep ➝ e’X generator
SIDIS LUND MC (PYTHIA and PEPSI)
Generating (claspyth) low Q2 events for hadronic background and PID studies using modified PYTHIA
Generate (clasDIS) single and double-spin dependent processes using the modified PEPSI (LEPTO)
Exclusive events
exclusive g (DVCS), p/h using GPD models
exclusive eKL, eppp,..
phase space for multiparticle final states
inclusive e-
clasDIS
P(GeV)
p in DVCS
p- in eLX
GET MATERIAL FROM HARUT
q(deg)
L S
MX(epp-X)
M2X(eKX)
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

12. Simulating EM Background
Can be set with gcards:
<option name="LUMI_EVENT" value="124000, 250*ns, 4*ns" />
<option name="LUMI_P" value="e-, 11*GeV, 0*deg, 0*deg" />
<option name="LUMI_V" value="(0.,0.,-4.5)cm" />
<option name="LUMI_SPREAD_V" value="(0.01, 0.01)cm" />
124K e-/event = 1035 s-1cm-2 on 5cm LH target
Start beam before the target, 100 micron wide
DVCS event with no EM background
DVCS event with EM background at full CLAS12 luminosity
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

13. High Level Analysis Tools Radiative Corrections
Needed to account for distortion of measured observables (cross section and asymmetries) induced by radiation:
distortion of observable magnitude
modulation of angular distributions
Use and extension of algorithms and codes utilized for 6 GeV program:
RadGen for inclusive/DIS processes
ExcluRad for single meson production
Dvcsgen for DVCS
HapRad and SidisGen for SIDIS …
Dedicated Workshop at JLab in May 2016:
Set of precision calculations of radiative corrections for different processes including the full set differential cross section
Development of generators including the radiative photon emission
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

14. Physics Analysis Example Deeply Virtual Compton Scattering
Accessing Generalized Parton Distributions amplitudes
Exercise of the physics analysis chain, from event generation to physics observable extraction
GET NEW SLIDE FROM LATIFA
Missing Mass cut
Angular cut
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

15. Analysis Organization
CLAS Coordinating Committee
Physics Working Groups
Analysis Review Committee
Collaboration Governance
Calibration & Commissioning Group
Contact Person
Analysis Coordinator
Software & ReconstructionGroup
Calibration Team
Analysis Team
Experts
First Experiment
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

16. CLAS12 Analysis Readiness
Timeline
TASK
2016
2017
Oct.
Nov.
Dec.
Jan.
Feb.
Mar.
Apr.
May
June
July
Aug.
Sept
SIMULATION
RECONSTRUCTION
CALIBRATION
ANALYSIS
EVENT SELECTION TOOLS
PHYSICS ANALYSIS TOOLS
EXP. SCHEDULE
New release ready for physics studies
New release ready for physics studies
Ready for KPP
Ready for Physics
First release
Test of analysis chain on pseudo data
KPP Run
Engineering Run
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

17. CLAS12 Analysis Readiness
User Involvement
Software workshops and demonstrations at CLAS Collaboration meetings
3 times/year
Last week meeting: >30 participants on site and 15 remotely connected
Analysis workshop planned at next Collaboration meeting
Weekly meetings of software and CalCom groups
experts
developers
users
Mailing list and forums (
Documentation:
CLAS12 wiki: CLAS12_Wiki
CLAS12 docs:
GEMC:
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

18. CLAS12 Analysis Readiness
Summary
Simulations ready to generate realistic pseudo data
Reconstruction released for user to study physics reactions
Calibration tools developed with real and pseudo data
Now: ready for Feb commissioning run
July 2017: ready for Fall 2017 engineering run
Analysis tools under development:
Event generators
Event selection and data handling tools
High level physics analysis tools
Test of physics reaction analysis
Analysis organization and management defined
Ready for physics in Fall 2017
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

19. CLAS12 Analysis Readiness
Backup
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

20. CLAS12 Analysis Readiness
Event Reconstruction
Single track resolution and multi-track event reconstruction
Electron momentum and angular resolution
TDR:
σ(p)/p ~ 1%
σ(θ) ~ 1 mrad
σ(φ) ~ 3 mrad
σ(p)/p vs. p
σ(θ) vs. p
σ(φ) vs. p
ADD TRACKING EFFICIENCY
Missing mass from exclusive final states
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

21. Monitoring and Calibration GUI
ECMon
Monitoring and Calibration GUI
Current Features
Common JAVA framework for PCAL and EC.
Pixels dynamically generated from geometry database.
Mouse-over navigation of detector elements.
Live updating of detector response and calibration results.
Monitoring
Occupancy: strips, pixels, fADC and TDC data.
fADC data: pulse shape, noise, fitter settings.
Single event: visualize/analyze hits, peaks, clusters.
Pedestals: offsets, noisy channels.
EPICS data (scalers and HV) for status monitoring.
Calibration
Single pixel filters for cosmic muon hits.
Optimization of pixel selection (statistics, geometry).
Fits to pixel data: PMT gains and light attenuation.
Validation using GEMC simulations.
ECAL: 1296 pixels
PCAL: 6916 pixels
Further Development In Progress
Energy cluster reconstruction and trigger debugging support.
Energy calibration using physics data (e-, p0, and MIP pions).
Timing calibration (offsets, time-walk).
EC, PCAL relative alignment using cosmic muon pixel tracks.
Work by C. Smith (JLab), N. Compton (Ohio), T. Chetry (Ohio)
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

22. FTOF and CTOF Calibration
Work completed / in progress:
Classes for each calibration step provide standard interface to generate constants
HV, left-right timing, attenuation length tested with cosmic and GEMC data
Effective velocity and time walk – testing done with GEMC data
Work planned:
Further testing with GEMC data
Paddle to paddle offsets
Documentation
L. Clark (Glasgow)
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

23. CLAS12 Analysis Readiness
HTCC Gain Matching
The Calibration Suite for the HTCC is written using Common Tools provided by the CLAS12 Software group
The Suite is multithreaded and object oriented
The object oriented design of this software will save time
Due to the design, it could be adapted to detector systems that find the SPE peak position in other ways (e.g. LTCC)
Fitting Benchmarks
4 Threads
The task: fit 1488 histograms
~2min per fit serially
Java .jar file and .hipo data files were able to run across all platforms and OSes with no issues
32 Threads
64 Threads
Work by Y. Sharabian (JLab), N. Markov (UConn), W. Phelps (FIU)
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

24. CLAS12 Analysis Readiness
DC Calibration Status
trkDoca/docaMax
Time (ns)
Time-to-distance calibration program written (in ROOT) and tested on cosmic data available from one R1 chamber.
To be done: a GUI-driven automated suite:
Java version of the time-to-distance minimization routine to calibrate the parameters (In Progress).
GUI using common tools (Started)
Putting together various components:
Decoder (Done)
Time Delay Corrections (Done)
Reconstruction (Done)
Minimizer (Underway)
Writer of results to CCDB
Superlayer 1
Residual (cm)
= 246 microns
K. Adhikari (Mississippi), M. Mestayer (JLab)
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

25. CLAS12 Analysis Readiness
SVT Calibration Suite
Work in progress:
Creating CCDB tables
Porting to COATJAVA 3.0
S.N.R. > 15
short strips
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

26. CLAS12 Analysis Readiness
CND Calibration Suite
COATJAVA 2.4 based suite nearly complete - will be updated for use with 3.0 when required
Developed with cosmic data → some small changes will be required for GEMC data testing (plans in place for this)
CND group will fully test suite using initial documentation as part of an iterative suite improvement process
Complete:
timing offset correction
noise removal
time resolution
effective velocity
In progress:
light attenuation (80%)
energy calibration (initial work will be imported then optimized)
G. Murdoch (Glasgow)
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

27. CLAS12 Analysis Readiness
Forward Tagger
INFN – Genova, Edinburgh U., CEA-Saclay
System checkout and commissioning with cosmic rays
Combined Cal+Hodo+Trk monitoring GUI
Now upgrading to COATJAVA 3.0
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

28. MVT Calibration Status
1 ) Low-level calibration procedure developed that needs to be incorporated in the calibration suite: - Extract and store pedestals and RMS pedestals. - Store time resolution and sampling constants (number of samples, rate).
Pedestals
RMS Pedestals
2) Need now to implement the high-level calibration code that consists of: - Computing and correcting for Lorentz angle in the reconstruction Storing masks for dead/noisy strips in the database.
CEA - Saclay
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

29. CLAS12 Analysis Readiness
Alignment of the SVT
Track-based alignment of SVT requires fitting many parameters: Nsectors x Nlayers x Ntrans x Nrot = 66 x 2 x 3 x 2 = 792
Program millepede does linear least squares with many parameters.
Uses matrix form of least squares method and divide the elements into two classes.
Global parameters – the geometry misalignments. Same in all events.
Local – individual track fit parameters. Change event-to-event.
Calculate first partial derivatives of the fit residuals with respect to the
local (i.e. fit) parameters and global parameters (geometry misalignments).
Manipulate the linear least squares matrix to isolate the global parameters
(geometry) and invert the results to obtain the solution.
Type 1 tracks –
Horizontal sensors.
Apply to a ‘simple’ example – Type 1 tracks.
Simulated cosmics for validation.
Shift layers 1-2 (Region 1) by 20 microns in x.
millepede reproduces shifts in the range mm.
Apply to Type-1 cosmic ray sample from SVT.
5.9M events; fixed layer 4 to SVT residual.
Millepede misalignments added to reconstruction, residuals ~25 mm.
Resolution in sectors mm.
Code for more complex events now being tested.
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

30. CLAS12 Analysis Readiness
Alignment of the SVT
Ideal Geometry Validation and Testing
Calculate ideal fiducial location on each module.
Observed significant difference with engineering drawings - up to 100 µm.
Worked with engineers to correct differences.
Ideal geometry now well defined with parameters from engineering drawings.
Differences < 3 mm
Geometry package
Common Java utility to access geometry for gemc simulation and reconstruction.
Generate shifts from ideal geometry to measured fiducial results.
Processing fiducial survey data in alignment shifts – validating with simulated tracks.
Putting full inventory of material in SVT gemc simulation.
Fiducials form
triangle on
sensor.
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness

31. Physics analysis example Deeply Virtual Compton Scattering
Cross-sections
Beam Spin Asymmetry
Φ (°)
Ebeam = 6.6 GeV
Ebeam = 8.8 GeV
Accessing Generalized Parton Distributions amplitudes
JLAB Software Review, 11/10/2016
CLAS12 Analysis Readiness