1. EFA/DHCal development at NIU
Vishnu V. Zutshi
Northern Illinois University

2. V. Zutshi, Cornell LC Workshop
Introduction
Primarily interested in exploring the (semi)digital hadron calorimeter option in general, with scintillator as the active material in particular
This talk focuses on the algorithm development aspects
Results are preliminary
11/28/2018
V. Zutshi, Cornell LC Workshop

3. V. Zutshi, Cornell LC Workshop
The Calorimeter
30 layer Silicon-Tungsten ECAL
5mm x 5mm transverse segmentation
34 layer Scintillator-Steel HCAL
Transverse cell size varied between 2-9cm2
Cells are projective in q and f
No support structures/dead spaces
Immersed in 5T magnetic field
11/28/2018
V. Zutshi, Cornell LC Workshop

4. V. Zutshi, Cornell LC Workshop
N vs. E
11/28/2018
V. Zutshi, Cornell LC Workshop

5. V. Zutshi, Cornell LC Workshop
Sampling Weights
Determined from 10 GeV charged pions
Minimize:
(1/N) S (E0 – aiLi) where
E0 is the incident energy
ai is the weight for layer i
Li is the energy/number of hits in lyr i
For simplicity i=2 considered here
11/28/2018
V. Zutshi, Cornell LC Workshop

6. V. Zutshi, Cornell LC Workshop
Energy Resolution
single threshold
11/28/2018
V. Zutshi, Cornell LC Workshop

7. V. Zutshi, Cornell LC Workshop
Energy Resolution
dual threshold
11/28/2018
V. Zutshi, Cornell LC Workshop

8. V. Zutshi, Cornell LC Workshop
Energy Resolution
11/28/2018
V. Zutshi, Cornell LC Workshop

9. V. Zutshi, Cornell LC Workshop
“Density”
Need a hierarchy for cells, in the absence of an energy measurement
Clumpiness of surroundings ?
A simple-minded realization of this:
di = k S (1/Rij) where Rij is the angular distance between cell ‘i’ and cell ‘j’
11/28/2018
V. Zutshi, Cornell LC Workshop

10. V. Zutshi, Cornell LC Workshop
Position Resolution
Since Energy-Flow invariably involves associating clusters near an extrapolated track…….
Best way to this in a digital calorimeter?
Used charged pions in the HCAL
Resolution is defined w.r.t. the extrapolated Monte Carlo generated track
11/28/2018
V. Zutshi, Cornell LC Workshop

11. V. Zutshi, Cornell LC Workshop
Position Resolution
5 GeV p
11/28/2018
V. Zutshi, Cornell LC Workshop

12. V. Zutshi, Cornell LC Workshop
10 GeV p
Measured relative to the energy weighted resolutions
Cell area at first layer=2cm2
11/28/2018
V. Zutshi, Cornell LC Workshop

13. V. Zutshi, Cornell LC Workshop
10 GeV p
Cell area at first layer=6cm2
Cell area at first layer=9cm2
11/28/2018
V. Zutshi, Cornell LC Workshop

14. V. Zutshi, Cornell LC Workshop
Clustering
Local density maxima chosen as seeds
Membership of each cell in the seed clusters decided with a distance function
Only unique membership considered here
Calculate centroids
Iterate till distortion is below some threshold
11/28/2018
V. Zutshi, Cornell LC Workshop

15. V. Zutshi, Cornell LC Workshop
p0  gg
Density weighted q-f
High asymmetry
11/28/2018
V. Zutshi, Cornell LC Workshop

16. V. Zutshi, Cornell LC Workshop
p0  gg
Density weighted q-f
11/28/2018
V. Zutshi, Cornell LC Workshop

17. V. Zutshi, Cornell LC Workshop
10 GeV p0
Recon. energy
Recon. mass
18%
11/28/2018
V. Zutshi, Cornell LC Workshop

18. V. Zutshi, Cornell LC Workshop
20 GeV p0
Recon. energy
Recon. mass
11/28/2018
V. Zutshi, Cornell LC Workshop

19. V. Zutshi, Cornell LC Workshop
Energy Asymmetry
A = abs(Eg1 – Eg2)/(Eg1 + Eg2)
11/28/2018
V. Zutshi, Cornell LC Workshop

20. V. Zutshi, Cornell LC Workshop
S+  pp0 p
Density weighted q-f p0
11/28/2018
V. Zutshi, Cornell LC Workshop

21. V. Zutshi, Cornell LC Workshop
S+  pp0
Eflow
Cal only
Erec/Egen
11/28/2018
V. Zutshi, Cornell LC Workshop

22. V. Zutshi, Cornell LC Workshop
S+  np+ n
ECAL
Density weighted q-f
HCAL p
11/28/2018
V. Zutshi, Cornell LC Workshop

23. V. Zutshi, Cornell LC Workshop
“Jet” Reconstruction
A 0.7 simple cone (no splits or merges) algorithm used
Operates on a pt ordered list of final state, visible MC particles
Pt weighted q-f used as centroid
11/28/2018
V. Zutshi, Cornell LC Workshop

24. V. Zutshi, Cornell LC Workshop
Z  jj
Generated mass
11/28/2018
V. Zutshi, Cornell LC Workshop

25. Further “Jet” Reconstruction
Use the generated jet centroid as the seed
Grab clusters within 0.7 of seed
Extrapolate MC tracks into calorimeter
Replace E with p for matched clusters
11/28/2018
V. Zutshi, Cornell LC Workshop

26. V. Zutshi, Cornell LC Workshop
Jet Erec/Egen
~60% better
Eflow
Calorimeter only
11/28/2018
V. Zutshi, Cornell LC Workshop

27. V. Zutshi, Cornell LC Workshop
Jet Erec/Egen
Eflow digital (2cm2 cells)
Eflow analog
11/28/2018
V. Zutshi, Cornell LC Workshop

28. V. Zutshi, Cornell LC Workshop
jet-jet mass
EF analog
2 cm2 cells
50% improvement over calo
EF digital
Calo only
11/28/2018
V. Zutshi, Cornell LC Workshop

29. V. Zutshi, Cornell LC Workshop
jet-jet mass
6 cm2 cells
Digital(1-bit) ~10% worse
11/28/2018
V. Zutshi, Cornell LC Workshop

30. Photon Reconstruction (in jets)
11/28/2018
V. Zutshi, Cornell LC Workshop

31. V. Zutshi, Cornell LC Workshop
Short Term Goals
Better track extrapolation taking into account the energy loss and scattering in the material
Introduce particle id to supplement pattern recognition
Alternative measures of density
The long road to optimization
11/28/2018
V. Zutshi, Cornell LC Workshop

32. V. Zutshi, Cornell LC Workshop
Summary
A first version of a ‘digital’ Energy Flow algorithm implemented with encouraging results
Detailed optimization studies beginning
Not discussed here, but hardware prototyping studies proceeding
Hope to cross-fertilize hardware design with algorithm development
11/28/2018
V. Zutshi, Cornell LC Workshop