What we learned from DC1 B-physics validations

1. What we learned from DC1 B-physics validations pp  B(J/y(mm) K0) X M.Smizanska, Lancaster University for B-physics validation team.

2. DC1 B-physics validation teams: M.Smizanska, et al, DC1 B-validation, UK-physics meeting

3. List of Physics processes M.Smizanska, et al, DC1 B-validation, UK-physics meeting

4. Software tools, status at the start of DC0: • Detector simulation • ‘TDR’ detector layout was obsolent for several years, but the changes were not implemented in ATLAS software. atlsim 98_2 was identical to 97_6 for the ID description, which corresponded to a description in ID TDR. • … so an impact of important changes in the ID: increasing the radius of b-layer, eliminating second pixel layer and some other parts in the endcap - had to be estimated by ATLFAST – using simple approximations of the resolutions derived from TDR ones. • Physics performance for conferences was for a long time presented for TDR layout. • DC1 validation was important step forward: first publicly presentable B-physics results with new ID layout. M.Smizanska, et al, DC1 B-validation, UK-physics meeting

5. Software tools, status at the start of DC0, cont • ATHENA, Generators • TDR B-physics generator Atgenb – a branch of Atgen an ATLAS interface package that stop to be supported in 97. • In DC0 – Atgenb – rewritten to PythiaB – ATHENA algorithm, was in use for DC1 production. • Reconstruction: • atsim, atrecon – longest survivors over TDR-DC0-DC1 … finally were useful for ATHENA validations – most of our DC1 done in parallel using atrecon (or atlsim!) and ATHENA • ATHENA-reconstruction much progress during DC1 … • … but we did not reach the same performance for the 3 packages in DC1. The sources of differences are more-less understood, but all these packages stop at DC1. All manpower - to DC2. M.Smizanska, et al, DC1 B-validation, UK-physics meeting

6. Detector Layouts in DC1 validations M.Smizanska, et al, DC1 B-validation, UK-physics meeting

7. Software in DC1 validations *) part of Complete and Initial also with iPatrec, 6.0.3 M.Smizanska, et al, DC1 B-validation, UK-physics meeting

8. Performance results M.Smizanska, et al, DC1 B-validation, UK-physics meeting

9. Mass reconstruction Core of mass distributions similar with Complete and Initial layouts and Complete-300mm. Degradation vrt TDR: 10-15% M.Smizanska, et al, DC1 B-validation, UK-physics meeting

10. Complete vs Initial layout: reconstruction of B-signal mass xKalman6.5.0 optimized track finding strategy Example for channel BsJ/y(m6m3)f(K+K-) All four tracks of B reconstructed e = 82.5% (5% B in tails) e =77.9% (6% B in tails) … and B-vertex reconstructed e = 82.3% (5% B in tails) e = 77.7% (6% B in tails) Initial Layout: 1. Efficiency to reconstruct B only 4.5% smaller then in Complete. 2. Only 0.3 % fails the vertex fit in both Complete and Initial layouts. Complete Layout Initial Layout M.Smizanska, et al, DC1 B-validation, UK-physics meeting

11. All four tracks of B reconstructed e = 83% (4% B in tails) e =77% (11% B in tails) … and B-vertex reconstructed e = 82% (4% B in tails) e = 67% (8% B in tails) …the same events withdefault xKalman track search strategy – failed for Initial layout, ok for Complete layout. Example for channel BsJ/y(m6m3)f(K+K-) Initial Layout: 1. More B’s in tails. 2. Efficiency to reconstruct B tracks only 6% smaller, 3. however next 10% fails the vertex fit Complete Layout Initial Layout M.Smizanska, et al, DC1 B-validation, UK-physics meeting

12. B-hadrons - proper time resolution Core of proper-time distribution similar in Complete and Initial layouts. Degradation vrt TDR: 20-35%. Degradation Complete 400mm vrt 300mm : 14% M.Smizanska, et al, DC1 B-validation, UK-physics meeting

13. B-hadrons proper-time resolution, optimized xKalman Example for channel BsJ/y(m6m3)f(K+K-) Both Complete and Initial layout similar: Bs proper-time reconstruction: 8% in tails only 0.3% fails vertex fit in both layouts Complete Layout Initial Layout M.Smizanska, et al, DC1 B-validation, UK-physics meeting

14. All four tracks of B reconstructed and B-vertex reconstructed e = 82% (7% B in tails) e = 67% (16% B in tails) B-hadrons proper-time resolution, default xKalman Example for channel BsJ/y(m6m3)f(K+K-) Bs proper-time resolution: 1. Complete layout 7% in tails 2. Initial layout 16% in tails … and lower efficiency of track reconstruction and less sucessful vertex fits -> all these factors lead to decrease of efficiency. After final selection cuts in this channel Initial : Complete 3:5 Complete Layout Initial Layout M.Smizanska, et al, DC1 B-validation, UK-physics meeting

15. Efficiency of reconstruction of B-signal including vertex fit Initial vs Complete, optimal xKalman M.Smizanska, et al, DC1 B-validation, UK-physics meeting

16. Complete layout vrt Initial layout similar performance. • … so are we going to miss second pixel layer? • The simulation was optimistic • inefficiencies underestimated • no misalignement • degradation appears at higher multiplicities - already at L=2 x 1033 cm-2 s-1 M.Smizanska, et al, DC1 B-validation, UK-physics meeting

17. Single-track performance: wrong hit on track in b-layer; dependence on: layout, multiplicity and pT. Complete layout vrt Initial similar - if only a signal event simulated. Degradation at higher multiplicities (already at L=2 x 1033 cm-2 s-1 ). M.Smizanska, et al, DC1 B-validation, UK-physics meeting

18. Athena7.0.0 versus atrecon… different performance ATHENA worse by  ~10% than atrecon6.5.0 atrecon 'private'  ~8% worse than atrecon6.5.0. Degradation due to pixel clusters errors. Other -smaller factor:  ATHENA more inefficiencies   Can ATHENA7.0.x be corrected ??  M.Smizanska, et al, DC1 B-validation, UK-physics meeting

19. Athena7.0.0 versus atrecon… different performance • We finish DC1 with ATHENA reconstruction in which we are aware of errors • Degraded performance in proper-time vrt atrecon. • Can ATHENA 7.0.x be improved?? NO • cannot invest time for old code – need people for 7.3.0… • DC1 Simulation not realistic anyway, for instance misalignment… M.Smizanska, et al, DC1 B-validation, UK-physics meeting

20. Conclusions • Initial vrt Complete layout - similar performance – if no pileup, with optimized track search strategy in atrecon6.5.0. • Track-finding efficiency -7% for pt (0.5-1.0) GeV, only -2%. for pT>1GeV • Tracks with wrong hit in B-layer: Initial 3%, in Complete 2% for pT<1GeV • Efficiency of B-signal reconstruction Initial vrt Complete: lower by ~6% - due to track search inefficiency. Only 0.3% fails vertex fit in both Initial and Complete layouts. • Comparisons with other layouts • Mass resolution: degraded 10-15% vrt TDR, • Time resolution: degraded 20-30% vrt TDR, and 14% 400mm vrt 300mm. • Still to be done in DC1: Signal events with minimum bias. M.Smizanska, et al, DC1 B-validation, UK-physics meeting

21. Conclusions,cont • We are aware of insufficiencies of DC1 and we understand their impact on performance. • This will alllow us to use reasonably the DC1 software validation results as a starting point to validate DC2 software. M.Smizanska, et al, DC1 B-validation, UK-physics meeting