Comparison between Aasim and Calibob Motivation: - two different MC to extract opticalproperties - estimate systematic uncertainty Comparison for MC inputs: - with default optical properties - with no scattering Thanks to Stephan and Aart for their help
Input Aasim and Calibob simulation have: L_abs=50 m, L_scatt=50 m, eta = 0.17 Run=58120 No efficiency corrections Remark: Calibob based on KM3 Aasim without simulation of electronics
Comparison Compare MCs along same line Following histogram have same integral Check line 2 (similar results for other lines) Compare - higher part of the line (photo electron regime) - lower part of the line
Line 2 (each om with interval [-0, 300] ns) Black: calibob Red: aasim Tails seem to agree Scattering seems ok Floor 12 Floor 13 Floor 14 Floor 15 Floor 16 Floor 17
Log scale: Line 2 (each om with interval [-0, 300] ns) Black: calibob Red: aasim Floor 12 Floor 13 Floor 14 Floor 15 Floor 16 Floor 17
Log scale: Line 2 (each om with interval [-0, 300] ns) Black: calibob Red: aasim Floor 12 Floor 13 Floor 14 Floor 15 Floor 16 Floor 17
Line 2 far photo-electronregion (each bin is the integral of a PMT) PMT get different amount of light - scattering ? - angular acceptance ? - binning ? - shadowing ? - etc. ? Aasim more pronounced variations Black: calibob Red: aasim Floor 12 Floor 13 Floor 14 Floor 15 Floor 16 Floor 17
Line 2 near (each bin is the integral of a PMT) Near OMs Histogram missing: Floor 7 OM 0 Floor 10 OM 0 Floor 5 Floor 6 Floor 7 Floor 8 Floor 9 Floor 10 Floor 11
Input: No scattering Aasim/Calibob simulation have no scattring: L_abs=50 m, L_scatt>10^5m, eta = 0.17 Run=58120 No efficiency corrections
No Scattering: Line 2 fare (each bin is the integral of a PMT) Black: calibob Red: aasim Smaller difference =>difference generated from scattering Stephan already knows Floor 12 Floor 13 Floor 14 Floor 15 Floor 16 Floor 17
No scattering: Line 2 near (each bin is the integral of a PMT) Black: calibob Red: aasim For floors which are near OB differences in integral even for MCs without scattering => some other effects also play a role Near OMs Histogram missing: Floor 7 OM 0 Floor 10 OM 0 Floor 5 Floor 6 Floor 7 Floor 8 Floor 9 Floor 10 Floor 11
Conclusion Aasim and Calibob disagree, but we can use it as uncertainty of our systematic Important check to do: Can we find the correct input optical properties when using e.g. Calibob runs as MC templates and an Aasim run as unknown data? If yes we have enough sensitivity to extract optical properties
Find best Calibob run for Aasim run (aasim as data) Input Scatt= 50 m Abs = 50 m Minimum at Scatt=50 m Abs =45 m This is valid for Floors between Floor 12 and Floor 20 Method has enough sensitivity to distinguish optical properties
Data vs MC
Data vs Calibob Super-Super-Histogram (all lines except OB line) 180 PMTs
Chi2 for all lines at same time 180 PMTs
Old way Chi2 is sum of 11 lines Chi2/11
because of ARS token ring effect, select photoelectron region OM selection Some OMs are rejected: OMs too close to the OB Floor > 13 because of ARS token ring effect, select photoelectron region OMs too far away Floor < 21 because of missing statistics OMs whose efficiency ε < 0.5 or ε > 1.5 because of large extrapolation Backwards looking OMs because of PMT acceptance uncertainty OMs very inclined Led emission uncertainty OMs after visual inspection of their distributions This selection can introduce a possible bias
All cuts 180 PMTs Numbers give Chi2/ndf Minimum seems to be slightly flatter (45 – 60 m)
Include backward looking and very inclined 242 PMTs Minimum stable
Include also visual inspection rejected 263 PMTs Minimum stable
Increase photo electron region cut (0.35->0.55) 324 PMTs Minimum stable at 45-60 m
Increase photo electron region cut (0.35->0.75) 349 PMTs Minimum stable
Decrease photo electron region cut (0.35->0.25) 235 PMTs Minimum flat and shifts
Decrease photo electron region cut (0.35->0.15) 192 PMTs Minimum shifts
Conclusion How to get to the final result? In principle we have two different MC. We will never get good chi2. In reality they are very bad. Should we compare Aasim with data?