IC40 Cascade Analysis
Stephanie Hickford
University of Canterbury
Last updated:

---

Stephanie's Homepage Analysis Data CORSIKA Signal Level 2 Trigger (level 0) Pole filter (level 1) Level 2 Level 3 Cut variables Reconstructions Development of cuts Level 4 Cut variables Reconstructions Development of cuts Level 5 Cut variables Development of cuts Level 6 Cut variables Development of cuts Final Sensitivity Remaining events Resolutions Data/Monte Carlo Comparison Excess in COG Level 5 TMVA Summary Of Issues Feature extractor Hit binning bug Good DOM list DOMcalibrator Systematics Ice properties DOM efficiency Cross-section Seasonal variation Unblinding Discussion Diffuse call Analysis call Results Plots Events

Data/Monte Carlo Comparison This page deals with the discrepancy between Experimental Data and CORSIKA that is seen in all analyses. The first section tracks the discrepancy for this analysis at each level, and the second section is an in-depth look at the discrepancy at the BDT level. Excess in COG This plots below show the differences between Experimental Data and Monte Carlo for the Centre of Gravity reconstructions. The ratio of Data/MC is tracked at each level. Level 2 Data = 1.62998 × 101 Hz Monte Carlo = 1.24808 × 101 Hz Excess of Data over Monte Carlo = 1.30599 Statistics Data: burn sample runs 111460, 111490, 111530 (6.74 hours) Monte Carlo: ~6 days unweighted single, double and triple Signal: dataset 2182 Level 3 Data = 1.74854 × 100 Hz Monte Carlo = 0.924745 × 100 Hz Excess of Data over Monte Carlo = 1.89083 Statistics Data: burn sample from August only (54.01 hours) Monte Carlo: ~6 days unweighted single, double and triple Signal: dataset 2182 Level 4 Data = 2.54148 × 10-2 Hz Monte Carlo = 3.330405 × 10-2 Hz Excess of Data over Monte Carlo = 0.769201 Statistics Data: whole burn sample Monte Carlo: all Signal: dataset 3221 Level 5 Data = 2.08697 × 10-3 Hz (6556 events) Monte Carlo = 2.49128 × 10-3 Hz (7611 events) Excess of Data over Monte Carlo = 0.83771 Statistics Data: whole burn sample Monte Carlo: all Signal: dataset 3221 Final level Data = 6.3666 × 10-7 Hz (2 events) Monte Carlo = 0 Hz (0 events) Excess of Data over Monte Carlo = undefined Statistics Data: whole burn sample Monte Carlo: all Signal: dataset 3221 Figure 1: Reconstructed Centre of Gravity at each cut level using a) COGX, b) COGY, c) COGZ, d) Ratio of experimental data to all Monte Carlo in COGZ. The green horizontal line is the mean excess of data over Monte Carlo. Level 5 TMVA This plots below are identical to those from TMVA on my level 5 webpage with the 10% burnsample substituted in for signal. This makes no sense in terms of analysis, but is a useful tool in comparing the differences between Experimental Data and CORSIKA at the BDT level cut. Figure 2: Variables used in TMVA. Figure 3: Correlation Matrices for TMVA. a) Signal. b) Background. Figure 4: Other TMVA plots. a) Overtraining check. b) Cut efficiencies. c) ROC curve.

---

Supervisors: Dr. Jenni Adams and Dr. Suruj Seunarine
email: Stephanie Hickford