CHICOS Analysis

CHICOS Maximum Likelihood Code
CTShower Code (May not be most recent.)
Shower Test Page (V. 3.03 Testing may not be most current. Be wary of using v. 3.02 as you may have to kill it manually.)
It seems like the reconstructor frequently gets thrown off by large hits, and that the problem has something to do with the HitLogLikelihood (timing probability function). I spent some time playing with other variations to see if I could get all the showers to reconstruct. On the previous page, only using HitLogLikelihood for hits < 100 MIP allows all but one shower to reconstruct. (02-03-07 Westridge shower.) That shower has one hit with 25 MIP, and ignoring it allows the shower to reconstruct. However, simply moving the cutoff for using HitLL to 20 does not allow the shower to reconstruct. Not sure what is going on there.

V. 303c, all data No contribution from HitLogLikelihood. (7 fail.)

V. 303c, no accidentals (4 fail.)

V. 303d, all data Only take timing info for hits < 20. (1 fails.)

V. 303e, all data KS test instead of HitLogLikelihood. (33 fail, many others are nonsense.)

V. 303f, all data Cutoff of 20 for both time and intensity probabilities. (Mostly nonsense, some fail.)

V. 303g, all data Like 3.01 (no cutoffs) but with TDF 1 (Agasa). (13 fail.)

V. 303h, all data Same, with HitLL cutoff at inten < 100. (6 fail.)

V. 303i, all data P_T = P_POISSON(t_delay,T_mode_time) (all but 6 fail.)

So two questions - 1.) Why does HitLogLikelihood cause the reconstructor to go haywire when there are large hits, and 2.) What is the problem with the shower on 02-03-07?