Mirror of the best candidate showers page.
I notice that the showers which fail seem to have at least one large hit (>100). Changed HitLogLikelihood_Primitive so that if both I_event and I_LDF are both >10, P=1. This allows 04-19-03 to find a fit, but not the other showers that failed. Playing with the threshold and adding the same change to IntensityLogLikelihood_Primitive doesn't seem to make a difference.
01-29-03 Better error bars.