Speaker:
Tom Eichlersmith
(UMN)
- Difference in data size between L1L1 and L1L2?
- Re-calculate mass resolution for L1L2?
- Overly optimistic exclusion estimate could be coming from multiple factors
- May be because we optimized on this set, maybe have a different test set?
- Could need to scale denominator (the Allowed) when scaling up, will need some smoothing method to avoid propagating statistical fluctuations
- Optimization strategy should be using exponential re-weighting somewhere to help bring cuts closer to the target
- Expand search window to 220MeV in invariant mass to align with bump hunt