1) Apply a "tau DLV" to our standard sample. This would define a new standard sample that Harold would also switch to. In other words, we add a veto of events where we explicitly reconstruct the ttbar->tau decay.

2) Invert the tau DLV to define a control sample. This consists of the events rejected in 1). We use this to provide a MET template for the unidentified ttbar->tau that remains in the (new) standard sample, normalizing it to the data for MET<100.

3) We calculate MT for the control sample and cut on it 40<MT<90 to reduce susy contamination.

Of course, there is still a lot of "proof of principle" that is needed, and we need to know how to handle ttbar->tau events in which the tau decays to e,mu. We need the cheat info for the tau decays for this, which Sudan is also working on.

-- Bill

-- HaroldNguyen - 08-Sep-2010