Evidence December 9, 2006Posted by gordonwatts in physics.
Dugan gave a great Wine & Cheese seminar and announced the results. I don’t have the final copy of his talk, or I’d post it here. The place was packed with people (sitting in the aisles) and Young-Kee got the lab to buy some fake bubble and gave us a nice toast at the end of the talk (thanks – I know that isn’t normal procedure, but I think the lab should do more things like that in the future!). In summary:
- Three methods to search for single top production on a common dataset: Decision Trees, Matrix Element, and Bayesian Neural Network. Listed in order of decreasing expected significance as measured on zero-signal ensemble tests.
- The DT sees single top with a significance of 3.4 sigma, the ME is 2.9, and the BNN is 2.4. All measured, again, on the zero signal ensembles.
- As several people from CDF pointed out (repeatedly :-)) we were amazingly lucky. Our DT’s expected significance is about 2.1 — we see 3.4. Of course we measure a high cross section: 4.9 + 1.4 – 1.4 pb. CDF has an expected sensitivity in their best analysis of 2.5 sigma. They were unlucky, however, and saw only 2.3 (I’m recalling those last two from memory). So CDF has, if we measured sensitivity the same way, a better analysis, but we have better data. I guess we at DZERO need to be a little bit humble about that (even though that is hard to do sometimes!).
Afterwards there was a party. We started a bar tab and invited all of Fermilab to come. I took some video (to come later) and some pictures. But the pictures don’t do the get together justice: I forgot about my camera and didn’t start taking pictures until just before I had to leave: a number of people had already left.
As Aurelio pointed out: this is just the start for both experiments. We’ll combine and extend our techniques, get to the 5 sigma range, and then do the thing we really want to do: start measuring its properties!
Perhaps I missed it, but I don’t see an announcement, and I don’t see a copy of Dugan’s talk posted yet. A PRL is coming after that, and since that isn’t the most useful thing for an analysis this complex, a PRD is coming soon after that. And we thought making this announcement would give us some breathing room!