New paper approved!

My colleague Giorgio Cortiana (right) gave a Paper Seminar in CDF last week on the recent measurement of the top quark mass we (well, mostly he) carried out on the dataset of top quark pairs we isolated in 311 inverse picobarns of Run II data. The paper seminar is the last formal step the authors of an article based on CDF data have to make before entering the final phase of the publication process – and tomorrow we will indeed start that, by sending our draft to Physical Review D.

Giorgio has been working with me in Padova since 2000, first as a summer student, then as a undergrad, then as a grad student, and now as the recipient of a research grant. Together, we have written no less than 22 internal CDF notes, and published two papers in refereed journals – well, three with this one, if referees allow.

It is unfortunate that we will probably have little chances of working together in the future, given the ramping down of my involvement in CDF, while he is going to be fully in CDF for a while longer – he is presently heavily involved in the upgrade of the Level-2 calorimetric trigger of CDF -which is in its final testing phase, and in several other tasks connected to the Higgs searches.

So, the paper. It is a measurement of the top quark mass in events that other analyses have never considered as a mine from which to dig out top pair production events. These events, collected by a multijet trigger, are selected by requiring the signal of a energetic neutrino: a large imbalance in the energy emitted orthogonally from the beams direction.

The neutrino is a very good tag of the decay of a W boson, and the presence of four or more hadronic jets -one of which has to be b-tagged- enrich the data of top pair production events.

The top quark decays almost exclusively to a W boson and a b-quark. The former can decay to a pair of jets or to a lepton-neutrino pair, while the latter produces a b-jet. In the case of “single-lepton” top pair production events such as those collected in our sample, there is a total of two b-jets and two W bosons, with one W decaying to a jet pair and the other decaying leptonically -yielding our precious neutrino signature.

The primary charged lepton produced together with the neutrino can be lost if it gets mistaken for a jet (tau decays do that most of the time), or it fails the identification criteria, or it leaves the interaction region in the direction of the beam, where CDF has no coverage for electrons or muons. We salute these unfortunate disappearances!, because otherwise those top candidate events would make it to the official “single lepton” datasets, and our little analysis would be left with no data to play with. Instead, by explicitly vetoing well-identified charged leptons, we collect data that other analyses discard, and we are free to mine them for top pairs.

It seems straightforward, right ? However, ours is the first analysis that exploits this category of events. The data already allowed us to produce a successful (and quite precise: when it was approved it was CDF’s third-best) measurement of the top pair production cross section [see Phys. Rev. Lett. 96, 202002 (2006)], and now we used it to measure the top quark mass. A bold move, since the lack of a charged lepton and the insufficient information from the neutrino make it really tough to fully reconstruct the kinematics of the decay.

When one does not have all the information to reconstruct invariant masses of the two produced top quarks, one can use a global variable, easy to measure and understand, which is correlated with the top mass. In our case, it is the “Ht”: the sum of jet energies plus missing transverse energy. The distribution of measured Ht in the data carries information on the mass of the top quark.

Above you see the Ht distribution for data (black points with error bars) as is interpreted by a two-component fit to a background template (in red) and a mass-dependent signal template (in blue). On the right, a sketch of the likelihood dependence on the top mass shows that the minimum is at 172 GeVish.

At the end of the day, our measurement is not very precise. It could jolly well be ignored in the average of CDF determinations without losing more than 2% of the information. We in fact measure Mt=172.3+-10.8(stat)+-10.8(syst) GeV, for a total error six times larger than the most precise one produced so far by CDF, using a matrix-element method and single lepton top decays (which however used three times more statistics).

“Your error bar is very large, so what is the justification for publishing your paper?“, asked a rather annoying colleague at the Paper Seminar. A silly question to come from an experienced physicist, if you ask me. If the name of the game were just “shrink the error bar”, and nobody cared to publish new ideas although still not competitive, our field would dry out as a desert creek. Do not let us kid ourselves: the analysis deserves to be recorded in print. And so it will.