Bumps part II – more evidence for scalar quarks? January 30, 2007Posted by dorigo in personal, physics, politics, science.
As I discussed yesterday , the situation in CDF was red-hot in 1999, after Giromini’s group had shown they were seeing a bump which resembled a resonance in the dimuon mass distribution, and had proposed a fancy explanation which would fit the bump as well as two other intriguing results they had found in the previous years, and had been unable to get approved for publication.
It was in fact some kind of an accumulation point, reached after years of debates on obscure technical details affecting the measurement of the top quark cross section, maturated into a controversy on the origin of a handful of top candidate events (13) which were unexplainable with ordinary means, and saturated into little less than a garage fight between proponents of the analysis -who wanted to publish those 13 “superjet” events – and top group coordinators and spokespeople.
The funny thing was that, arguably, the scientific method had been followed after all – by a combination of Giromini’s intuition and the steering force of a skeptical collaboration.
So, before I delve into discussing significance issues, let me make a flash-back and tell in a logical sequence the events that had gotten us in front of the dimuon mass bump.
The facts: following the top quark discovery by CDF and D0 in 1995, Giromini’s group had spent the years 1996-1998 measuring the top quark pair production cross section in the freshly collected Run I dataset, and had been in competition with another big group, mainly composed of American physicists who had more political support in the collaboration. The two groups had found different results, mainly because of the evaluation of something called the “b-tagging scale factor”, which Americans claimed to be unity (data and Monte Carlo jets would be tagged at the same rate), and Frascati claimed to be 1.25 (a higher Monte Carlo tagging rate). Frascati’s group had been silenced after many controversial debates in the group meetings, and the top cross section had been measured at (7.6 +1.8 -1.5) pb (submitted paper here ).
A higher scale factor meant a smaller cross section for tt production. And sure enough, after three more years of malignant external scrutiny of his work, stomped feelings, and embittered relationships, Giromini had been proven right (see the scale factor measurement as a function of jet transverse energy above), and an improved measurement was finally published at (5.1+-1.5) pb in September 2001 ( submitted paper here). For what is worth, the theoretical prediction at 1.8 TeV is today 4.9 pb if I remember correctly.
Now the plot thickens. For the sample used to measure top quark pairs in the single lepton final state (so-called “W+jets”, containing a leptonic W decay plus several hadronic jets) contained an excess of events with peculiar characteristics. A smaller top pair cross section meant this excess was even more prominent: 13 events were found with a jet of very large energy, containing both a secondary vertex (a “b-tag”, made by tracks which would not fit to the interaction vertex but were produced by the decay of a long-lived particle) and an electron or a muon. Electrons and muons are indeed produced by b-quark decay, but the 13 events with these “superjets” were way more abundant than what one would expect to get (4+-1 events) from normal processes. CDF had an excess of superjet events.
So what do you do when you observe an excess of events of a particular class ? Simple: you study their kinematic characteristics, to see if these are what you expect from any one of the processes you expect to have contributed to the data sample at hand.
The kinematics of those 13 events was weird. A statistical analysis examining a complete set of kinematic observables from the events showed they were utterly incompatible with known processes – at a level exceeding 5-sigma. One could indeed start to fancy what exotic process could produce such a weird signal when a jet with a secondary vertex always contained a charged lepton. A scalar quark with a 100% branching ratio to leptons was a possible candidate – if a bit nutty – explanation.
The proposed explanation by Giromini generated further more controversy, and additional endless scrutiny. I was then asked to be part of the “Oversite godparenting committee” which had the uncomfortable assignment of determining the soundness of the analysis. We did many tests, produced ancillary results, studied independent samples. No mistake was found, either in the background prediction (4+-1 events), or in the statistical tests. Below is a plot of the probability of observed kinematic distributions for the superjet sample and for a control sample of similar events containing no lepton in the “superjet”. It seemed really hard to find a sample with kinematics as weird as that of those intriguing events!
Then, Paolo Giromini had a brilliant idea. If those events were the result of some fancy scalar quark produced together with the W, one could seek it in other datasets. And the Frascati group looked in two samples they knew well: dijet events with an electron or muon in the jet. Those samples had been used to measure the scale factor mentioned above.
Sure enough, a careful investigation of the rate of multi-lepton events revealed a startling excess of jets with two leptons in the same jet, and with a very small angle between them. More weird stuff! And this time, the statistics were not small: it was hundreds of events with those characteristics, so one could study them in detail with advanced statistical means.
The publication process of these two anomalies took more years of pain. And while the review process was in full swing, Giromini had another idea: let’s see if this scalar quark makes a bound state! If it produces leptons at small angle, it must have spin zero, and maybe – just maybe – if it got bound in a squark-antisquark pair in a p-wave configuration, it would have the same behavior of spin-1 vector mesons such as the J/psi or Upsilon. So let’s look in the sample where the latter show up: dimuon resonances!
Above you can see a dimuon mass distribution of opposite-signed pairs (in blue) and same-sign pairs (in red). The peaks are respectively the J/psi, the psi(2s), the Upsilon(1s), (2s), and (3s) resonances. Quite amazing, if you ask me. These are really bound states of two tiny things orbiting around each other, “resonating” for a tiny fraction of a second before disintegrating into two muons!
Giromini looked closer. He knew his alleged particle had to have a mass of something between 6 and 8 GeV – his multi-lepton signal had convinced him the mass of the scalar quark would be in the 3 to 4 GeV range. He knew what the width of his resonance had to be. And he could guess what the rate would be.
[To be continued…]