Ready for another turn into National-Enquirer mode of particle physics reporting ? I have a figure to discuss. It is a result now a few months old, but one which received little attention -less than it should have, perhaps. I myself got to see it only a few weeks ago in a presentation given by Jacobo Konigsberg, CDF spokesperson, at a workshop in Bologna.
The scandalistic cut of this article is manifest in the title: facing a dearth of exciting reports of new physics discoveries, we are bound to now and then swerve off the path of our usual responsible handling of two-sigma effects, odd deviations, and assorted zoology. It is, I believe, a necessary resource to rely on, if we are to keep the interest of Science readers on particle physics.
The effect discussed here is one found by the CDF collaboration while looking for the decay of a heavy Z' boson, a particle of very similar characteristics to those of the Z boson, the electrically neutral weak-force carrier first discovered by Carlo Rubbia in 1983 and then studied with disgusting precision by the LEP collider at CERN in the following decade.
No Z' bosons exist, unless Nature - the bitch, not the magazine - has hidden it from our sight like a annoying dime rolled off to a dark corner: the dark corner of very high-energy, a niche that the Tevatron proton-antiproton collisions are starting to illuminate only now, thanks not so much to a higher arm reach, but to a brighter lamp. The brighter lamp is provided by the large "integrated luminosity" of collisions acquired since the start of Run II seven years ago. Luminosity is a measure of how many collisions are produced in a given amount of time, for given beam parameters. The larger is time-integrated luminosity - luminosity multiplied by time, that is- the brighter we can illuminate those dark spots.
The new CDF analysis searched a dataset corresponding to a luminosity of 2.9 inverse femtobarns, id est about 200 trillion collisions. Among them, a few Z' bosons must have been produced if those particles exist, and if their mass is not too far away from our arm's reach. The search involved hypothesizing that the Z' bosons yield a pair of "ordinary" W bosons when they disintegrate, and that the two W bosons in turn produce a mixed decay: one W yields an electron-neutrino pair, the other a pair of hadronic jets. The energetic electron guarantees that the event is efficiently collected and sorted out from the trillions not useful for the search.
It would take too much of your time to learn the basics on jets and neutrinos and how they are detected, while I wish to get to the money plot before you get tired of my scribblings. So let me just explain that from the observed jets and electron and missing energy, one can reconstruct the mass that the Z' boson would have had, were it the progenitor of those particles. One may thus select a sample of data which really looks like the decay of a Z' -by apparently containing the decay of a pair of W bosons- and study the invariant mass distribution. A Z' boson signal will appear in the histogram as a bump at a well-defined mass value; backgrounds will instead distribute all over the board.
The CDF experimenters took great care to understand the rate and properties of all backgrounds which may mimic the Z' decay signature. The two largest contributions are due to the production of a single W boson accompanied by two jets, and to the production and decay of two top quarks. The latter was a once very dear signals to us, but top quarks are now just an annoying background for many of the searches we are undertaking. Fortunately, we know them top quarks well enough to model their characteristics and rate.
The figure below shows the various backgrounds stacked with bright colours in the mass histogram (beware, the vertical axis is logarithmic, so equal areas do not represent equal number of events). The black points show the real data collected by CDF after all selection requirements are applied. You can see that at low values of reconstructed mass -say below 500 GeV- the points line up with the red line quite nicely, a sign that backgrounds have been estimated well there. However, at masses around 600 GeV something fishy is apparent. The data appear to depart from the red line! Is that a sign that a few Z' bosons have actually been netted ?
Of course nobody knows. How can you tell ? Either hypothesis is viable: the hypothesis that the standard model is valid (represented by the red curve) does not provide an outrageously bad fit, although a few points do depart from it at 600 GeV; the hypothesis that a 600 GeV Z' boson has joined the party (represented by the blue curve, which includes the shaded blue histogram representing the Z' resonance added to standard model backgrounds) is fitting the data better, but that is not enough to claim we have discovered anything.
An interesting fact from a sociological point of view: the CDF analysts do not even care to provide us with an estimated significance of the bump-like deviation, so I am bound to provide you with the estimate of my carefully trained eye: it cannot be far from a two-standard-deviations effect, give or take a couple of decimal points. Something of the same order of magnitude as this one, which generated a tsunami in the blogosphere two and a half years ago, or this other one (scroll down the page), which only made a few ripples eighteen months ago.
Extraordinary claims require extraordinary evidence. If we had given the same belief to the "no Z'" and to the "Z' at 600 GeV" hypotheses before performing the experiment, we could now conclude that the new physics hypothesis is more likely. However, we have such a solid faith in the standard model, and in the absence of new physics in the dark corners where we have not looked yet, that our "prior belief" does not get much shaken by the slightly better fit that the added Z' signal provides. In Bayesian terms, our "posterior belief" does not differ much from our prior, because we are dominated by prior information.
The analysis does not stop at the plot you can see above. In fact, that is just one of the many which were considered, because the search considered different optimizations of the selection cuts tuned on the different possible masses of the searched Z' boson. Using the absence of a significant signal in the mass spectra, a limit on the existence of the particle has been in the end derived. But that is another story.
So at the bottom of this article I just offer my due congratulations to the authors of the study: Chiho Wang, Byeong Rok Ko, Seog H. Oh, and Jared Yamaoka. I wish they will now take on the orthogonal sample of muon-triggered events, which may double the statistics, and even better, use all the available data that CDF has collected so far. In total, one may hope in a result based on a four-times larger dataset - which would make statistical uncertainties smaller by a factor of two. Who knows ? Maybe a 4-sigma Z' signal is awaiting to be unearthed!
For those among you who are interested in the gory details of this interesting analysis, here is the public note where CDF describes in full their findings. Otherwise, you may choose to visit the public page of the analysis.
Another Would-Be Z' Signal Awaiting Us At 600 GeV