Dzero published today measurements on W+jet events. One can take many lessons out of this measurement and the comparisons to “SM theory” (Figure 3, second panel). The latter is not a unique object and has to be treated with care. The … Continue reading
The CDF Wjj excess is puzzling in many respects. If a new gauge boson is the culprit, then it has to couple very weakly to leptons. Otherwise, it should have been observed at LEP or the Drell-Yan spectrum at hadron colliders. A Z’ of 150 GeV which decays exclusively to quarks could hide itself very comfortably in the large dijet QCD background at Tevatron and the LHC.
Backgrounds can be tolerated if we lower the collision energy. Awkwardly, the destiny of interpretations of modern era excesses can be decided by experiments which belong to glorious but almost ancient history of particle experimentation, such as the UA2 experiment. Can such exclusions be trusted given the limited knowledge of QCD at the time?
UA2 collected about 11/pb of proton/antiproton collisions at 630 GeV collision energy in 1989 and 1990. With these data, they put stringent limits on the existence of vector di-jet resonances. Their analysis is quite transparent. UA2 selected back to back jets in the central parts of their detector, vetoing very thoroughly events with a third jet. This seems very intuitive and wise when one searches for di-jet resonances. However, the selection and reconstruction of jets would have been very different nowadays, with more sophisticated and theoretically cleaner algorithms.
On the other hand the exposure of the analysis to theory uncertainties is as little as possible. The question that UA2 asked first was how well their di-jet data sample can be fitted by a smooth function and Gaussian bumps. These correspond to the QCD dijet production, the W-boson, the Z-boson, and a hypothetical dijet resonance X. This is a QCD blind question and can be answered statistically.
UA2 assumed a simple fitting function with very few parameters. An overall normalization for the luminosity, three parameters for an exponentially falling QCD background, three parameters for the W-boson Gaussian and two parameters for the X-particle. They could fix very confidently the Z and W Gaussians components from an already very good knowledge of the relative cross-sections, masses and widths of these particles.
A good fit of the data did not require a significant Gaussian component from a Z’ . Then, UA2 proceeded to place upper limits on the total cross-section for such a particle. This last step is however prone to theory and systematic uncertainties.
1. The width of a hypothetical Z’ was fixed from the width of the Z by assuming SM couplings. Most likely, a leptophobic Z’ that could explain the Wjj CDF anomaly is narrower than the Z-boson. But this could only make their limits more stringent.
2. A Gaussian is not exactly how a Z’ resonance appeared in their simulations. The MC spectrum had a rather long tail at lower masses. So, their smooth function parameterization may have been less than ideal. It shifted systematically towards lower values (as compared to MC) the mass peaks of hypothetical Z’ particles. They used an iterative fitting procedure to re-absorb the tail into the smooth QCD background, and estimated the inefficiency of the final Gaussian to capture the full cross-section due to a Z’ with Monte-Carlo. At the end, they made a conservative estimate considering the worst case scenario for this systematic shift of the true peak due to the parameterization.
3. UA2 estimated the acceptance of a Z’ signal due to their selection cuts using a version of PYTHIA written in 87! How accurate is this efficiency, in view of the advances of QCD and Monte-Carlo simulations in the last 20 years? This cannot be answered without a serious study. If they have been correct, then the total cross-section for a Z’ of 150 GeV which decays fully hadronicaly cannot be larger than ~130 pb; a limit which kills many of the Z’ explanations, unless something special happens (eg, purely left-handed couplings which are also different for up and down quarks).
Could they have underestimated the uncertainty of their efficiency? UA2 estimated an efficiency of (18+-1)%. Perhaps, none of us believes that this is extremely accurate. On the other hand, it would need to be wrong by quite a big factor in order for the Z’ explanations of the CDF Wjj excess to thrive unobstructed by this historic experiment.
The announcements of CDF of an excess in the dijet invariant mass distribution for events with a lepton and missing energy has thrown theorists to contemplation. Early models were trying to explain simultaneously the Wjj excess, the top-charge asymmetry, and a small excess in the dijet invariant mass of events with three b-quarks. The models were postulating a Z’. On one hand, this is a simple explanation but on the other hand it was required some magic with non-universal couplings to quarks. Are we asking too much of ourselves to understand three anomalies in one go? Let’s set the bar a bit lower. Can we just explain one anomaly, let’s say the Wjj excess, but without magic couplings and flavor problems? An answer to this question is coming from a very nicely written paper by JoAnne Hewett and Thomas Rizzo; they consider a Z’ which couples to all quark families with universal couplings. They find that family independent couplings are indeed allowed! But, they also find that such a gauge boson cannot be agnostic of the SU(2) group. This makes the building of a complete theory which evades electroweak precision tests very difficult.
The Wjj excess requires ~4pb of a cross-section which is not accounted for by the Standard Model CDF simulations. This is a rather large cross-section. If a light Z’ of 150GeV is responsible, it must have been produced copiously at hadron colliders alone, without being associated with W events necessarily. However, the dijet background of QCD is enormous at the Tevatron and the LHC. One may hope to see new resonances at the tails of inclusive dijet invariant mass distributions, but it is extremely hard to find one where the bulk of the events lie. Essentially, the only significant constraint for such a resonance is placed by the UA1 and UA2 experiments with collisions at a much lower center of mass energy.
According to the paper, the UA2 places a limit of a cross-section of 150 pb for a resonance at 150 GeV. Can a Z’ with universal family couplings survive this limit and at the same time explain the Wjj excess of 4pb at the Tevatron? The answer is yes as long as:
- The coupling to right handed quarks is suppressed
- The coupling to up quarks is significantly smaller than the coupling to down quarks.
Such a Z’ knows about all directions in electroweak physics: left/right and up/down!
It will be interesting to find out how it has managed to spy on the physics that we have been testing vigorously for so many years without us getting a glimpse back at it! According to the calculations of Hewett and Rizzo, such a particle at the LHC may have a very large cross-section for WZ’ production as well, although one cannot say for sure before a complete model is written. Other signatures, such as ZZ’ and gamma-Z’ are expected to have much smaller cross-sections at the LHC by factors of order ~50.
The CMS collaboration performed a first measurement of the cross-section for the production of pairs of W-bosons. This is an interesting process, since W-pairs are generated with well defined rules in the Standard Model theory. These rules have been exhaustively tested. Except one: the production of W-pairs through a Higgs boson. W-pair cross-section measurements will likely lead to the discovery of the Higgs boson, and will be a central source of information when studying its properties.
Currently, data from the sample of 40/pb are not statistically strong to draw any conclusions on the existence of the Higgs boson in the Standard Model. The CMS measurement is nevertheless an important first step in this direction.
W-bosons are unstable particles and decay very quickly into a doublet of quarks or leptons. The CMS measurement is restricted to the case that both W-bosons decay to
an electron or a muon and a neutrino.
The selected signatures are: e+e-, mu+mu- and e+ mu_& e- mu+.
Some cuts are designed to eliminate backgrounds such as Drell-Yan production of a single W or Z boson. Here come some details from the CMS paper on the lepton selection:
muons: pt> 20GeV, |eta|< 2.4
electrons: pt>20GeV, |eta| < 2.5
Isolation: cone DeltaR = 0.3, scalar sum of allE_T deposited in ECAL and HCAL must be less than 15% pt of muon/10% pt of electron.
Missing transverse momentum: E_t,miss > 20GeV, (suppressing Drell-Yan)
-Projected Missing transverse momentum: If missing Et and closest lepton from an angle less than 90 degrees then, projected Et is the missing Et in the direction of the lepton, otherwise it is the full Et,miss. For e+e- and mu+mu-, projEt>35GeV, for e+mu&-e-mu+ projEt>20GeV.
Invariant mass cut Mee, Mmumu not in Mz+-15GeV, and larger than 12 GeV.
A more tricky background is the production of top quarks which also decay to W-bosons. These events are associated with bottom quarks and should be accompanied with b-jets. WW production is predicted to occur with additional jet radiation too, however, this is an effect that occurs at higher orders in perturbation theory. To suppress the background due to top pair production a jet-veto is applied.
Jet Veto: Rejects jets with |eta| < 5 and pt>25GeV. jets constructed with the anti-Kt algorithm. An additional top-veto, based on b-tagging and soft-muon tagging, is implemented.
Predicting what is the efficiency of the jet veto maybe difficult from the theory side, especially when only very small values of jet-pt are allowed (as it indeed occurs
in the CMS analysis). To decrease their reliance on theory, CMS tested the same jet-veto procedure to Drell-Yan production of a Z-boson. One should anticipate that the jet veto efficiency is very similar for Z and WW production, due to an analogous structure of perturbative QCD corrections for the production of a colorless invariant mass from initial state quarks. For Drell-Yan production, the jet-veto efficiency can be experimentally measured with data. Apparently, the Monte-Carlo(s) used for the analysis do a very good job.
CMS improves the theoretical prediction on the WW jet-veto efficiency with the following estimation:
eff_WW_data = eff_WW_MC x (eff_Z_data/eff_Z_MC)
Interestingly, CMS finds that the ratio in parenthesis is close to 1.
The CMS study is a clever demonstration of how a powerful analysis tool
is the data itself.
It will be nice to see theoretically in the future how the jet-veto efficiency
in a potential gluon fusion Higgs production signal correlates to the jet-veto efficiency in Drell-Yan production. For Higgs production, the initial state is different and uncommon features should also be expected. But, may it be not a hopeless undertaking.
CMS measures a total WW cross-section of about 40pb and an uncertainty of roughly 60%, in agreement with the NLO QCD Monte-Carlo MCFM.
The uncertainty is dominated by statistics. Obviously, this will improve very quickly with additional data. WW pair production is a process to keep an eye on for the years to come. Eventually, it will determine the destiny of many of our ideas of physics at the TeV scale.
The experimental collaboration also report the ratio to the W cross-section, as it has been measured in a previous publication by CMS. This also agrees with NLO QCD predictions. In the ratio, they can get rid of the luminosity uncertainty. The systematic uncertainty does not look to be much different. Some more details on how the combination of errors is being made to estimate the uncertainty of the ratio would have been very welcome. Systematic uncertainties due to parton densities could also be reduced in the ratio, but there is no mention of it in the publication.
In the second part of the paper, CMS details some of the implications of their measurement on the phenomenology of the Higgs boson which I find interesting. Not so much for the actual results which inevitably rely on a very small set of data, but mainly for the methodology followed.
