Wednesday, 23 July 2014

Higgs Recap

On the occasion of summer conferences the LHC experiments dumped a large number of new Higgs results. Most of them have already been advertised on blogs, see e.g. here or here or here. In case you missed anything, here I summarize the most interesting updates of the last few weeks.

1. Mass measurements.
Both ATLAS and CMS recently presented improved measurements of the Higgs boson mass in the diphoton and 4-lepton final states. The errors shrink to 400 MeV in ATLAS and 300 MeV in CMS. The news is that Higgs has lost some weight (the boson, not Peter). A naive combination of the ATLAS and CMS results yields the central value 125.15 GeV. The profound consequence is that, for another year at least,  we will call it the 125 GeV particle, rather than the 125.5 GeV particle as before ;)

While the central values of the Higgs mass combinations quoted by ATLAS and CMS are very close, 125.36 vs 125.03 GeV, the individual inputs are still a bit apart from each other. Although the consistency of the ATLAS measurements in the  diphoton and 4-lepton channels has improved, these two independent mass determinations differ by 1.5 GeV, which corresponds to a 2 sigma tension. Furthermore, the central values of the Higgs mass quoted by ATLAS and CMS differ by 1.3 GeV in the diphoton channel and by 1.1 in the 4-lepton channel, which also amount to 2 sigmish discrepancies. This could be just bad luck, or maybe the systematic errors are slightly larger than the experimentalists think.

2. Diphoton rate update.
CMS finally released a new value of the Higgs signal strength in the diphoton channel.  This CMS measurement was a bit of a roller-coaster: initially they measured an excess, then with the full dataset they reported a small deficit. After more work and more calibration they settled to the value 1.14+0.26-0.23 relative to the standard model prediction, in perfect agreement with the standard model. Meanwhile ATLAS is also revising the signal strength in this channel towards the standard model value.  The number 1.29±0.30 quoted  on the occasion of the mass measurement is not yet the final one; there will soon be a dedicated signal strength measurement with, most likely, a slightly smaller error.  Nevertheless, we can safely announce that the celebrated Higgs diphoton excess is no more.

3. Off-shell Higgs.
Most of the LHC searches are concerned with an on-shell Higgs, that is when its 4-momentum squared is very close to its mass. This is where Higgs is most easily recognizable, since it can show as a bump in invariant mass distributions. However Higgs, like any quantum particle, can also appear as a virtual particle off-mass-shell and influence, in a subtler way, the cross section or differential distributions of various processes. One place where an off-shell Higgs may visible contribute is the pair production of on-shell Z bosons. In this case, the interference between gluon-gluon → Higgs → Z Z process and  the non-Higgs one-loop Standard Model contribution to gluon-gluon → Z Z process can influence the cross section in a non-negligible way.  At the beginning, these off-shell measurements were advertised as a model-independent Higgs width measurement, although now it is recognized the "model-independent" claim does not stand. Nevertheless, measuring the ratio of the off-shell and on-shell Higgs production provides qualitatively new information  about the Higgs couplings and, under some specific assumptions, can be interpreted an indirect constraint on the Higgs width. Now both ATLAS and CMS quote the constraints on the Higgs width at the level of 5 times the Standard Model value.  Currently, these results are not very useful in practice. Indeed, it would require a tremendous conspiracy to reconcile the current data with the Higgs width larger than 1.3 the standard model  one. But a new front has been opened, and one hopes for much more interesting results in the future.

4. Tensor structure of Higgs couplings.
Another front that is being opened as we speak is constraining higher order Higgs couplings with a different tensor structure. So far, we have been given the so-called spin/parity measurements. That is to say, the LHC experiments imagine a 125 GeV particle with a different spin and/or parity than the Higgs, and the couplings to matter consistent with that hypothesis. Than they test  whether this new particle or the standard model Higgs better describes the observed differential  distributions of Higgs decay products. This has some appeal to general public and nobel committees but little practical meaning. That's because the current data, especially the Higgs signal strength measured in multiple channels, clearly show that the Higgs is, in the first approximation, the standard model one. New physics, if exists, may only be a small perturbation on top of the standard model couplings. The relevant  question is how well we can constrain these perturbations. For example, possible couplings of the Higgs to the Z boson are

In the standard model only the first type of coupling is present in the Lagrangian, and all the a coefficients are zero. New heavy particles coupled to the Higgs and Z bosons could be indirectly detected by measuring non-zero a's, In particular, a3 violates the parity symmetry and could arise from mixing of the standard model Higgs with a pseudoscalar particle. The presence of non-zero a's would show up, for example,  as a modification of the lepton momentum distributions in the Higgs decay to 4 leptons. This was studied by CMS in this note. What they do is not perfect yet, and the results are presented in an unnecessarily complicated fashion. In any case it's a step in the right direction: as the analysis improves and more statistics is accumulated in the next runs these measurements will become an important probe of new physics.

5. Flavor violating decays.
In the standard model, the Higgs couplings conserve flavor, in both the quark and the lepton sectors. This is a consequence of the assumption that the theory is renormalizable and that only 1 Higgs field is present.  If either of these assumptions is violated, the Higgs boson may mediate transitions between different generations of matter. Earlier, ATLAS and CMS  searched for top quark decay to charm and Higgs. More recently, CMS turned to lepton flavor violation, searching for Higgs decays to τμ pairs. This decay cannot occur in the standard model, so the search is a clean null test. At the same time, the final state is relatively simple from the experimental point of view, thus this decay may be a sensitive probe of new physics. Amusingly, CMS sees a 2.5 sigma significant  excess corresponding to the h→τμ branching fraction of order 1%. So we can entertain a possibility that Higgs holds the key to new physics and flavor hierarchies, at least until ATLAS comes out with its own measurement.


Dipankar said...

Flavor violating decay of the SM higgs in the quark sector seems possible via 1-loop. As an example, SM Higgs can, in principle, decay into a b-sbar pair throuh a W-top loop. Is there any symmetry that prevents this? I don't know how small the branching fraction would be though.

Anonymous said...

Jester, could you elaborate on the specific assumptions you mentioned in the Higgs width measurement?

"Nevertheless, measuring the ratio of the off-shell and on-shell Higgs production provides qualitatively new information about the Higgs couplings and, under some specific assumptions, can be interpreted an indirect constraint on the Higgs width."

Why can the measurement not be considered model-independent?

Jester said...

To interpret it as a width measurement you need to assume that off-shell and on-shell Higgs couplings are the same. As explained in 1405.0285, this assumption is not valid in general; in fact, it's generically invalid in the presence of new large contributions to the Higgs width.

Jester said...

@Dipankar, I can't find in the literature how much is BR(h->bs) in the standard model, but certainly it's small (loop and Cabibbo suppressed) and unobservable.

Anonymous said...

Indeed, it would require a tremendous conspiracy to reconcile the current data with the Higgs width larger than 1.3 the standard model one.

Can you explain this comment? What about the current data suggests this?

Jester said...

It comes from the Higgs signal strength measurements. If the width is increased, the signal strengths in all decay channels are universally suppressed, and at some point become inconsistent with the existing data. To avoid that, one would need to change other Higgs couplings in a correlated way, such that the signal strengths do not change too much. That's what I call the conspiracy.

Anonymous said...

Too difficult for me, but i'm glad you're back.

Michel Beekveld

west said...

While the particulars of the relevant particle physics in this post are beyond me, I know enough to be a bit mystified by the usage of ML estimates and confidence intervals that contain non-physical values. Its not as if the problem of constructing intervals with proper coverage is an unimportant issue in statistics (see Feldman-Cousins). At a gut level, seeing some of those data points on figures for #2 & #5 make me question whether I should take the results seriously at all.

Does anyone have an explanation as to why my skepticism is misplaced?

Alex said...


There are surely people around who know much more about statistics than me and can give you a more in depth answer, but I can tell you this much: it is not nonsensical to include negative signal strengths here. What we are shown here is (more or less) the best fit of #(expected background events) + mu*#(expected higgs events) compared to #(data events). The data can always fluctuate down below the expected background, and thus make the best fit value for mu negative, so there's a real probability that mu<0 is actually measured. Thus, I would say it makes sense to include it in the range of the error bar. If in an extreme case we measure mu<0 with the 2-sigma error bar not even including the 0, it's either 1. the look elsewhere effect, 2. or we can assume that we underestimate the SM backgrounds and/or there is no SM higgs, or 3. there is other weird new physics interfering destructively with the background.

Anonymous said...

" At the same time, the final state is relatively simple from the experimental point of view, thus this decay may be a sensitive probe of new physics."

I wouldn't agree with this statement. Measuring taus is difficult due to their pion decay modes, which are ubiquitous at hadron colliders. In fact tau are probably the most difficult particle to id.

Dipankar said...

Jester, can you give any reference (may be the pdf of a talk) for the updated value of diphoton signal strength from ATLAS?