Winter 2017 LHC Results: The Higgs Is Still There, But...
    By Tommaso Dorigo | April 4th 2017 05:32 PM | 3 comments | Print | E-mail | Track Comments
    Snow is melting in the Alps, and particle physicists, who have flocked to La Thuile for exciting ski conferences in the past weeks, are now back to their usual occupations. The pressure of the deadline is over: results have been finalized and approved, preliminary conference notes have been submitted, talks have been given. The period starting now, the one immediately following presentation of new results, when the next deadline (summer conferences!) is still far away, is more productive in terms of real thought and new ideas. Hopefully we'll come up with some new way to probe the standard model or to squeeze more information from those proton-proton collisions, lest we start to look like accountants!
    This is also a good moment to have a look at the most exciting results produced by the scientific collaborations. You can do so by browsing the "experimental summary talk" presented at the Moriond conference last week (all talks are at this link), if you have the knowledge and brain equipment to venture into a hostile landscape of tons of graphs and figures with little captioning. Or we can do it bit by bit, by examining some of the interesting things together. 

    One thing I would like to look at today is the reconstructed mass distributions of Higgs candidate events produced by ATLAS and CMS with the 2016 data. The 36 inverse femtobarns of data collected at 13 TeV last year have more statistical power than all of the data collected until then by the LHC experiments, so there is definitely something new to learn from those distributions... Were it not for the fact that not all data has been analyzed yet for all analyses. Oh well, they're a lot of data anyways!

    The first one, shown above, is the diphoton mass distribution in the vicinity of 125 GeV, produced by ATLAS (only 14 inverse femtobarns here); the lower panel shows the background-subtracted residuals, where you can better see that the bumplet is still there, and it is definitely not so small anymore: it consists in several  hundred events. A statistical analysis of those events can provide a wealth of information about production modes, spin-parity assignments of the Higgs boson (in case you still doubted it is a scalar particle, e.g.), and other issues we need to investigate in more detail.

    The corresponding plot from the CMS collaboration is shown below. I venture to suggest that it is even prettier than the ATLAS one! One thing to note in this pair of graphs is that they show, in the y axis, a "sum of weights" proportional to the signal to noise ratio of the category from which the event is taken. Each analysis is the combination of several distinct searches for photon pairs. Depending on the details of the events containing the Higgs decay signature, the events get categorized in disjunct classes. The fact that the signal-to-noise ratio is different in the various classes (because backgrounds are comparatively larger or smaller) makes the procedure of "divide et impera" advantageous, as it produces a more significant result, in statistical terms.

    About the four-lepton distribution: this is a lower-statistics search, but the signal in this case is much, much cleaner. Here we not only have four charged leptons of high momentum to select the data with: one of the two pairs of leptons must yield the Z boson mass, as the Higgs boson produced them via the decay H -> Z Z* -> 4 leptons. Note that unfortunately one Z has an asterisk: it is "off-mass shell", meaning that it does not have the nominal mass of the Z boson (91 GeV). In fact, the Higgs boson's mass is only 125 GeV, and this is smaller than twice the Z mass: the decay to two on-shell Z's would violate energy conservation. 

    The peak in the CMS graph shown above, obtained with 2016 data, is impressively nice. Unfortunately I could not find the corresponding ATLAS peak - will look it for you later. Ah, in case you wondered: yes, there are two peaks - the one on the left is the background coming from Z decays that produce four leptons through an internal photon conversion. 

    Well, so - from a qualitative standpoint, nothing in the above graphs is news. The Higgs boson continues to decay to photon pairs and to four leptons. But how about the other, more difficult decay modes ? In particular, how about the decay to b-quark pairs, which the standard model predicts to be the most frequent one (every second Higgs boson should decay into two b-quarks, in fact) ?

    Surprisingly, that decay mode sees the collaboration agonizing in the search of a conclusive proof that it does take place at the predicted rate. The signature is very hard to put in evidence because b-quarks are a very common outcome of a proton-proton collision, unlike leptons or photons. Despite the search in a number of different kinematical and final state configurations accompanying the wanted H->bb decay - top quark pairs, vector-boson tagging forward quarks, single top quarks, vector bosons, etcetera - that decay mode has not been proven to exist yet. If you look at the summary table below (extracted from Bill Murray's slides), which lists the current values of the measured ratio μ between observed and predicted H->bb decay rate in various channels, you immediately see that there is a dearth of such a signal.

    A layperson's average of the five determinations would turn out a result of 0.2+-0.4, which is two sigma-ish away from the expected value (1.0). Is there something fishy going on ? Well, relax, no - all is good. I do believe it is a downward fluctuation. If you believe otherwise, how about a $100 bet that the H->bb branching ratio will end up being what the SM predicts ? My bank account needs some cash injection...


    Tommaso Dorigo is an experimental particle physicist, who works for the INFN at the University of Padova, and collaborates with the CMS experiment at the CERN LHC. He coordinates the European network AMVA4NewPhysics as well as research in accelerator-based physics for INFN-Padova, and is an editor of the journal Reviews in Physics. In 2016 Dorigo published the book “Anomaly! Collider physics and the quest for new phenomena at Fermilab”. You can purchase a copy of the book by clicking on the book cover in the column on the right.


    Tommaso Dorigo is an experimental particle physicist, who works for the INFN at the University of Padova, and collaborates with the CMS experiment...

    View Profile


    Tommaso, you said "... The peak in the CMS graph shown above, obtained with 2016 data, is impressively nice. Unfortunately I could not find the corresponding ATLAS peak - will look it for you later ...". ATLAS has shown a corresponding peak in Figure 4(a) of ATLAS-CONF-2016-079 which peak is very much like the CMS peak but ATLAS-CONF-2016-079 is dated 4 August 2016 and so is based on only 14.8 /fb of data which is only the first 1/3 or so of the full 2016 LHC run whereas the CMS graph is based on 35.9 /fb of data which is the full 2016 LHC run. As far as I know, ATLAS-CONF-2016-079 has not yet been updated which is hard for me to understand because it seems to me that the basic analysis work has already been done back in August 2016 to produce ATLAS-CONF-2016-079 and it seems to me that all ATLAS has to do is to plug in the full 2016 LHC run data into the analysis structures that ATLAS obviously has had in hand since August 2016. Tony
    Hello Tony,
    thanks - I suspected the situation was the one you depict. Let's wait and see :)
    Hi T, I was wondering what happened to the Higgs boson in the 13 TeV data. Thanks for your note.