A Discrepancy to Keep an Eye On

Today (as I sit in a waiting room for jury service) I’ll draw your attention to something that has been quite rare at the Large Hadron Collider [LHC]: a notable discrepancy between prediction and data.  (Too rare, in fact — when you make so many measurements, some of them should be discrepant; the one place we saw plenty of examples was in the search for and initial study of the Higgs particle.)  It’s not big enough to declare as a definite challenge to the Standard Model (the equations we use to describe the known particles and forces), but it’s one we’ll need to be watching… and you can bet there will be dozens of papers trying to suggest possibilities for what this discrepancy, if it is real, might be due to.

The discrepancy has arisen in the search at the CMS experiment for “multileptons”: for proton-proton collisions in which at least three charged leptons — electrons, muons and (to a degree) taus — were produced. Such events are a good place to look for new phenomena: very rare in the Standard Model, but in the context of some speculative ideas (including the possibility of additional types of Higgs particles, or of superpartner particles from supersymmetry, or new light neutral particles that decay sometimes to lepton/anti-lepton pairs, etc.) they can be produced in the decays of some unknown type of particle.

At CMS, as reported this week at this year’s Supersymmetry conference, the number of events with

  • an electron/positron pair or a muon/antimuon pair that probably didn’t come from a Z particle,
  • plus a tau lepton that decays to hadrons,
  • plus another electron or muon or tau,
  • with the total energy in the leptons/antileptons and any additional jets being rather small by LHC standards
  • along with some amount of unbalanced (“missing”) momentum from undetected particles (perhaps neutrinos, or perhaps something unknown, if we’re very lucky; note taus always produce neutrinos in their decays, so we expect some missing momentum)

is a bit too large. Note that taus can decay both to hadrons or to a muon or electron, so it’s possible that some of these events are actually from a tau/anti-tau pair plus an electron/positron or muon/anti-muon pair. As far as I can tell at present, CMS hasn’t provided enough detail for us to say much more at this point. 

CMS results for events with four charged leptons, showing an excess in events with a single hadronic tau and low total energy.  There is an excess for various amounts of missing transverse momentum from undetected particles.

CMS results for events with four charged leptons, showing an excess in events with a single hadronic tau, no b jet, low observed total energy (H_T), and no lepton/anti-lepton pair consistent with being decay products of a Z particle (off-Z).  Moderate excesses are seen for various amounts of missing transverse momentum (E_T^miss) from undetected particles.  Each entry contains the number of events observed (at left) and the number predicted +- the uncertainty in the number predicted (at right).

It’s interesting, however, that CMS actively pointed out this excess, rather than waiting for us to notice it.  The excess is somewhere between 1.5 and 2.5 standard deviations (where 3 would be striking and 5 definitive), depending on how you treat the issue of the look-elsewhere effect.  I quote from CMS: “The three relevant entries in Table 2 refer to an observed (expected) yield of 15 (7.5±2), 4 (2.1±0.5),  and 3 (0.6±0.24) events for the three different ETmiss bins. The probability for an expected yield of 10±2.4 to fluctuate to 22 or more is about 1%. As this analysis includes 64 sets of channels like this, we estimate our chances to see one such fluctuation at about 50%. Alternatively, taking correlated systematics into account, the joint probability to observe three fluctuations like this in one of the 64 channels is about 5%. “

Since multileptons are known to be a classic way to look for supersymmetry (though remember there are many other possible ways to get multileptons!), CMS has provided a possible (but certainly not unique!) interpretation of this search (not of the excess events!) in terms of such particles, specifically sleptons (superpartner particles of electrons or muons or taus.) What it shows is that the limits that they can actually put on this possibility are quite a bit weaker than they would expect, given this amount of data, and that the excess events could potentially be explained by sleptons between 100 and 200 GeV. Again, it’s important to keep in mind that this discrepancy, even if due to a real effect, may have absolutely nothing to do with sleptons or anything supersymmetric at all. There are many possibilities to consider.

I should also perhaps draw your attention to Patrick Meade’s talk at the SEARCH 2013 workshop last week, concerning discrepancies in events ascribed to pairs of W and/or Z particles.  The problems he talked about there might not be entirely disconnected from the discrepancy we’re dealing with today.

A natural question, of course, is “What does ATLAS observe?” Well, on the one hand ATLAS doesn’t observe any notable excess in events of this type, but on the other hand, they do the analysis quite differently, and so there’s an issue of comparing apples and oranges. Only with some work that I haven’t yet done can one tell whether ATLAS’s lack of a moderate excess is really evidence against the discrepancy at CMS being from a real physical effect.  For all I know right now, it might be that the effect is real and just slipped by ATLAS’s methods.

One bit of weak evidence against this being a real new phenomenon is that there’s no corresponding excess in the most promising three-charged-lepton events at CMS. Specifically, in events similar to ones discussed above except that they lack a tau decaying to hadrons, and focusing especially on events with large missing transverse momentum where the backgrounds are smallest, there’s no excess in the CMS data. You might have hoped that there would at least be a small excess here, since sometimes the hadronic tau might not be detected, and sometimes the tau will decay to a low-energy muon or electron, giving events with four electrons, positrons, muons or anti-muons, in which it is quite common, when the leptons have low energy, that only three are detected.   But again, this isn’t a strong statement at this point.

CMS results for three leptons and low amounts of total energy [actually, transverse momentum] in leptons and jets.  No significant excesses are seen in channels where one might have expected them, given the results in the earlier table.

CMS results for three leptons and low amounts of total energy [actually, transverse momentum] in leptons and jets. One might have hoped for an excess in the first numerical column, fourth and fifth rows, or perhaps the seventh and eight rows, but one sees nothing dramatic there.  Other natural places to look for discrepancies also do not show anything compelling at this time.

In any case, we don’t have a smoking gun here, nor do we have smoke drifting out of multiple cracks, so it’s too early to do anything but what we often have to do: either wait for more data, or look for more wisps of smoke hiding somewhere in the ATLAS or CMS data.

7 responses to “A Discrepancy to Keep an Eye On

  1. ….. The jury remains out! :)

  2. It is as noteworthy as the discrepancy in the h->WW signal strength at CMS. I hope they materialize in something deeper.

  3. Is there some reason that some seemingly significant events were flagged while the others were?

    The second from the top entry in the same column: observed 2; expected 0.7+/-0.33 seems almost as significant as the other data points.

    There are also a couple of similarly high excesses in the column to the left of the column with the circled entries – the one immediately to the left of the 15 observation entry that is circled, and the entry in the same column that reads 2 observed; expected 0.18 +/- 0.13 which lacks a corresponding tau entry, both of which are also “off-Z” results with no b jets and low observed total energy.

  4. “The excess is somewhere between 1.5 and 2.5 standard deviations (where 3 would be striking and 5 definitive), depending on how you treat the issue of the look-elsewhere effect.”

    Is it even worth talking about a discrepancy that falls below 3 standard deviations?

  5. Pingback: Word of the Week: Discrepant | Nigel's random thoughtsNigel's random thoughts

  6. Pingback: Allgemeines Live-Blog vom 23. – 29. August 2013 | Skyweek Zwei Punkt Null

  7. History is full of these little weird fluctuations. For example see here the first discovery of “some Higgs” at 5sigma significance:

    http://www-public.slac.stanford.edu/sciDoc/docMeta.aspx?slacPubNumber=SLAC-PUB-3380

    So I am sorry, but you just need to move on, there is nothing of interest here yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s