Category Archives: LHC News

The Two-Photon Excess at LHC Brightens Slightly

Back in December 2015, there was some excitement when the experiments ATLAS and CMS at the Large Hadron Collider [LHC] — especially ATLAS — reported signs of an unexpectedly large number of proton-proton collisions in which

  • two highly energetic photons [particles of light] were produced, and
  • the two photons could possibly have been produced in a decay of an unknown particle, whose mass would be about six times the mass of the Higgs particle (which ATLAS and CMS discovered in 2012.)

This suggested the possibility of an unknown particle of some type with rest mass of 750 GeV/c².  However, the excess could just be a statistical fluke, of no scientific importance and destined to vanish with more data.

The outlook for that bump on a plot at 750 GeV has gotten a tad brighter… because not only do we have ATLAS’s plot, we now have increasing evidence for a similar bump on CMS’s plot. This is thanks largely to some hard work on the part of the CMS experimenters.  Some significant improvements at CMS,

  1. improved understanding of their photon energy measurements in their 2015 data,
  2. ability to use 2015 collisions taken when their giant magnet wasn’t working — fortunately, the one type of particle whose identity and energy can be measured without a magnet is… a photon!
  3. combination of the 2015 data with their 2012 data,

have increased the significance of their observed excess by a moderate amount. Here’s the scorecard.*

  • CMS 2015 data (Dec.): excess is 2.6σ local, < 1σ global
  • CMS 2015 data (improved, Mar.) 2.9σ local, < 1σ global
  • CMS 2015+2012 data: 3.4σ local, 1.6σ global
  • ATLAS 2015 data (Dec. and Mar.): 3.6σ local, 2.0σ global to get a narrow bump [and 3.9σ local , 2.3σ global to get a somewhat wider bump, but notice this difference is quite insignificant, so narrow and wider are pretty much equally ok.]
  • ATLAS 2015+2012 data: not reported, but clearly goes up a bit more, by perhaps half a sigma?

You can read a few more details at Resonaances.

*Significance is measured in σ (“standard deviations”) and for confidence in potentially revolutionary results we typically want to see local significance approaching 5σ and global approaching 3σ in both experiments. (The “local” significance tells you how unlikely it is to see a random bump of a certain size at a particular location in the plot, while the “global” significance tells you how unlikely it is to see such a bump anywhere in the plot … obviously smaller because of the look-elsewhere effect.)

This is good news, but it doesn’t really reflect a qualitative change in the situation. It leaves us slightly more optimistic (which is much better than the alternative!) but, as noted in December, we still won’t actually know anything until we have either (a) more data to firm up the evidence for these bumps, or (b) a discovery of a completely independent clue, perhaps in existing data. Efforts for (b) are underway, and of course (a) will get going when the LHC starts again… soon!  Next news on this probably not til June at the earliest… unless we’re very lucky!

So What Is It???

So What Is It? That’s the question one hears in all the bars and on all the street corners and on every Twitter feed and in the whispering of the wind. Everybody wants to know. That bump seen on the ATLAS and CMS two-photon plots! What… IS… it…?


The two-photon results from ATLAS (top) and CMS (bottom) aligned, so that the 600, 700 and 800 GeV locations (blue vertical lines) line up almost perfectly. The peaks in the two data sets are in about the same location. ATLAS’s is larger and also wider. Click here for more commentary.

Well, to be honest, probably it’s just that: a bump on a plot. But just in case it’s not — just in case it really is the sign of a new particle in Large Hadron Collider [LHC] data — let me (start to) address the question.

First: what it isn’t. It can’t just be a second Higgs particle (a heavier version of the one found in 2012) that is just appended to the known particles, with no other particles added in.   Continue reading

Is This the Beginning of the End of the Standard Model?

Was yesterday the day when a crack appeared in the Standard Model that will lead to its demise?  Maybe. It was a very interesting day, that’s for sure. [Here’s yesterday’s article on the results as they appeared.]

I find the following plot useful… it shows the results on photon pairs from ATLAS and CMS superposed for comparison.  [I take only the central events from CMS because the events that have a photon in the endcap don’t show much (there are excesses and deficits in the interesting region) and because it makes the plot too cluttered; suffice it to say that the endcap photons show nothing unusual.]  The challenge is that ATLAS uses a linear horizontal axis while CMS uses a logarithmic one, but in the interesting region of 600-800 GeV you can more or less line them up.  Notice that CMS’s bins are narrower than ATLAS’s by a factor of 2.


The diphoton results from ATLAS (top) and CMS (bottom) arranged so that the 600, 700 and 800 GeV locations (blue vertical lines) line up almost perfectly. (The plots do not line up away from this region!)  The data are the black dots (ignore the bottom section of CMS’s plot for now.) Notice that the obvious bumps in the two data sets appear in more or less the same place. The bump in ATLAS’s data is both higher (more statistically significant) and significantly wider.

Both plots definitely show a bump.  The two experiments have rather similar amounts of data, so we might have hoped for something more similar in the bumps, but the number of events in each bump is small and statistical flukes can play all sorts of tricks.

Of course your eye can play tricks too. A bump of a low significance with a small number of events looks much more impressive on a logarithmic plot than a bump of equal significance with a larger number of events — so beware that bias, which makes the curves to the left of the bump appear smoother and more featureless than they actually are.  [For instance, in the lower register of CMS’s plot, notice the bump around 350.]

We’re in that interesting moment when all we can say is that there might be something real and new in this data, and we have to take it very seriously.  We also have to take the statistical analyses of these bumps seriously, and they’re not as promising as these bumps look by eye.  If I hadn’t seen the statistical significances that ATLAS and CMS quoted, I’d have been more optimistic.

Also disappointing is that ATLAS’s new search is not very different from their Run 1 search of the same type, and only uses 3.2 inverse femtobarns of data, less than the 3.5 that they can use in a few other cases… and CMS uses 2.6 inverse femtobarns.  So this makes ATLAS less sensitive and CMS more sensitive than I was originally estimating… and makes it even less clear why ATLAS would be more sensitive in Run 2 to this signal than they were in Run 1, given the small amount of Run 2 data.  [One can check that if the events really have 750 GeV of energy and come from gluon collisions, the sensitivity of the Run 1 and Run 2 searches are comparable, so one should consider combining them, which would reduce the significance of the ATLAS excess. Not to combine them is to “cherry pick”.]

By the way, we heard that the excess events do not look very different from the events seen on either side of the bump; they don’t, for instance, have much higher total energy.  That means that a higher-energy process, one that produces a new particle at 750 GeV indirectly, can’t be a cause of big jump in the 13 TeV production rate relative to 8 TeV.  So one can’t hide behind this possible explanation for why a putative signal is seen brightly in Run 2 and was barely seen, if at all, in Run 1.

Of course the number of events is small and so these oddities could just be due to statistical flukes doing funny things with a real signal.  The question is whether it could just be statistical flukes doing funny things with the known background, which also has a small number of events.

And we should also, in tempering our enthusiasm, remember this plot: the diboson excess that so many were excited about this summer.  Bumps often appear, and they usually go away.  R.I.P.


The most dramatic of the excesses in the production of two W or Z bosons from Run 1 data, as seen in ATLAS work published earlier this year. That bump excited a lot of people. But it doesn’t appear to be supported by Run 2 data. A cautionary tale.

Nevertheless, there’s nothing about this diphoton excess which makes it obvious that one should be pessimistic about it.  It’s inconclusive: depending on the statistical questions you ask (whether you combine ATLAS and CMS Run 2, whether you try to combine ATLAS Run 1 and Run 2, whether you worry about whether the resonance is wide or narrow), you can draw positive or agnostic conclusions.  It’s hard to draw entirely negative conclusions… and that’s a reason for optimism.

Six months or so from now — or less, if we can use this excess as a clue to find something more convincing within the existing data — we’ll likely say “R.I.P.” again.  Will we bury this little excess, or the Standard Model itself?

CMS and ATLAS present their results

CMS results are being presented by Jim Olsen of Princeton University.

CMS has magnet problems this year due to cooling system problems but was able to record 3/4 of the data with the magnet on.

The diboson excess widely discussed this summer is, perhaps not surprisingly, not confirmed.  Same for the old dilepton excesses.

With certain assumptions, limits on gluinos jump from 1.3 TeV – 1.4 TeV to 1.6-1.7 TeV.

Big improvement in limits on “Black Holes” or anything else dramatic at very high energy (as we saw also in my post yesterday about ATLAS multijet events.)

Top-primes — limits jump to about 950 GeV relative to 800, again with assumptions.

Some new limits on invisible particles.  W’ resonances ruled out up to 4.2 TeV if they decay to leptons, to 2.4 TeV if they decay to top quark + bottom antiquark (with assumptions.)  No dijet bumps or other unusual dijet behavior.  No dilepton bumps up to 2.5 – 3.1 TeV for simple assumptions.

Diphotons (with 2.6 inverse fb of data)! (Olsen shows an event at 745 GeV).  All diphoton events used.  Peak?  Yes!!  BUT: local 2.6 standard deviations, and with the look elsewhere effect, only 1.2 standard deviations. Not impressive.   Such a peak is not inconsistent with previous results, but doesn’t look like a signal.  Still… combining old and new data we see a signal at 3 standard deviations local, 1.7 standard deviations globally after look elsewhere effect.

Also the peak is rather ragged, though this doesn’t imply anything in particular; it is worth noting.  If you assume the peak comes from a wider bump, the significance goes down.

Now on to ATLAS, with results presented by Marumi Kado (from the French Laboratoire de l’Accelerateur Lineaire and Orsay).

ATLAS has 1.2-1.5 times more useable data than CMS.  This could be important.

Look for Higgs in four leptons.  Big statistical fluke!  They see fewer events than expected! This is, of course, no big deal… if you expect 6 events it is no surprise if you happen to see 2.

No peak in two Z’s at higher mass (i.e. no heavy Higgs seen.)  Some improvement in searches for Heavy Higgs particles decaying to taus at higher mass.

Limits on gluinos (with assumptions) go from 1.2-1.4 TeV to 1.4-1.8 TeV. (Got an improvement by looking for boosted top quarks in the case where gluinos decay to top quarks.)  Bottom squarks (with assumptions) — limits go from 650 GeV to 850 GeV.

The excess in Z + jets + invisible particles in high energy events remains in Run 2, a little smaller than in Run 1 but still there.  [Run 1: 10 expected, 29 observed; Run 2: 10 expected, 21 observed.] CMS still doesn’t see it.  What’s the story here?

Dijets (as I wrote about yesterday.)  Kado shows the highest-energy dijet event ever observed by humans.  Nothing unusual in photon + jet. Nothing in dileptons — limits on typical Z’ bosons in the 3-3.4 TeV range, W’ decaying to leptons limited up to 4.1 TeV,

DIPHOTONS. Here we go.

A completely generic search for photon pairs; nothing special or unusual.  Looking for bump with narrow width up to large width.  3.6 standard deviations local, global significance is 1.9 standard deviations.  Looks amusingly similar to the first hint of a Higgs bump from four years ago!  Large width preferred, as much as 45 GeV. Local significance goes up to 3.9 standard deviations, 2.3 after look elsewhere.  Mass about 750 GeV.  Hmm.  No indication as to why they should have been more efficient than in Run 1, or why such an excess wouldn’t have been seen at Run 1.

WW or ZW or ZZ where there was an excess in Run 1.  As with CMS, no excess seen in Run 2.  WH,ZH: Nothing unusual.

Ok, now for the questions. The diphoton bump seen, with moderate significance in ATLAS and low significance at CMS, is very interesting, but without more information and more thought and discussion, it’s premature to say anything definitive.

Kado says: Run 1 two-photon data was reanalyzed by ATLAS and it is compatible with the Run 2 bump for large width at 1.4 standard deviations, less compatible for narrow width at more than 2 standard deviations.  They have not combined Run 1 and Run 2 data yet.

Kado says: the diphoton excess events look like the background, with no sign of extra energetic jets, invisible particles, etc; nothing that indicates a signal with widely different properties sitting over the standard two-photon background.  (Obviously — if it had been otherwise they could have used this to reduce background and claim something more significant.)  There are about 40 events in the peak region (but how wide is he taking it to be?) Olsen: CMS has 10 events in the same region, too little to say much.

Conclusion?  The Standard Model isn’t dead yet… but we need to watch this closely… or think of another question.




Exciting Day Ahead at LHC

At CERN, the laboratory that hosts the Large Hadron Collider [LHC]. Four years ago, almost to the day. Fabiola Gianotti, spokesperson for the ATLAS experiment, delivered the first talk in a presentation on 2011 LHC data. Speaking to the assembled scientists and dignitaries, she presented the message that energized the physics community: a little bump had shown up on a plot. Continue reading

First Big Results from LHC at 13 TeV

A few weeks ago, the Large Hadron Collider [LHC] ended its 2015 data taking of 13 TeV proton-proton collisions.  This month we’re getting our first look at the data.

Already the ATLAS experiment has put out two results which are a significant and impressive contribution to human knowledge.  CMS has one as well (sorry to have overlooked it the first time, but it isn’t posted on the usual Twiki page for some reason.) Continue reading

LHC Starts Collisions; and a Radio Interview Tonight

In the long and careful process of restarting the Large Hadron Collider [LHC] after its two-year nap for upgrades and repairs, another milestone has been reached: protons have once again collided inside the LHC’s experimental detectors (named ATLAS, CMS, LHCb and ALICE). This is good news, but don’t get excited yet. It’s just one small step. These are collisions at the lowest energy at which the LHC operates (450 GeV per proton, to be compared with the 4000 GeV per proton in 2012 and the 6500 GeV per proton they’ve already achieved in the last month, though in non-colliding beams.) Also the number of protons in the beams, and the number of collisions per second, is still very, very small compared to what will be needed. So discoveries are not imminent!  Yesterday’s milestone was just one of the many little tests that are made to assure that the LHC is properly set up and ready for the first full-energy collisions, which should start in about a month.

But since full-energy collisions are on the horizon, why not listen to a radio show about what the LHC will be doing after its restart is complete? Today (Wednesday May 6th), Virtually Speaking Science, on which I have appeared a couple of times before, will run a program at 5 pm Pacific time (8 pm Eastern). Science writer Alan Boyle will be interviewing me about the LHC’s plans for the next few months and the coming years. You can listen live, or listen later once they post it.  Here’s the link for the program.

Dark Matter: How Could the Large Hadron Collider Discover It?

Dark Matter. Its existence is still not 100% certain, but if it exists, it is exceedingly dark, both in the usual sense — it doesn’t emit light or reflect light or scatter light — and in a more general sense — it doesn’t interact much, in any way, with ordinary stuff, like tables or floors or planets or  humans. So not only is it invisible (air is too, after all, so that’s not so remarkable), it’s actually extremely difficult to detect, even with the best scientific instruments. How difficult? We don’t even know, but certainly more difficult than neutrinos, the most elusive of the known particles. The only way we’ve been able to detect dark matter so far is through the pull it exerts via gravity, which is big only because there’s so much dark matter out there, and because it has slow but inexorable and remarkable effects on things that we can see, such as stars, interstellar gas, and even light itself.

About a week ago, the mainstream press was reporting, inaccurately, that the leading aim of the Large Hadron Collider [LHC], after its two-year upgrade, is to discover dark matter. [By the way, on Friday the LHC operators made the first beams with energy-per-proton of 6.5 TeV, a new record and a major milestone in the LHC’s restart.]  There are many problems with such a statement, as I commented in my last post, but let’s leave all that aside today… because it is true that the LHC can look for dark matter.   How?

When people suggest that the LHC can discover dark matter, they are implicitly assuming

  • that dark matter exists (very likely, but perhaps still with some loopholes),
  • that dark matter is made from particles (which isn’t established yet) and
  • that dark matter particles can be commonly produced by the LHC’s proton-proton collisions (which need not be the case).

You can question these assumptions, but let’s accept them for now.  The question for today is this: since dark matter barely interacts with ordinary matter, how can scientists at an LHC experiment like ATLAS or CMS, which is made from ordinary matter of course, have any hope of figuring out that they’ve made dark matter particles?  What would have to happen before we could see a BBC or New York Times headline that reads, “Large Hadron Collider Scientists Claim Discovery of Dark Matter”?

Well, to address this issue, I’m writing an article in three stages. Each stage answers one of the following questions:

  1. How can scientists working at ATLAS or CMS be confident that an LHC proton-proton collision has produced an undetected particle — whether this be simply a neutrino or something unfamiliar?
  2. How can ATLAS or CMS scientists tell whether they are making something new and Nobel-Prizeworthy, such as dark matter particles, as opposed to making neutrinos, which they do every day, many times a second?
  3. How can we be sure, if ATLAS or CMS discovers they are making undetected particles through a new and unknown process, that they are actually making dark matter particles?

My answer to the first question is finished; you can read it now if you like.  The second and third answers will be posted later during the week.

But if you’re impatient, here are highly compressed versions of the answers, in a form which is accurate, but admittedly not very clear or precise.

  1. Dark matter particles, like neutrinos, would not be observed directly. Instead their presence would be indirectly inferred, by observing the behavior of other particles that are produced alongside them.
  2. It is impossible to directly distinguish dark matter particles from neutrinos or from any other new, equally undetectable particle. But the equations used to describe the known elementary particles (the “Standard Model”) predict how often neutrinos are produced at the LHC. If the number of neutrino-like objects is larger that the predictions, that will mean something new is being produced.
  3. To confirm that dark matter is made from LHC’s new undetectable particles will require many steps and possibly many decades. Detailed study of LHC data can allow properties of the new particles to be inferred. Then, if other types of experiments (e.g. LUX or COGENT or Fermi) detect dark matter itself, they can check whether it shares the same properties as LHC’s new particles. Only then can we know if LHC discovered dark matter.

I realize these brief answers are cryptic at best, so if you want to learn more, please check out my new article.