Of Particular Significance

The December 2011 Higgs Update

Here is the real-time post from December 13th, 2011, during and following the Higgs update.  Because it was real-time, it doesn’t have the quality control that real articles do, so it’s here for historical purposes only.

[FINAL FINAL UPDATE: My informal so-called `poll’ (which really wasn’t a poll but a gathering of opinions through conversations about the science) was becoming an eyesore to me, as I had worded what I had found very poorly… so I’ve attempted to fix it.]

[FINAL UPDATE: This post was a real-time post, and so it changed as I learned things.  A more stable, complete and probably more accurate post, benefiting from conversations with a number of experimental physicists at CERN, is now available; this post now remains here merely for historical interest.]

[UPDATE: ATLAS and CMS are more compatible than I first believed.  This was due to misreading a plot by 1 GeV — and it matters a bit.  See final conclusions.  This doesn’t change the general feeling that things are inconclusive, but it makes it more likely that, with more data, it will turn into something conclusive down the line.  So I am a bit more optimistic now that the news will eventually be positive, sometime next year.]

[UPDATE: I have been chatting with my colleagues, all particle physicists working at or visiting CERN, and finding out how many say that the evidence presented today convinces them that the Higgs has been found that there is strong evidence for a Higgs signal in the data (I should have written that much more carefully, but was rushing; mea culpa).  So far — Experimentalists — No: 9,  Maybe: 1, and a good one; will explain shortly.  Yes: 1 (Tomaso Dorigo, of Quantum Diaries, is the unique individual so far, (he uses the term “firm evidence”) and I disagree with his assumptions and his conclusions, which in my view mix theoretical bias with experimental data to draw a more positive conclusion about the weight of the evidence — somewhat akin to assuming guilt to help in proving it.)  Theorists — No: 6, Yes: 0.  This is not a scientific poll, just a “poll” of senior scientists I happened to be talking to, and not even a “poll”, since I didn’t ask a prepared question … but anyone who tells you the community as a whole is convinced is either confused or making it up.]

[Update: I should remind everyone that all numbers quoted here are my personal best attempt to copy down the numbers that were presented by the experiments, and to interpret them correctly.  This isn’t trivial to do correctly as the numbers go flying by.  Some parts of the CMS presentation, especially, were very hard to interpret.  All official information comes from the experiments, and should be obtained from them directly, and not taken from this post.]

Well, it’s going to be an interesting day.  It’s a bit of a circus here at CERN, with all the media here, and all the blog posts and articles and YouTube videos and twittering cluttering the web.  It reminds me of the release of a new iPhone.  I was speaking with a senior person here last night who was involved with the discoveries of the W and Z particles at CERN back in the 80’s, and he remarked how things have changed since then.  It was a much more peaceful time to do science at the frontiers.

Based on what I explained to you yesterday, here’s my scorecard for today: (will be updating this as results come  in)

Huge round of applause for accelerator division after comments by the Director General.

ATLAS (talk to be given by spokeswoman Fabiola Gianotti)

Gianotti emphasizes great performance of LHC accelerator and ATLAS detector, and the challenge of having many more collisions simultaneously than were originally planned for (“pile-up”).

  • Are there any significant excesses in the Higgs –> two photon search consistent with a Higgs signal, and what possible Higgs masses do they suggest?  Yes, 126 GeV, exceeds expectation for a Standard Model Higgs (did ATLAS get lucky?)
  • How significant are they statistically both before and after the look-elsewhere effect?  2.8 sigma locally; much more than expected for any Standard Model Higgs signal,  if there were one — BUT ONLY 1.5 sigma after look-elsewhere effect
  • Are the events in any signal distributed as expected (if you restrict the data to photons in certain regions of the detector, or photons which hit something in the tracker, do you still see the signal?)
  • Are there any events in the Higgs –> ZZ –> two leptons+two antileptons search (here lepton=electron or muon) at the same Higgs masses as suggested by the two-photon events? How does the answer affect the significance of the signal(s)?  THREE EVENTS: 123.6, 124.3, 124.6.  A little bit more than expected from any Higgs signal.  And no events nearby, above or below.
  • Assuming the number of the Higgs –> ZZ events is not zero, if we remove one of these events, by how much does the significance of the signal(s) change? For ZZ only, 2.1 goes down to 1.5; that probably reduces the full result by half a sigma, but that’s a guess on my part.
  • Does the tricky and not so reliable Higgs –> WW –> lepton+antilepton+neutrino+antineutrino search increase the significance of the signal(s) appreciably? Apparently no (?); ATLAS only uses  less than half their data in this channel — they are not comfortable with their understanding of missing momentum from the neutrinos
  • What is the total significance of the signal(s) with and without the use of the Higgs –> WW search?  Altogether 3.6 sigma  (2.8 photons, 2.1 ZZ, 1.4 WW); expected for Standard Model Higgs would be 2.4 sigma
  • What is the probability of the signal(s) appearing by accident in a data sample of this size (i.e. including the look-elsewhere effect)?  1% (2.3 sigma)  Not bad at all.  Not overwhelming either.
  • ATLAS excludes 114-115, 135-136 using photons — BUT NOT other regions near and on other side of the little peak at 125!
  • ATLAS excludes down to 135 using ZZ
  • Together ATLAS leaves only 115 to 131 GeV
  • ATLAS does not attempt to exclude more in high mass range

Gianotti adds one expects to need 5 times current data needed to get 5 sigma on ATLAS alone, if there is a Higgs there.

My conclusion: interesting, intriguing, not convincing by itself, not even that compelling yet.  (Remember CDF had four ZZ events at around 327 GeV this year.)  We really have to see the region on either side of 125 excluded before we can really be confident that 125 — 126 is preferred.  Let’s see what CMS has to say.

CMS (talk to be given by spokesman Guido Tonelli):

Funny moment when Tonelli can’t figure out how to get the ATLAS talk off the screen and replace it with the CMS  talk.  CMS, unlike ATLAS, has results both at low mass and at high mass.  Tonelli emphasizes how challenging it has been to do a full analysis on data that was collected as recently as 6 weeks ago.   CMS, unlike ATLAS, uses all its data in all channels, which allows it to study the high mass range;  Tonelli emphasizes that all results are preliminary and subject to some change (as was also surely true for ATLAS).

Tonelli taking his time, showing every single channel that plays a role, even the rather insensitive ones, such as Higgs -> taus and Higgs –> bottom quark/antiquark.  This must mean that there are little excesses everywhere that contribute to the significance of the result that is coming.

To do good Higgs –> photons gets tough with high numbers of simultaneous collisions (must assign the two photons to the correct proton-proton collision, harder than at ATLAS); but efficiency to do this correctly dropped only from 86% to 80%  as data rate skyrocketed from spring 2011 to fall 2011.

  • Are there any significant excesses in the Higgs –> two photon search consistent with a Higgs signal, and what possible Higgs masses do they suggest?  Barely.  Wiggles around 120,  and 123; very hard to read.
  • How significant are they statistically both before and after the look-elsewhere effect?  Couldn’t tell.
  • Are the events in any signal distributed as expected (if you restrict the data to photons in certain regions of the detector, or photons which hit something in the tracker, do you still see the signal?)  Tonelli claims yes… too fast to read.
  • Are there any events in the Higgs –> ZZ –> two leptons+two antileptons search (here lepton=electron or muon) at the same Higgs masses as suggested by the two-photon events? How does the answer affect the significance of the signal(s)?  Two events around 125, three around 119-120, two near 130.  A bit higher than background expectation.
  • Assuming the number of the Higgs –> ZZ events is not zero, if we remove one of these events, by how much does the significance of the signal(s) change?  A lot; not sure how much.
  • Does the tricky and not so reliable Higgs –> WW –> lepton+antilepton+neutrino+antineutrino search increase the significance of the signal(s) appreciably?  Along with others, yes.
  • What is the total significance of the signal(s) with and without the use of the Higgs –> WW search?  2.6 sigma max across the region.
  • What is the probability of the signal(s) appearing by accident in a data sample of this size (i.e. including the look-elsewhere effect)? 1.9 sigma, which isn’t great.
  • CMS excludes 132-238 using Higgs –> WW, expected to get 129 (can bring the exclusion down to 129 [with 127 expected] with very specialized analysis, hard to know whether to believe it — theory errors?)
  • CMS Higgs –> ZZ excludes 134-158, as well as 340-460.
  • Altogether CMS combining everything excludes 127-600 (600!!!) but expected to go down to 117.
  • Cannot exclude between 115 and 127

In my view, completely unconvincing on its own. 120 GeV is just as good as 123 GeV, and both are unconvincing since the whole region between 115 and 127 is not excluded yet.  Also, CMS had to combine many channels together to get their significance.  Neither Higgs –> photons nor Higgs –> ZZ , or even the two together, make a clear case.

Taken together (a combination which must be done by others, since I believe this has not been formally done by the experiments and will not be presented):

  • Are any/all of the ATLAS and CMS signals roughly consistent with each other?  Roughly.  Not spectacularly.  ATLAS and CMS have very different ZZ signals.  The photon signals do not really line up
  • Do the signals appear at the same possible Higgs mass?  Not as bad as I read initially; the difference is 2 GeV, and that’s not too bad.  Not as close as I would like to see.  CMS has 123 I MISREAD the CMS plot initially: when you combine two photon with ZZ, CMS prefers 124, not 123. and is not so consistent with 126, which is what ATLAS has.  I am not sure ATLAS is consistent with CMS’s 123 yet, or how consistent both are with 124-125.
  • Are they of comparable size?  ATLAS has a larger two-photon signal; either they got lucky and saw more signal, or they got lucky and saw more background in this region.
  • What is the significance of the combination of the signals in Higgs –> two photons?  Guessing for now: Combination will probably give us something broad, not a very narrow peak. reasonable.
  • How does it change if Higgs –> ZZ –> two leptons + two antileptons is included?  Guessing for now: Significance will go up, but the rate will be in excess of Standard Model Higgs rate.
  • What is the probability of the signal(s) appearing by accident in a data sample of this size (i.e. including the look-elsewhere effect)?  Not small enough yet.
  • (Full combination of results, including Higgs –> WW, cannot be reliably performed without knowledge of precisely how the measurements were carried out, and can only be done by the experimenters themselves.)
  • What is the probability of the signal(s) appearing by accident in a data sample of this size (i.e. including the look-elsewhere effect)?  ???

Director General conclusion: Standard Model Higgs is neither found nor excluded yet;    “stay tuned for next year”.

Leave a Reply

Search

Buy The Book

A decay of a Higgs boson, as reconstructed by the CMS experiment at the LHC