Final Word from the Tevatron on the Higgs Hunt

By John Conway | July 2, 2012 8:37 am

Last September 30, at 3:00 in the afternoon, after a quarter century of operations, the Tevatron collider at Fermilab collided its final proton and antiproton. Since then, physicists from the two big Tevatron experiments CDF and D0 have been analyzing the complete data set, totaling 10 inverse femtobarns, squeezing every last bit of statistical significance in the search for the Higgs boson.

Fourteen years ago I helped lead a study at Fermilab of how much data we would need to collect in “Run 2”, then slated to begin in 2000, in order to observe the Higgs boson. How much data we’d need depended on the mass of the Higgs, because as the mass increases the number of Higgs bosons produced drops, and also the way that the Higgs boson decays changes. We also assumed that we would upgrade the detectors, an assumption that later turned out to be unfilfilled; in 2003 the DOE canceled the $50 million upgrade to the inner silicon tracking detectors. Anyway, our predictions were that to get a golden 5-standard-deviation discovery, it would take on average about 20 inverse femtobarns of collisions. If the Higgs boson was lighter, it would take less data. But could the Tevatron do it?

The early years of Run 2 saw slow progress in bringing up the Tevatron luminosity. Numerous technical problems plagued the machine, but one by one they were conquered and significant data samples began to accumulate. The LHC at CERN, initially slated to start colliding in late 2005, was also delayed due to its own challenges, giving the Tevatron experiments an opportunity to glimpse the Higgs boson. Late in the decade, it began to look like a horse race when in 2008 the LHC suffered a catastrophic failure after its first week of operation. Data were pouring in at the Tevatron, some 5 inverse femtobarns were being analyzed by mid 2009, and it was difficult to predict what would happen.

In late 2009, though, the LHC turned on at low energy, and in early 2010 the machine began colliding at energies three and a half times higher than those at the Tevatron, which were just under 2 TeV (trillion electron volts). That advantage was crucial: more energy means more collision at a given mass for the Higgs bosons, since we are actually colliding the constituents of the protons. But still, by late 2010 the LHC had accumulated only 0.035 inverse fembobarns, far too little to observe the Higgs boson, while the Tevatron was expected to reach 10 inverse femtobarns by the end of summer 2011.

Then the LHC hit the gas, and in 2011 started colliding in earnest, still at 7 TeV, and by summer of last year the LHC experiments had a full inverse femtobarn to analyze, and it became clear that by the end of the year, if the LHC kept going as well, it would amass a sample at least as sensitive, if not more, than the Tevatron experiments would have. And indeed, hints of a signal for the Higgs boson at about 125 GeV began to emerge at the end of the year, with the LHC experiments CMS and ATLAS analyzing nearly five inverse femtobarns.

At the 2012 winter conferences, the CMS and ATLAS experiments showed a dramatically reduced mass window remaining in which the Higgs boson could live, spanning the narrow range from about 115-130 GeV. Due to the peculiarities of the available search channels at the two machines, the lower end of this mass range favored the Tevatron, which is more sensitive to the decays of the Higgs boson to bottom-quark pairs, while the higher end favored the LHC, which is sensitive to the much rarer decays to pairs of photons or Z bosons, both of which give very sharp mass peaks. If the hints of the beast at 125 GeV turned out to be real, then the LHC would win the race for discovery.

And as I said at the outset, the name of the game for the Tevatron was clear: extract as much statistical significance as possible. The analyzers did an amazing job, and today unveiled the final answer: a broad excess in the data, consistent with a 125 GeV Higgs boson decaying to b-quark pairs. Combining both experiments’ data the statistical significance is equivalent to about 2.5 standard deviations at that mass. This is shown in the plot by the fact that the black curve, which shows the probability that the background alone could look at least as signal like as what is observed, dips down to nearly the three-standard-deviation level.

On Wednesday, at CERN, the LHC experiments will reveal the results of combining the large 2011 data sample with an even larger 2012 data sample at a higher energy, 8 TeV. It is obviously a tremendously exciting time – stay tuned!

CATEGORIZED UNDER: Higgs, Science, Top Posts
  • Ripu Jain

    Are we expecting the Higgs to be exactly as predicted or will the particles have some weird unknown properties to it? Does this complete the supersymmetry or does it create more gaps?

  • MKS

    Of course the real reason for the delay is the cognitive dissonance caused by observing the following on the Higgs Boson “Sorry for the inconvenience. Copyright Deity, all rights reserved. And that means you” :3

  • Pingback: Live-Blog zu den Higgs-igen Enth├╝llungen im Juli « Skyweek Zwei Punkt Null()

  • Clayton Brown

    To watch John (and Robin Erbacher, his wife) actually participate in Fermilab’s search for the Higgs, check out (or re-watch) the film The Atom Smashers, our portrait of earlier days in the search. The film chronicles the struggles John, Robin, and others faced as they spun up the Tevatron just as John describes in this post, all the while dealing with budget cuts and a political climate often at odds with research. Don’t miss the 1980s Donahue episode where Leon Lederman attempts to explain the Tevatron to a studio audience. You can check it out on iTunes, Netflix, or Amazon. Here’s the PBS link to the film:

    And here’s where you can get a copy:

  • John Conway

    Ripu: our very first job, once we discover the Higgs boson, is to measure its properties carefully to see if it is the particle predicted by the standard model, or deviates in some way from that. There are many reasons to believe that the Higgs boson is more complicated – there may be more Higgs bosons to find, or what we are seeing may not be a Higgs boson at all.

    If supersymmetry is there in nature, then we’ll definitely have more Higgs bosons out there to discover, but the lightest one can look very standard model like. To really discover supersymmetry, we need to see the supersymmetric partners of the quarks and leptons…that has not happened yet!

  • Oded

    I’m completely confused by this graph. Is it an overlay of 2 different graphs with 2 different y-axis together?

    On the one hand, there is the dashed line which seems to represent “expected results without higgs”, with standard deviation “bars” around it. It is unclear what the “expected results” are (number of collisions, number of detections, whatever). Regardless, whatever the dashed line is measuring, I would figure, the further an “observed” measurement is from the dashed line, that would mean “higher sigma” for real result.

    On the other hand, on the right side, it says “1-sigma”, “2-sigma”, etc.. what’s that??
    On which y-axis is the solid line working? What is going on here?

  • Entropy

    Assuming a given probability distribution, a p-value can be converted into a “sigma” and vice-versa. A p-value says “the probability that the background could have fluctuated this far from its true value is x” while a sigma says “a fluctuation this far from the expectation is a y standard deviation shift.” The probability of shifting by y standard deviations can be converted to a probability (essentially a p-value) for some distribution. For example, for a Gaussian distribution, the actual value will lie within 2 standard deviations of the mean something like 95% of the time, so the probability of background randomly fluctuating up by 2 sigma or more is 2.5%, so the p value is .025.

    “Expected” is basically a measure of how sensitive the Tevatron thinks it ought to be. For example, they should be able to see a roughly 4 sigma deviation if the Higgs had mass 165 GeV (the large sensitivity here is, I’m guessing, because the Higgs is almost exactly twice the mass of two W bosons and so will decay to a WW pair very, very often, and W bosons often decay leptonically, and leptonic decays are comparatively straightforward to observe). In any event, they do not see that deviation and so can exclude a Higgs in that region.

    They expect to be able to see a 2 sigma deviation at 125 GeV if the Higgs exists. Instead they see a 3 sigma deviation. This is consistent with a Higgs, indeed it is consistent with a Higgs with production cross section greater than the SM Higgs.

  • Pingback: Higgs: What’s In A Name? | Whiskey…Tango…Foxtrot?()

  • Oded

    Ahh I see, I completely misunderstood what “Expected” there meant. Usually in these graphs, “expected” is “expected results of measurements, assuming no higgs anywhere”. That obviously made no sense here, hence my confusion…

    Thanks Entropy! (it is a bit odd thanking entropy.. Damn you entropy!! Always moving time forward!!)

  • Chris Greene

    Hi, does anyone know much about the plotting software used to generate those plots? It looks like R, has CERN created their own plot package? Or just cobbled together some ad-hoc scripts?

  • Robin

    Chris: CERN has created their own plot package (used worldwide in particle physics) called ROOT, which interfaces to C++/C and has an imbedded C-interpreter.


Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Cosmic Variance

Random samplings from a universe of ideas.

See More

Collapse bottom bar