Plot Of The Week: Improved Projections On ATLAS Higgs Reach
    By Tommaso Dorigo | November 4th 2010 09:29 AM | 10 comments | Print | E-mail | Track Comments
    About Tommaso

    I am an experimental particle physicist working with the CMS experiment at CERN. In my spare time I play chess, abuse the piano, and aim my dobson...

    View Tommaso's Profile
    The ATLAS collaboration has just released an important study of the sensitivity to a standard model Higgs boson. For the first time precise predictions are made for LHC running at a centre-of-mass energy of 7 TeV (but also 8 and 9 TeV are considered, given the possibility that next year the energy is bumped up a bit), and for most of the sensitive channels together.

    The public document is long and detailed, and I have no time to discuss its intricacies with you here, nor do I believe that you would actually want me to. But I do want to discuss one of the most significant figures in the note. It is shown below.

    In the figure, you see the 95% confidence-level limit that ATLAS predicts it will extract with one inverse femtobarn worth of proton-proton collisions on the Higgs production rate, in multiples of the Standard Model prediction. This is the variable plotted on the vertical axis: a horizontal line at 1.0 indicates where the SM prediction is, of course, such that when the coloured curves sink below it we can say that ATLAS expects to exclude the Standard Model Higgs for that particular mass value (the value plotted on the horizontal axis). An important thing to note: these limits are expected, at each mass value, if the Higgs boson has a mass significantly different, or if it does not exist. If the Higgs is there, the limit gets worse... until one may claim it is actually there!

    Another explanation concerns the brown and yellow areas on the left and on the top of the figure. They display the exclusion already performed at the Tevatron and at LEP II. You are accustomed to see the LEP II limit as a brick wall -a vertical line- but ATLAS took the pain to plot the effective values of the cross section ratios excluded for masses below 114 GeV. Not that this makes much of a difference: if the SM Higgs exists, it cannot have a cross section much smaller than what theorists compute!

    Let us return to the new ATLAS results, however. By examining the curves, which are the result of a careful combination of search limits extracted from half a dozen different Higgs decay channels, one gets to see how wide is the exclusion region at ATLAS reach as a function of the integrated data by the experiment. With the 1/fb predicted to be available at the end of 2011, chances are that the lower limit will be around 128 GeV (black curve intersecting the R=1 horizontal line). But with 2/fb the limit already moves down to 122 GeV, and 5/fb allow ATLAS to exclude all the mass values, given that masses below 114 GeV are already ruled out by the LEP II experiments.

    It is interesting to note that ATLAS predicts to reach these results single-handedly: the inclusion of similar sensitivity curves by CMS would approximately double the effective luminosity, such that the 1/fb prediction for the lower mass limit on the Standard Model Higgs from a LHC combination corresponds to the 2/fb curve for ATLAS alone, and is probably going to be close to 122 GeV.

    And what about a possible first hint of the Higgs ? If the particle exists, a 3-standard deviation evidence may be found by ATLAS: by quoting the paper,

    Evidence at the 3-sigma level could be found in a limited range from 1 /fb, but if two experiments like ATLAS were combined there would be a reasonable chance of 3-sigma evidence over most of the mass range from 131 to 430 GeV.
    If 2/fb at 8 TeV is collected, and expected or planned analysis improvements are implemented, then the median exclusion region covers a range of Higgs boson masses from 114 to well over 500 GeV.
    Interesting! Maybe we will know where the Higgs sits by the end of next year, after all!


    Nice post... I only disagree with your comment on the yellow LEP limit. It *is* important: after all, don't we all know
    that the SM is not the final story? So, we should be well prepared to find a non-SM Higgs and one should look for it in all corners where it could be hiding.

    Hi VA,

    well, I agree that the information plotted here is all valuable, and that the yellow curve is better than a box -actually, the latter is just wrong, of course. However, don't let's forget that the plot assumes SM production AND decay. I know that SUSY may mimic the SM very well, but I doubt we should confuse this plot for a generalized sensitivity one. F

    Very useful article to me, Tommaso. I have been following the LHC Higgs prospects. Thanks!

    It seems that also at the low Hihhs mass the LHC experiments can do 1-2 x SM. The friendly competition with Tevatron will be even more intense, as they have equal chance to reach three sigma, especially if Tevatron runs another three years.

    Sure - especially since, if the Tevatron continues to run, the LHC will most likely do the same.

    What would the curve be for the luminosities collected up to now, if both experiments ATLAS and CMS are added? Is there a chance that the combined curve might be already better than the Tevatron data at least for some parameters?


    As seen in a previous post by Tommaso, each experiment had collected 0.04 fb-1, and since those days let's say 0.05 fb-1 per experiment, which means 0.1 fb-1 together CMS and ATLAS. As today the proton proton collisions stopped at LHC for a few months (while they have heavy ion colliosions) this is the data to be used for analysis for the next months. It is not clear to me if at this point the LHC experiments together could do better than the Tevatron experiments together. Also, I saw in the article that the numbers come after estimated improvements. They will not necessarily be ready in the next months. But I feel LHC is surely catching up with the Tevatron.

    Hi Frank,

    well, the analyses that are used to compute the curves in the plots above have not been run on the data yet AFAIK. I would guess that one just needs to take the 1/fb line and move it up by a factor of sqrt(1/fb / (2x0.05/fb))= sqrt(10). Uncertainties at this stage are largely statistical, so they scale with sqrt(N), N being the data size; therefore, sensitivity scales with signal/error= N/sqrt(N)=sqrt(N) too.

    What we get is, cow more cow less (an Italian way of saying "roughly speaking"), that the LHC experiments would be unable to limit the Higgs even combining refined analyses on the present data at this stage.

    So where is the corresponding CMS analysis??

    There is one - I cannot link it right now, will do so later.
    Great post! Exciting times ahead. :)