Banner
    Plot Of The Week - A SUSY Higgs At 150 GeV ?
    By Tommaso Dorigo | June 15th 2010 10:35 AM | 12 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 CDF Collaboration has recently produced results of a search for Supersymmetric Higgs bosons in events with three or more bottom-quark jets. Here I wish to give just the highlights of this analysis, but before I do I will try to spend 5' on making sense of the previous sentence.

    In good order, below I explain first of all 1) what is CDF, 2) what is the Tevatron, 3) what are Higgs bosons from Supersymmetry, 4) how can these be sought with bottom quarks, and 5) what are bottom quark jets. After I am done with these five explanations, those of you who are still here will no doubt appreciate the results I am showing today. Can I make it in 5 minutes ? Sure, but can you read forty lines of text in the same amount of time ?

    Five simple explanations in five minutes of your time

    Minute 1 - First of all, CDF. It is arguably the longest-lasting physics experiment in the history of physics, with a project dating back to 1979, and first data collected in 1985. CDF is a general purpose experiment: a 5000-Ton device designed to detect and measure all products of proton-antiproton collisions. It is built like an onion around the interaction point, from which scores of particles created in the collisions fly out at light speed three million times a second. At the core of the onion there are many precise layers of silicon detectors and sensitive wires that record the passage of individual charged particles within a strong magnetic field: from the lining up of many measurement points, the helical trajectories of charged particles bent by the magnet are reconstructed; the bending is inversely proportional to the momenta, so the latter can be determined from the former. Outside, layers of heavy metals interspersed with detecting elements allow to determine the collective behavior of streams of particles, destroying them and measuring their total energy. A picture of CDF during the installation of the silicon detector is shown on the left.

    Minute 2 - The CDF experiment collects the results of high-energy interactions produced when protons and antiprotons are accelerated to light speed and then aimed at one another by the Tevatron accelerator at Fermilab (see an aerial picture of the site of the lab on the right). The Tevatron is a synchrotron - it is built with powerful magnets curving the orbit of the particles, alternated with sections where these are increasingly accelerated by electric fields. A synchrony is necessary because as particles increase their energy, magnets need to increase their power accordingly, to keep them on the same curved orbit. After being injected and accelerated, protons and antiprotons are brought to collide at 2 Tera-electronvolt energy, the now second-highest energy available for particle physics studies (7 TeV collisions have started to be produced at CERN ten weeks ago).  In such energetic collisions, new states of matter can be produced. Higgs bosons, maybe ?

    Minute 3 - The Higgs boson is a particle predicted over forty years ago by theorists, to explain the phenomenology of fundamental interactions among elementary particles. A Higgs boson must exist if the Standard Model (SM), our currently accepted theory of nature, is correct. However, at least five different such particles must exist if the world includes not only all the particles we have discovered in the last thirty years, but also a whole new set of bodies with similar properties, a "mirror world" of Supersymmetric (SUSY for insiders) particles. One of these five Higgses, in particular, should be quite similar to the Standard Model one, but it might be produced in larger numbers. So the Tevatron, while still not sensitive enough to detect SM Higgs bosons, might have a shot at discovering a SUSY Higgs!

    Minute 4 - One of the ways by which neutral Higgs bosons might be produced in larger numbers than SM ones is by means of "associated production" with bottom quarks. This is because a SUSY Higgs might have a large "coupling" with b's, i.e. attach more readily to those quarks, and would thus be readily radiated by a energetic bottom quark. And not just that: if the b-H coupling is large, then the Higgs prefers to decay to pairs of bottom quarks, too. This makes it convenient to search for events containing at least three independent bottom-quark jets: one or more from the hard collision, and two more from the decay of the Higgs.

    Minute 5 - You might not know what a bottom quark is: in that case, look at the picture on the left, which summarizes the elementary particles of nature. Bottom quarks, labeled "b", are the second-heaviest quarks in the collection of six we know. Quarks are the elementary constituents of protons and neutrons, as well as of many other short-lived particles; particles composed of quarks are called hadrons. When an energetic collision between protons creates a bottom quark of high energy, the latter moves away from the collision point by fragmenting into a collimated stream of hadrons. This is what we call a "jet". By measuring jets in the CDF detector, we infer the energy of the originating quark.

    ... And we are done! It was not so hard, was it ? If you did not know anything about these five things, and you got this down in the reading, please drop me a line and let me know what you found hard to understand, how long you took to read the text above, and whether you considered dropping out. But later. Now we have something better to do - we can try and figure out what CDF did with the data it collected since 2002, and what is the rather surprising result they have obtained.

    A hint of Supersymmetric Higgs bosons

    The CDF analysis, which is an update of a former search performed with less data, takes events with at least three jets of hadrons, and selects those where all jets contain a signal of having likely originated from a bottom quark. Since the probability that a generic jet is originated by a bottom quark is small -of the order of a few percent-, the requirement kills backgrounds quite effectively.

    Bottom-quark jets can be identified thanks to the fact that among the hadrons they originate there is one so-called B hadron: these particles may travel a few millimeters away from the interaction point before decaying, and this crucial fact allows their identification, thanks to precise silicon tracking detectors contained in the core of CDF: several particles can be seen to radiate from the point of decay of the B hadron  instead than from the collision point. The point of convergence of these secondary particles is called a "secondary vertex".

    Even after selecting jets wherein a secondary vertex is spotted, one cannot be sure that the jet is a bottom quark one. Lighter quarks may pollute the selection, but there is an additional characteristics that allows a further discrimination: the total mass of the tracks radiating from the vertex. B hadrons are heavy -they weigh over five times more than a whole proton- so they tend to yield much higher total mass. This fact is exploited by the search to discriminate b-quark backgrounds from non-b-quark ones. Below you can see the different vertex mass for jets originated by light quarks (black), charm quarks (in red), and bottom quarks (in blue). Since there are three b-tagged jets, the three values of vertex mass are combined in a single variable "x_tags", as shown in the cartoon below the graph.



    A Higgs boson decaying to two b-quarks would produce two energetic jets, but we do not know whic h jets to pick to try and reconstruct the Higgs mass. If we could always choose the two right ones, the resolution on the Higgs mass would help a lot in discriminating it from backgrounds, since backgrounds do not "resonate" -the mass does not show a peak. The analysis makes an educated guess of which jets should come from the Higgs: the two highest-transverse-energy ones. This is a sensible choice, and the resulting resolution on the dijet mass is good enough to provide discrimination from backgrounds. The shape of the invariant mass of the two leading-transverse-energy jets for Higgs bosons of different masses is shown below.



    It is clear that the mass resolution is not excellent; however, these distributions are still peaking well enough that they provide the necessary discrimination power with respect to all backgrounds.

    In the end, a two-dimensional fit is performed using the vertex mass of b-jet candidates and the dijet mass. Backgrounds are due to quantum chromodynamical processes, whereby two hard jets are produced by the main proton-antiproton collision, and two more are produced by colour radiation. Different combinations of b-quark production processes yield different shapes in the dijet mass distribution, such that attention is required when computing the total expectation from Standard Model proceses. The authors took great care to model the different contributing background shapes, basing their estimates on data with and without secondary vertex tags.

    And what are the results ? Results show that the data appears to be well modeled by the sum of backgrounds. However, an even better fit is produced if together with backgrounds, a significant Higgs decay signal is added to the mix! Below you can see the two different ways by which CDF data may be interpreted: the first shows backgrounds alone, the second shows a Higgs contribution, with a 150 GeV mass hypothesized for the supersymmetric particle.





    Now, in statistical terms, we have two competing hypotheses, H_0 and H_1, to interpret our data. H_0 is "no Higgs", H_1 is "a Higgs of 150 GeV is present in the data". By comparing the quality of the fit in the two hypotheses, one may derive the significance of the observation -how much do the data "prefer" the H_1 hypothesis. Please note that since H_1 is a nested hypothesis within H_0, H_1 will always fit the data at least as well as H_0: H_1 in fact reverts to H_0 when the assumed number of Higgs candidates is brought to zero. Because of that, the likelihood of the H_1 fit L(H_1) is always larger than the likelihood of the H_0 fit L(H_0).

    A simple calculation allows to derive the significance of the Higgs hypothesis from L(H_1)/L(H_0), the ratio of the two likelihood values. The significance appears to exceed slightly 2 standard deviations. Yes, just another two-sigma effect! The standard model has survived much higher-significance claims in the past, and my opinion is that it will survive this one, too. Regardless of our personal bias, the observation is interesting, and the tentative signal requires our attention. We need to improve this search, add more data, and see if those two-sigmas become zero or four!

    Below I show one more technical plot produced by the analyzers: the exclusion region of the parameter space of supersymmetry that the search allows. Two of the SUSY parameters are chosen to determine what the data can rule out: a variable called "tangent beta", the tangent of an angle which is related to how strongly the Higgs "couples" to bottom quarks, and the mass of one of the Higgs bosons (the mass of the five Higgs bosons are related to one another, and to plot the data the A Higgs boson is usually chosen as SUSY parameter).



    In the exclusion plots, a wide blue band (centered on the hatched limit line) is overlaid to the actual line showing the boundary of the excluded region (excluded values of the parameters lie above the black thick line). The band shows the range of possible exclusion region boundaries that the analysis was expected to produce, at 1-sigma (dark blue) and 2-sigma (light blue). The band gives a feeling of where the data fell -whether it allowed a more stringent limit than expected, or whether it failed to limit the Higgs boson presence in the data. In the CDF analysis a 150 GeV Higgs boson is allowed by the data, and this reflects in the limit being weaker than expected for that value of mass. This is shown clearly by the departure of observed and expected limit for 150 GeV mass.

    Comments

    lumidek
    Your choice of the topic is highly appreciated. Still, what is the ability to find a SUSY Higgs good for if the SUSY Higgs has to be 130 GeV or lighter, or is that wrong?
    rholley
    Me no like-um acronym, so searching Wikipedia I found that CDF stands for Collider Detector at Fermilab.

    There I read that the “onion” consists of 7 layers (#1 being the beam pipe, #7 the muon detector.)  I guess Wikipedia has got it right, but is that so?
    Now to the article.  
    The standard model has survived much higher-significance claims in the past, and my opinion is that it will survive this one, too.
    Which seems to imply that a 4-sigma deviation would hit the Standard Model hard.  Is that the case?  Does a 150 GeV Higgs go with or against the Standard Model?


    Today on SciBlog we also have WMAP: No Dark Matter Or Dark Energy?!  Perhaps there is going to be a flurry of head-scratching quite soon.


    Robert H. Olley / Quondam Physics Department / University of Reading / England
    dorigo
    Hi Robert,

    Wikipedia considers "layer" a whole detecting element (like "calorimeter system" is a layer). In my explanation of the inner tracker I considered a "detecting layer" one physical surface where the particle leaves a hit, like a silicon wafer (300 microns thick).

    Cheers,
    T.
    dorigo
    Hi, here the cross section x branching ratio limits are interpreted in a search not for the light MSSM Higgs (which would have to be lighter than 130 GeV), but for the ``heavy'' Higgs bosons, A and H. M_A is a free parameter of the model and can easily range between 100 GeV (or so, where the LEP bounds come in) up to a few TeV. In this part of the parameter space one finds M_A \approx M_H, and also the coupling to b quarks are very similar. So the ``signal'' would correspond to the combined A/H production and decay (which, of course, was taken into account in the CDF analysis). On the other hand, the very large value of tan_beta corresponding to the ``excess" looks very ugly. tan_beta larger than 60 or so easily leads to trouble like non-perturbative bottom Yukawa couplings or RGE running to a GUT scale does not work anymore. Cheers, Sven
    lumidek
    Thanks! It would surely be interesting for the light and heavy Higgses to be this close to one another....

    Still, do I understand well that the peak around 140-160 GeV is not unexpected because it reflects that it's easier to distinguish the Higgs decays for this mass range - than for more realistic ranges such as 115-130 GeV? Isn't this higher "resolution" for the mass range around 150 GeV the main reason why the Tevatron could have excluded the Higgs at 165 GeV or how much it was?


    dorigo
    Hi Lubos,

    The Higgs (a SM one) can be distinguished better if it has a mass around 165 GeV because it decays favourably to W pairs, a channel with smaller backgrounds. At 130 GeV, the SM decay to look for is two b-quark jets as here, and there it is hard. The problem of the SM search is that there is no free parameter (tan beta) with which to "cheat" inflating the coupling to b-quarks and thus the cross section.

    In this MSSM search the Higgs is looked for in the bb decay final state across the full mass range, because the larger coupling to b-quarks granted by a large tan(beta) allows it.

    The fact that the analysis sees a bump at 150 has nothing to do with resolution. Resolution deteriorates as the mass increases, as the plot above shows; however, backgrounds peak at low mass so there is an advantage in the higher-mass points.

    Cheers,
    T.

    PS the previous message was from Sven Heinemeyer, who could not post his comment here for some glitch in the system.
    lumidek
    Dear Tommaso,...

    well, yes, the WW decay is surely easy to study. But isn't it true that even when the (light) Higgs mass is 150 GeV, it's easier to distinguish this (light) Higgs from the background than when the (light) Higgs mass is 115-130 GeV? For very light Higgses, you get the dominant messy decays to b+b_bar, and some to tau+tau-. For 150 GeV Higgses, the decays to two "slightly off-shell" W's are still very different from the other background, aren't they?


    When I localize what I really don't understand about the chart, it's the "expected" exclusion. What is the "expectation" based on? A priori, the MSSM allows you very - almost arbitrarily - high values of tan beta, doesn't it? Is the expectation based on the Standard Model without SUSY, and the goal is to rule out MSSM relatively to SM? It's the only possible explanation I can identify - maybe it's written somewhere in the article...

    If that's so, I still don't understand why the chart is interpreted as a hint for a 150 GeV heavier Higgs. Is it just because slightly bigger values of tan beta are allowed for these masses i.e. because there's a slightly bigger area before the upper thick black line in the chart than for e.g. 130 GeV? What are the higher values of tan beta good for? There can still be a valid MSSM with reasonable values of tan beta for 130 GeV Higgses, too - can't they? Moreover, I don't understand in what sense the deviation around 150 GeV reflects the actual observations rather than the theoretical differences between the models. If I believe that MSSM is more likely correct than not, and I do, then the "higher deviation of this produced chart from SM around 150 GeV" could be simply a theoretical feature of the MSSM, including the MSSM with different Higgs masses, couldn't it?

    If the goal were just to say that one can find a 2-sigma evidence for new physics somewhere over here, it seems correct. Well, it's still 95% confidence level only. And I do believe there's new physics, anyway. But I am simply missing the steps that lead the writer to believe that the heavier Higgs mass around 150 GeV is becoming more likely by this measurement.

    BTW there were some recent papers about the very high tan beta regimes, e.g.:
    http://arxiv.org/abs/1004.1993


    The loop-induced contributions from the "wrong" Higgs doublet become very important.

    Cheers
    LM


    dorigo
    I fail to understand whether you are referring to MSSM Higgses in the first paragraph. The fact you insist in stressing "light" seems to indicate that. But branching ratios are different. However, for the specifics of the question, I agree, at 150 GeV the Higgs is easier to spot than at 115. This is true both at the Tevatron and the LHC.

    As for the expectation band: it is based on pseudoexperiments, where you assume there is no MSSM Higgs, only SM backgrounds. You try to fit different Higgs masses, and for each you determine what limit on tan(beta) you can set given the allowed signal. You get a distribution of upper limits, from which you can extract 1- and 2-sigma bands. By comparing the expected limits with the observed ones, you get a feeling of how well the observed data match expectations. At 150 GeV the analysis cannot exclude large tan(beta) values because the data allow a sizable number of Higgs events there (larger tan(beta) implies more Higgs produced).

    Don't be deceived by the last "limit" graph - that is the final result, but the "signal" is not there, it is rather in the mass distribution. There, you can see that adding a MSSM Higgs of 150 GeV fits the data better than not having it in the mix.

    I am not sure I answered well your comment, because I am rather in a rush... Ask for more clarifications if you are still in doubt.

    Cheers,
    T.
    lumidek
    Tx, concerning 1st paragraph, I know nothing specific about the differences how hard it is to detect the SM Higgs and the light MSSM Higgs at these masses, so if I don't distinguish them, it's comparable with my knowledge....

    I kind of understand your expectation exercise. It's just not the usual exercise whose logic I am familiar with. It may still be useful for something - but I don't directly see what it is useful for and what kind of conclusions may be deduced from such convoluted what-if exercises.
    Hi Tommaso,

    i really have to wonder: L(H_1)/L(H_0)>1? H_0 has one dof less thatn H_1, so if they produce the same chi^2, then L(H_1)/L(H_0)<1. or are you doing this differently at CDF?

    dorigo
    Hi Chris,

    no, I was not describing the actual calculations of CDF, but rather the general idea. There are no two chisquared values -this is a log-likelihood fit- but a chisquared-distributed quantity C can be derived by taking the difference in twice the log-likelihoods of the two hypotheses. You then calculate the probability with Prob(C,NDOF), where NDOF are the number of degrees of freedom added to H_1 wrt H_0 (and the two hypotheses should be nested into one another, or the trick will not always work).

    Cheers,
    T.
    Oh, I have actually looked at this post one month ago, as the comments of mine prove.

    In the MSSM context, can this observed Higgs simply be the heavier neutral "H" Higgs? Because I assumed it had to be the lighter "h", and it was above 130 GeV, I ignored this posting immediately after I read about it.