Banner
    Tevatron Higgs Results Confirm LHC Signal!
    By Tommaso Dorigo | March 7th 2012 03:45 AM | 39 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
    2.2 standard deviations. That is what the combination of CDF and DZERO searches for the Higgs boson yield, according to a release of the interactions news wire.

    The talks on Higgs searches are scheduled for this morning at the Moriond Electroweak conference in La Thuile, a nice ski resort in the Italian alps.

    The money plot is the one below, which shows the upper limit on the Higgs boson production rate, in units of the Standard Model expectation (y=1 on the vertical axis corresponds to the SM Higgs production rate, for the particular mass identified by the abscissa in the graph).



    The Tevatron experiments can exclude Higgs masses above 147 GeV at 95 % confidence level, a region which has already been wiped out last year by the data resulting from more powerful collisions analyzed by the LHC experiments. However, what is most important in the graph is that the black curve describing the observed 95% exclusion "departs" from the dashed curve at the center of the "brazil flag" band, for Higgs masses in the whereabouts of 125 GeV. The upward departure signifies that the experiments predicted they would see fewer Higgs-like events in that region from background sources alone.

    The experiments quantify the "significance" of the discrepancy as a 2.2 standard deviations effect: way too little to discover a new particle, but yet enough to convince many of us that the signals observed by ATLAS and CMS in the corresponding searches are probably due to the real thing. As a reminder, the two LHC experiments found upward fluctuations in their searches which amount to discrepancies of 3.5 and 3 standard deviations from the background-only hypothesis, if the Higgs boson mass is assumed to have a mass in the whereabouts of 125 GeV).

    In the end it is just a matter of what your prior beliefs are. 3-sigma, 4-sigma, 5-sigma... Let's not get obsessed by these figures. Above 3 standard deviations a pure statistical fluctuation is already very, very unusual (save the infamous "look-elsewhere effect", of course); but systematic effects may make even "one-in-a-billion" fluctuations become commonplace. And since systematic uncertainties not accounted for by the experiments can and do occur, as the recent saga of superluminal neutrino measurements clarify, one needs to compare one's personal belief about the likelihood of the claim with the actually observed result. The 6-sigma evidence of superluminal motion of neutrinos found by Opera last fall was obviously not enough for scientists to get convinced of the breaking of the theory of relativity; a genuine 4-sigma Higgs signal is all it takes to convince me.

    To put the above another way: "extraordinary claims require extraordinary evidence". Well, the existence of a 125 GeV Higgs is not at all extraordinary to me: I would rather say it is a rather expected thing, given the mass of evidence on the correctness of the Standard Model and the indirect information on the possible range of masses of the Higgs that the Standard Model itself provides.

    I was surprisingly attacked by prof. Matt Strassler three months ago for saying that the LHC results on the Higgs were "firm evidence" for the existence of the Higgs boson. Now these 2.2 additional standard deviations -a one-in-fifty effect- do not allow me to change that objective, scientific evaluation: firm evidence was (technically, a above-three-sigma effect), and firm evidence remains. But the independent find by the Tevatron colleagues does raise the odds to which I am willing to bet that the Higgs boson is indeed there.

    UPDATE: Now I can offer much more information on the Tevatron searches. I decided I would simply add it here rather than in an independent posting. So let us examine the material in more detail.

    First of all, one should ask oneself whether the Tevatron experiments are capable of observing signals known to exist: the Z->bb decay in WZ and ZZ production events is a very good test of the power of the searches for the corresponding WH and ZH (with H->bb) production processes: WZ and ZZ have cross sections over ten times larger than WH and ZH, but the H->bb decay is more frequent than the Z->bb one by a factor of 5. So if the experiments can demonstrate that they see the Z->bb signal in WZ and ZZ events, one can then trust the Higgs results much more.

    The proof is given by the figure on the right, which shows a background-subtracted mass distribution for pairs of b-quark jet candidates (both single and double b-tagged events are used here) associated to a leptonically-decaying W or Z boson. The blue histogram shows the background uncertainty, the red and yellow histograms describe the expected WZ and ZZ signals, and the black points are experimental data (after background subtraction). One clearly observes an excess in the Z mass region -and some additional excess at higher mass.

    Next, let us look at a very interesting figure which shows the data presented as a function of the "signal purity". In other words, data from different searches is classified according to the expected value of signal to noise corresponding to their characteristics. The bins with higher relative signal content are the rightmost ones (technically, the abscissa has the logarithm of the signal-to-noise). Here, the hypothesis is made that the Higgs boson has a mass of 125 GeV. Expected backgrounds are collectively displayed in blue, and the expected SM signal contribution is in red. The data, shown by black points with stat-only error bars, nicely agree with the sum of the two processes, hinting at the fact that some Higgs events might indeed be present. Of course we are talking about a very small signal, and the data are not incompatible with backgrounds alone in this particular view; but the figure is still important to examine.

    A histogram closely connected to the one above is the one showing the number of events observed and expected for the two hypotheses (background-only, in black, and background-plus-signal, dashed red) as one sums bin contents in the histogram above starting from the right, and continuing leftward. In other words, the figure shows a cumulative distribution considering the bins ordered by signal to noise. Because of the ordering, the leftmost part of the integrated distribution allows one to see the signal contribution more clearly, while once data with larger relative background content is added as one move toward the right, the signal becomes less and less important in a relative sense. As you can see, the data points consistently follow the signal-plus-background hypothesis (actually overshooting it). This imples that there is a signal-like component in the data. However, do not be deceived by the error bars of the data points: they are 100% correlated, because of the way the histogram is constructed. That means that to visually size up the discrepancy of data with background alone one can only count the distance of the data points from the black curve once, in all the graph. That means a roughly 2-sigma discrepancy, similarly to what I announced at the beginning of the article.

    Now let us go to the background p-value. This is a number which quantifies how likely it is that the data comes from background sources alone. Since we are making multiple hypotheses for the Higgs boson mass in the analysis, and each mass point is independently studied and optimized, one ends up with a different p-value for each different Higgs mass hypothesis. Do not get confused: we are still discussing the background-only hypothesis, but the data changes along the abscissa, which should thus rather be labeled "Higgs mass hypothesized in the corresponding data selection and analysis", if one wanted to be clear. Anyway, the black curve shows the observed data p-value; the dashed curve shows what p-value one would expect to observe (technically, this is the median value of a possible distirbution of p-values) if the Higgs boson did exist with the mass corresponding to the abscissa value. We also get to see a band of expectations, indicating that for 125 GeV the observed p-value is perfectly in line with what one would expect if the Higgs were indeed there with that mass.

    A different view of the same information is provided in another "brazil band" graph which I actually prefer to the one above: this is a plot of the logarithm of the likelihood-ratio discriminant between the two different hypotheses of background-only and background-plus-signal. By taking into account the two hypotheses together, one can plot also the background-only expected p-value here (in black, dashed, with brazil-band surrounding it), while the signal-plus-background hypothesis is this time in red. This way, one can verify how well the data matches with background-only predictions for Higgs mass values away from the 125 GeV point that we are all convinced by now is close to the true value. See below.


    From the figure it is clear that there is a departure from the background-only expectation only for Higgs masses in the 115-135 GeV region. The agreement of full and dashed black curves in the rest of the spectrum is instead reassuring that background expectations are in line with observed event counts.

    At this point one might well wonder which, among the various decay modes that have been investigated by the CDF and DZERO collaborations before putting together the information in the above graphs, is the one contributing the most to the observed signal-like excess. Of course such a question can only be answered meaningfully by providing a plethora of different results in the independent channels. However, there is a shortcut, provided by the figure on the left. This shows the p-value of the background-only hypothesis as a function of the Higgs mass hypothesized by the different data selections and analysis. The figure only receives as input the CDF searches for H->bb decay modes: that is all WH and ZH production processes yielding the H->bb decay together with a leptonic W or Z boson signature. The latter may result in charged leptons and missing energy, charged lepton pairs, or even just missing energy, when the Z decays to neutrino pairs. Anyway, the figure shows that the H->bb signature of CDF is alone producing an almost 3-standard-deviation effect at 125 GeV. This is the single most significant observation among the various Tevatron channels.

    Having seen the above figure, one might wonder whether the combined Tevatron result is not just due to a bad underestimate of backgrounds producing b-tagged jets in W/Z +jets samples. This would be a rather nasty attitude, however: I recall the figure I pasted above, where the WZ and ZZ signals demonstrate that CDF and DZERO appear capable of precisely estimate backgrounds in the search of such signals. However, it is true that the excess is mostly coming from the bb final state. The global picture from CDF can however be appreciated by another graph, which changes view from the "null hypothesis" perspective to the one of fitting the possible Higgs signal in the data. By taking all the search channels together, CDF produces a global fit for the signal strenght assuming that the branching fractions in the individual channels are the ones predicted by the Standard Model. The result is the one below.


    You can see that for a Higgs mass hypothesis of 125 GeV the CDF data is in line with expectations (which correspond to the value of 1 of the vertical scale, i.e. the Standard Model prediction). This is another nice way to look at the CDF results. I unfortunately do not yet have available a corresponding graph for the DZERO searches, but will try to paste it here later today.

    More information is coming as I collect it from today's La Thuile talks... Stay tuned!

    Comments

    From that graph it looks like they have an even better signal at 200 GeV.

    Hi Tommaso,

    do you think that Tevatron will be able to pip the LHC to the post and provide proof of Higg's existance before the year is out? They have after all already produced their data and - thanks to LHC - know where to look.

    Cheers,
    Martin

    dorigo
    Hi Martin,

    the results above really are the swan song for Tevatron Higgs searches. Any adjustment will only modify negligibly the picture. The Tevatron does not have the sensitivity to reach for a Higgs signal, especially in the 120-130 GeV region, as is evident by the figures I have pasted to the post now.

    Best,
    Tommaso
    Do they actually have Exbibyte of raw data lying around that they are currently pumping through a Hadoop Cluster? The fact that updates come out 6 month after shutdown is amazing.

    vongehr
    So Matt "attacking" you was bad, but you doing similar to me here was fine (because of your private definition of "firm evidence")?
    The 6-sigma evidence of superluminal motion of neutrinos found by Opera last fall was obviously not enough for scientists to get convinced of the breaking of the theory of relativity;
    A 100 sigma evidence of FTL particles would not be enough to doubt the theory of relativity either (at least for those who understand what that theory is actually about).
    Sascha,

    Generally, the tone on this website is quite respectful. Only Lubos Motl sometimes drops by with his occasionally quite aggressive stance. But you have neither Lubos’ panache nor his scientific standing, i.e. you offer little in form of entertainment or scientific insight. Instead you seem to be spoiling for a (verbal) brawl.
    If you are generally interested in the topics discussed on this website, ask relevant questions and keep your manner courteous. Don’t act like somebody who thinks the world owes him and is slow in paying up.

    Cheers,
    Martin
    PS Please note that I used the reply button.

    vongehr
    Martin, you may like to consider the possibility of lacking the deeper insight necessary to grasp the hidden arrogance that Dorigo masterfully throws at people he thinks himself superior to, let alone the deeper level insight that I aim for. If you do not see the relevance of comments that point out double standards and almost pseudo-scientific argumentation by cheerleaders for naive scientism (for example against FTL particles without understanding modern relativity theory), than you plainly miss the whole point of science blogging! Hint: It has something to do with credibility and legitimacy of science in the democratic decision process of a technological world (it is not about who is one day earlier than me getting out a rumor about a two sigma bump).
    If you have difficulties grasping my articles, you are welcome to ask pertinent questions in the comment sections. Thank you.
    Hi Sascha,

    It is true; I have not noticed Tommaso's arrogance. He must indeed be masterful in hiding it. Practice makes perfect!
    Sorry about making insinuations about your scientific knowledge. I should have checked your profile first.
    And according to Tommaso's profile you are on his friend list. I see now that I was worrying needlessly.

    Cheers,
    Martin

    Martin, please tell me this last comment of yours is purely ironical.
    (And that's enough for pseudoscience today, sorry for adding to the noise in the thread...)

    Hi Tulpoeid,

    I'm sorry; I'm feeling openly arrogant today and will therefore - very condescendingly - not answer your request.

    Cheers,
    Martin
    PS Is Tulpoeid your nickname? What does it mean?

    "you may like to consider the possibility of lacking the deeper insight necessary to grasp the hidden arrogance ..."
    Holly cow! You might like to consider the possibility of stepping out get some fresh air!

    dorigo
    I have not said it was bad. I only note I was right in both cases. The evidence of LHC was (and is) firm, the superluminal neutrinos are a chimera.

    Cheers,
    T.
    vongehr
    You were wrong both times and did you know that you can use the "Reply to This" button? Oh you used it with Martin but not with me, hmmm, I wonder why. Arrogance - do you really need it?
    dorigo
    You are being a child as always. Your comments are way off topic. I will delete any other, so please save me the trouble and abstain from continuing.
    T.
    Tony Fleming
    Any chance you might give us some idea of how much and what work is involved in the (money) graph as shown please Tommaso
    Tony Fleming Biophotonics Research Institute tfleming@unifiedphysics.com
    dorigo
    Hi Tony,

    the graph is a summary of a large number of different analyses from the two experiments CDF and DZERO. Each experiment counts roughly 500 collaborators, and at least a fourth of them have worked in a way or another at the analyses for a sizable fraction of their time in the course of the last ten years (or even more in some cases). I would estimate that just from the point of view of data analysis techniques (including work at perfecting the identification and measurement of the physics objects that make up the Higgs decay signatures), each experiment probably devoted of the order of a million man-hours of work to reach the goal of producing that figure (and a host of others, of course).

    Maybe looking back at the article (updated a few minutes ago) you may see some of the graphs that were produced together with the one you point at, and get a better representation of the ingredients.

    Best,
    T.
    Tony Fleming
    Jeepers. Can't wait to see the title page and authors of THAT paper when it emerges!! Thanks Tommaso cheers Tony
    Tony Fleming Biophotonics Research Institute tfleming@unifiedphysics.com
    Not a physicist here, sorry if this is quite ignorant.

    I had been under the impression that Tevatron couldn't confirm a higgs signal above 115Gev or so, hence the necessity of the LHC in the higgs search. Obviously that's wrong, but could you clarify why this data is coming from Tevatron now, rather than much earlier? Did they have one last run at higher power, and it just took this long to present the data?

    Is it possible to combine the data sets from D0 and CDF with the Atlas/CMS data? Can we now consider the atlas/CMS to be, say, 4 sigma or some such number higher than the 3.5 previously reported?

    If you ignore 95 per cent exclusion zones as being too weak
    (and require at least 99 per cent exclusion)
    then
    the Tevatron (Wade Fisher slide 24)
    seems to show 2 peaks at 120 and 135 GeV
    and
    ATLAS (Sandra Kortner slides 11 and 14 and 24)
    seems to show excesses at 126.5 and 244 GeV
    and
    CMS (Marco Pieri slide 11)
    seems to show excesses at 125 and 136 GeV.

    If you just take the experimental results at face value
    and do not view them from a particular model (SM or other) viewpoint,
    and
    you say that
    Tevatron resolution does not allow you to distinguish
    its 120 from 125 or 126.5 and its 135 from 136
    but
    ATLAS and CMS resolution does distinguish between 125 and 126.5
    then
    would it be fair to conclude that
    none of the results are evidence or discovery of anything,
    but
    they are indications of interesting phenomena at
    125 GeV
    126.5 GeV
    136 GeV
    244 GeV
    that should be studied further.

    My speculative model explanation would be
    125 GeV - pion-type meson T0 (Tquark-antiUpquark)
    126.5 GeV - pion-type meson T0c (Tquark-antiCharmquark)
    136 GeV - SM Higgs (one state of multi-state)
    244 GeV - SM Higgs state near Higgs VEV

    As it is relevant to our wagers,
    with your position being a SM Higgs around 125 - 126.5 GeV
    (and the 136 and 244 eventually going away with more data)
    how do you account for the difference between
    the ATLAS 126.5 GeV and the CMS 125 GeV ?

    Tony

    PS - Thanks very much for the clear explanation of the histogram for Integrated Expected Signal. I had (as my comment on Resonaances showed) not understood it properly.

    What does a Higgs at 125 GeV tell us?

    He tells us: "I'm finally here! Good morn..." then zip it decays!

    Does the 2.2 sigma refer to (CMS jargon) local significance (the "standard" one usually cited) or to global significance (the one that takes into account the look-elsewhere effect)?

    Hi Tommaso,

    if one of my comments doesn't make it past the "Awaiting moderation", do I get a short email with a reason or two?

    Cheers,
    Martin

    dorigo
    HI Martin,
    no, I don't think so. I do not moderate comments; it is an automated system that takes into account the content and the origin of the comment. I can only apologize if something gets delayed - I can approve manually things held up for moderation but I don't always manage that well and promptly.
    Cheers,
    T.
    Hi Tomasso,

    its OK. The comment I posted yesterday appeared today - at least on my PC.

    Cheers,
    Martin

    It does include the LEE.

    gunn
    "_ It is not Higgs because  every physics event is interpretted by particles which similar well-known elementary particles - leptons, quarks and gauge bozons. Therefore, if anybody will claim that he had found Higgs then not believe - this is not Higgs. " _
    http://www.ptep-online.com/index_files/books_files/quznetsov2011.pdf
    ttp://arxiv.org/pdf/physics/0302013v3
    "But the independent find by the Tevatron colleagues does raise the odds to which I am willing to bet that the Higgs boson is indeed there."

    Do you not mean LOWER the odds you are willing to bet?

    dorigo
    Yes, sorry :)
    Cheers,
    T.
    Hi
    Without being a specialist, I try to understand
    1) all these articles focus on the mass of the Higgs boson according to its prediction but why a particule has a given mass is just what the Higgs boson is expected to explain; I suppose thus that you have eliminated mass ranges from some other predictions; is it possible to find somewhere an explanation behind those elimination?
    2) how that Higgs boson will elucidate the particule masses, including itself?
    3) in those experimenys, I suppose that the nature of that particules has been elucidated by other Higgs 's properties?
    which ones and why it should have these properties?

    Chris Austin
    Would the Tevatron be able to get a better fix on the Higgs mass if it considered only Z H events in which the Z decays to hadrons or charged leptons, and the b and \bar{b} from H -> b \bar{b} both decay to hadrons, e.g. b -> c \bar{u} d, (so no neutrinos), or is the rate for these too low?
    Chris Austin
    Near the end of the CDF results page for WH to l nu b bbar, which of the vector + Higgs channels is the one where the Tevatron expected to get closest to the SM cross-section x branching ratio, they state that for the WZ control channel, which was used to test the new HOBIT b-tagger, the fit for the total WZ cross section distributions yields 5.63 +1.79 -1.76 pb, in comparison to the SM prediction of 3.2 \pm 0.2 pb. Could this account for the 2.2 sigma excess?
    dorigo
    Hi Chris,

    indeed, part of the excess cross section  for WZ (with Z->bb) can be due to some Higgs events leaking out there. But it does not "account" for it... The analyses are different, the multivariate discriminants are optimized differently, etcetera.
    Cheers,
    T.
    lumidek
    Dear Tommaso, let me just say that I agree with all your evaluations of the Higgs data, at the LHC and the Tevatron, and the neutrino speed data from OPERA, too.
    Bonny Bonobo alias Brat
    Its nice to see you guys agreeing!
    My article about researchers identifying a potential blue green algae cause & L-Serine treatment for Lou Gehrig's ALS, MND, Parkinsons & Alzheimers is at http://www.science20.com/forums/medicine
    lumidek
    Almost always happy to make you happy, Helen.
    Bonny Bonobo alias Brat
    Almost always is a lot better than anyone else I know, thanks Lubos :)
    My article about researchers identifying a potential blue green algae cause & L-Serine treatment for Lou Gehrig's ALS, MND, Parkinsons & Alzheimers is at http://www.science20.com/forums/medicine
    I humbly request the exact conversion formulas, including the numerical values used to the last decimel for gigavolts, megavolts, ( "n" electronvolts ) etc; into mass in grams ( fractions thereof ) for comparison to the stated mass values of the really well known and measured particles such as Mn, Mp, Me, u. Please understand I am trying to figure out everything I can about this. I have no clue why I can't find the proper website that explains all this mathematically. The numbers and as many examples as you please can't possibly over explain to me. I'm interested in the "W" intermediate boson known to exist in the neutron. Why, how, when, by whom using what logic came it to be found between two protons for example. One day I walked past a shop in Paris displaying beautiful dresses, purses, accessories, and said to my daughter "Look at the size of the shoes! " " Dad the customers of this shop don't ovulate, gestate, lactate, but to fools they look better than the real things."

    dorigo
    Dear anonymous professor,

    I doubt there's much to figure out from a collection of numbers, especially if these are expressed in non fundamental units. In any case, the conversion from a gram to a GeV/c^2 can be found in the internet quite simply (look e.g. in pdg.lbl.gov), under "constants".

    The W does not "exist" in the neutron. A virtual W may be emitted by a down quark in the neutron when this turns into a up quark, thus making the neutron become a proton. In the process, the virtual W lives a very, very small time, a time so small that the W can fool the conservation of energy (not a correct explanation but one which can be understood), then turning into an electron-neutrino pair. This is called "beta decay" and was first formulated by Fermi as a four-body reaction (which at the time did not include quarks, nor the W, but was thought as a point interaction connecting a neutron, a proton, an electron, and a neutrino). Then in the fifties the intermediate vector boson theory was born: the weakness of the decay - its slowness - was ascribed to the fact that it were mediated by a very massive boson, a W. The W was discovered in 1983. Nowadays if you collide protons, or any other particle made of quarks (hadrons, that is) at sufficiently high energy, you can make real W bosons in large amounts. The LHC experiments have collected millions of them already in less than two years of running.

    Cheers,
    T.