Higgs at Fermilab2007

Z -> bb and A_FB

A_FB and Higgs

Tommaso Dorigo, in hisblog entry "New Higgs predictions from EW fits:M(h)=80 GeV!" said, on 19 January 2007:

"...  An improved version of the Mw-Mt plot has been producedfor winter conferences. I am still unaware of the numbers entering inthis new global fit, but here is the January 2007 result anyhow: themost likely higgs mass sits at 80 GeV, with an uncertainty of +36-26GeV. This supersedes the former result of 85+39-28 GeV.  

The plot shows the dependence between the W mass and thetop quark mass in the Standard Model, as a function of the unknownmass of the Higgs. Lines belonging to the green band represent thefunctional dependence between W and top mass for a fixed value ofHiggs mass. The red ellipse constrains the three particle massesusing electroweak measurements at LEP and SLD, while the blueellipse encrypts all direct measurements of the top and Wmasses. The fact that the intersection between the blue ellipse andthe red one is very small, and does not include any green region,implies that there is a tension between the lower limit on the Higgsmass obtained at LEP II (114 GeV) and the favored higgsmass, now at 80 GeV and thus about one sigma away from the allowedregion.  

This is becoming a story of the incredibly shrinking higgs. Butwait until my team blesses an improved jet energy scale measurementin CDF, using the Z->bb peak: I expect it to change things a bit,since most of CDF top quark mass determinations will be affected. Ifthe authors of the analyses include the new scalemeasurement into their results, that is. ...".


I [Tony Smith] commented on 23 January 2007:

"... Tommaso said

"… here is the January 2007 result anyhow: the most likely higgs mass sits at 80 GeV, with an uncertainty of +36-26 GeV…there is a tension between the lower limit on the Higgs mass obtained at LEP II (114 GeV) and the favored higgs mass, now at 80 GeV and thus about one sigma away from the allowed region. …".

Although one sigma is far from enough to say that the ElectroweakStandard Model is refuted by LEP II HIggs search, it does raise thequestion:

If the Higgs is really at least 114 GeV, then what parts of theElectroweak Standard Model are most likely to be changed in makingthe global fit to data match the experimental Higgs mass.

 According to the 2006 PDG review "10. Electroweak Model…" revised September 2005 by Erler and Langacker:

"… The global fit to all data, including the CDF/D0 average, m_t = 172.7 +/- 3.0 GeV, yields M_H = 89 +38 -28 GeV …

the goodness of the fit to all data is very good … Only g_L^2 from NuTeV and A^(0,b)_FB from LEP are currently showing large (2.7 sigma and 2.4 sigma) deviations. …

The strongest individual pulls toward smaller M_H are from M_W and A^0_LR …

A_LR was measured precisely by the SLD collaboration …

while A^(0,b)_FB and the NuTeV results favor high values …

the Z-pole forward-backward asymmetries at LEP … are given by A^(0,f)_FB … where f = e, mu, tau, b, c, s …

LEP 2 … Measurements were made … including … A_FB for b and c …

it appears that at least some of the problem in A^(0,b)_FB is experimental …

the uncertainty in A^(0,b)_FB is strongly statistics dominated. …

for … A^(b,c)_FB + M_Z …[gives]… M_H … 349 +250 -148 …[GeV]…

for … All data …[gives]… M_H … 89 +38 -28 …[GeV]…".

If the value of A^(0,b)_FB (pull toward high Higgs mass) were tobe given more weight


the value of A^0_LR (pull toward low Higgs mass) were to be givenless weight


would that give a Higgs mass more nearly in line with LEP IIsearch results?

Since I am very naive about all these things, I will ask

whether the fact that LEP measured A^(0,b)_FB whose value is apull that "favor[s] high values" of M_H

is related to the line in table 10.7 of the 2006 PDG Review:

"for … LEP 1 (no m_t) …[gives]… M_H … 168 +232 -91 …[GeV]…" ?

whether the fact that SLD measured A_LR whose value is one of the"strongest individual pulls toward smaller M_H"

is related to the line in table 10.7 of the 2006 PDG Review:

"for … SLD + M_Z …[gives]… M_H … 28 +26 -16 …[GeV]…" ?

Also, I note that the 2006 PDG Review says:

"… The forward-backward asymmetry, A_FB, for e+ e- finalstates in ppbar collisions has been measured by CDF and a value ofsbar_l^2 extracted … A^0,b)_FB … and the hadronicasymmetries are mainly sensitive to sbar_l^2 …",


another very naive question I have is

whether CDF results support the LEP result (pull toward highHiggs mass) ? ...".


Tommaso Dorigo replied 24 January 2007:

"... I am in no position to discuss the subtleties of the A_fbmeasurements, of which I am sure there are readers of this post whoare more knowledgeable than me. However, I can certainly say you areright about those two pulls in opposite directions by LEP and SLD aremeasured in strength by the fact that if used alone in the fits theyproduce quite different values of the higgs mass. The relativestrength of these pulls is, very simply, measurable by the distancebetween the higgs mass they predict with that of global fits (eg. 28vs 80, 168 vs 80) and their uncertainty. The SLD measurement isindeed suspicious if exposed bare as the PDG does in the line youmention.

As for Nutev, their measurement is indeed quite off the mark, butI think that taking it away from the soup "because it is unlikely"subtracts information illegitimately. If included, the higgs massmagically grows by quite a bit… but the inclusions pays theprice of a bad agreement with other measurements, so the "global"chisquare worsens significantly. I do not know by how much,unfortunately, since the blue band plot hides that information as faras I recall.

 I will add that CDF measures the W mass high, but the pullin the global fit is not large. It is much larger the pull due tosmall top masses, and here let me be a bit suspicious, because myinsider information happens to indicate that our jets are measuredslightly low - by maybe a couple of GeV in the top decay range - andonce that is taken into account, the top mass measurements increaseby a little bit, enough to make the higgs rise also a bit. I cannotbe more precise, since the b-jet energy scale measurement is notblessed yet, but it will be so in a few weeks, so stay tuned!

As for ppbar measurements of A_fb, I need to look those up!...".



Z -> bb andHiggs

 Tommaso Dorigo, in hisblog entry "A 2.1 sigma eccess of MSSM Higgs!" said, on 19January 2007:

"... let me dig out a plot from D0, showing their excess ofZ->bb events. Remember, if the higgs is produced directly, it hasa branching fraction to b-quark pairs, and should be present as anexcess in a plot such as D0's, albeit in modicum quantities.Here you can see the recent D0 plot.


Interesting! They have an excess at 160 GeV above QCD backgrounds(which are subtracted in this plot). I wonder if D0 is consideringthe use of that plot - or better, an updated, larger-statisticsversion of it - for setting limits on ... higgs bosons, or claiming asignal!

My question is not irrelevant, although slightly half-joking as ismy usual style. Indeed, Julien and I are just about to bless our ownZ->bb signal, based on half an inverse femtobarn of data. Ofcourse I cannot show you the plot, but guess what ? We also see abump at 160 GeV!

I got you curious, didn't I ? You will have to wait for three moreweeks to see the Z->bb plot by CDF, unfortunately! ...".

I [Tony Smith] was indeed curious, and commented on 24January 2007:

"... Lubos Motl in his blog on 23 January 2007 says about whatlooks like the same D0 plot shown in this blog entry:

"… I [Lubos] am unconvinced by the statement that there is an extra bump at 160 GeV in the chart above. The chart would be more likely to indicate a bump at 140 GeV. …".

Although the statistics are very inconclusive to say the least, itseems to me that Lubos's remark is correct. I am not disinterested,because my model (non-supersymmetric) has a Higgs at about 146 GeV....".


Tommaso Dorigo replied 24 January 2007:

"... if you look at D0's plot, the Z peaks at 80, not 91 GeV.

That means that b-jets were not corrected for the many effectsthat make them different from generic jets (on which jet energycorrections are tuned, since those are the jets you get when youtrigger on them at a hadron collider).

b-jets are indeed different: they have a harder fragmentationfunction, have a larger mass, but most of all they have a 23%branching ratio into leptons - so neutrinos are produced, and they donot leave a signal. Also, 13% of b-jets have muons inside, fromb->mu or b->c->mu or b->tau->mu decay chains. Andmuons, too, give a minimum-ionizing response in the calorimeter.

All that to say that if the Z peaks at 80 instead than 91, then abump at 140 means a particle at 160 GeV. ...".


Also on 24 January 2007 I commented:

"... thanks for explaining about the Z-peak being at 80 GeVindicating that the analysis that produced the chart was uncorrectedfor b-jet missing energy, so that the Z-peak showing up at 80 GeVwhen the Z-mass is known to be 91 GeV means that the Z-peak should becorrected by shifting it up 11 GeV to 91 GeV.

As you say "… then a bump at 140 means a particle at 160 GeV…".

Since my model value of 146 GeV Higgs is a tree-level calculation,and

since the state of uncertainty is such that it is OK to say that140 + 11 = 160 GeV (and the 140 only comes from eyeballing apreliminary chart),

I am with respect to my model happy to see a bump around 160 GeV,and am looking forward to the CDF Z to bb plot to be released nextmonth.

Further, the shift up by 11 GeV also would shift a much smaller(just one data point a bit above the base line, but it and itsneighbors do look like a peak) bump in the D0 chart around 170 GeV upto around 181 GeV, which is in the range where my model expects tosee a second Higgs peak (the model having 3 peaks for both the Higgsand the Tquark).

Such things are very highly uncertain, but they give me (probablyunwarranted, but it feels good anyway) optimism while waiting for LHCresults. ...".


Still later on 24 January 2007 Tommaso commented:

"... I think you know me well enough to not be surprised if I amskeptical about three-bumped stuff. However, I keep it an open-endedissue until the LHC nails it. And I would be disrespectfully happy ifyou turned out to be even half right.

As for the statistical information in plots, however, I usuallyget even more skeptical, since I know how to run pseudoexperiments,and when you do it long enough you get to understand a simple fact:just about any background-subtracted plot will show structures suchas those at which you hint above.

You of course realize that yourself when you say "such things arevery highly uncertain"… Yes. The CDF Z->bb plot will havefour times more statistics than the D0 one, and some structure above100 GeV will still be present (I cannot tell you exactly what orwhere for fear of being kicked out by CDF), but they will fail todemonstrate anything… We really need LHC data here. ...".

I agree that LHC data is needed to be very confident about suchstructures, but my personal confidence,

while very low for a bump at 160 GeV only in the D0 Z -> bbplot (just one plot),

goes up a lot if I see an independent plot of the same thing, alsoshowing a similar structure ( bump at 160 GeV ). ( Somethinglike that happened in the 1990s when a D0 histogram showed 3 Tquarkmass peaks similar to the 3 Tquark mass peaks in a CDF Tquarkevidence histogram. )

Since Tommaso said (quoted above):

"... You will have to wait for three more weeks [ from 19January 2007 to 9 February 2007 ] to see the Z->bb plot by CDF... Of course I cannot [ before 9 February 2007 ] show youthe plot, but guess what ?

We also see a bump at 160 GeV! ...",

I am, assuming that events due to 80+80 = 160 GeV Wpair have beentaken into account in background calculations by D0 and CDF,optimistic (subject to confirmation or refutation by LHC) that:


It is interesting that in the energy region from weak boson masses(80 to 91 GeV) to Higgs VeV (252 GeV) there are three bands (particlemasses are tree-level values calculated from my model, and so maydiffer a bit from experimental observations):

Low: 130 to 160 GeV - Tquark_low (130 GeV), Higgs_low (146GeV), W+W- pair (160 GeV)

Central: 170 to 190 GeV - Tquark_central (175 GeV), Higgs_central(180 GeV), Z0 pair (180 GeV)

High: 220 to 260 GeV - Tquark_high (225 GeV), Higgs_high (240GeV), W+Z0W- triple (252 GeV), Higgs VEV (252 GeV).



Apreamble to the CDF results on Z to bb decays February 13, 2007,posted by Tommaso Dorigo:

"... I have been searching for Z decays to b-quark pairs inproton-antiproton collider data for the last 11 years. Of course thatwas not the only thing I did research-wise in this long time span,but it has been some sort of trademark of my research activity.During my whole career I almost exclusively dealt with thereconstruction of particle decays in jet final states. I saw topquarks, W bosons, and Z bosons decays to jets -and in two cases,these were first-time observations. Let me describe what this isabout in a few lines.

The heaviest known particles - W and Z bosons, plus ofcourse the top quark, or the still-to-be-seen Higgs boson -can decay to quarks pairs (triplets, the top) carrying so much energythat the latter fragment into collimated streams of hadrons -particles themselves composed of quarks. Hadronic jets arecomplicated objects. Their efficient reconstruction and precisemeasurement in present-day experimental apparata is not trivial. Andthe measurement of the mass of the originating particle, whichdisintegrated to yield the quarks and eventually the jets we observe,is very challenging.

I begun my career as a particle physicist in 1992, whenI started collaborating with a group of physicists in Padova thatwas looking for the decay of top quark pairs into six hadronicjets - a final state apparently doomed by a huge background of QCDprocesses. Despite skepticism from our collaborators, my groupeventually found a signal of that decay (you can see the top quarkdecay contribution as a white area in the plot, best fitting theblack points if added on top of the yellow distributiondescribing the background shape),

and in 1997 we published a "first observation" paper describingour result. 

From 1996 to 1998, as a PhD student, I decided I would searchfor the decay of the Z boson to pairs of b-quark jets. As hard asfinding six-jet decays of top-antitop pairs had been, this iseven more difficult: you only have two jets in the event, andyour background is more than 1000 times larger than the expectedsignal, and almost impossible to tell apart. In fact, at the outsetmost people in my collaboration believed I was wasting my time -a fewexperts had tried and failed before I even started. But I wasconfident in myself, and I also knew that finding Z->bb decays wasof paramount importance: even failure would not be a waste of time!In fact, most of our hopes to find the Higgs boson at the Tevatronrely in the identification of its decay to pairs of b-quark jets, thevery same final state of Z decay. And the Z is only slightly lighterthan the Higgs is expected to be. But the Higgs has a much lowerproduction rate! How, then, can we hope to see a Higgs if we do notfirst find the Z in the same final state ? Moreover, a well-defined Zsignal allows one to tune the response of our detector to b-jets. Wecan test whether our jet energy is measured correctly, because weKNOW what the Z mass is (the LEP experiments measured it withextreme accuracy), and what we reconstruct in the detector canbe brought to match that knowledge, improving our measurements withb-jets - the top quark mass, or the Higgs mass if the latter isfound. Finally, the Z signal becomes a testing ground for anyalgorithm that attempts to increase the resolution of the b-jetenergy - again, an issue of paramount importance in the search forthe Higgs: if you are looking for a tiny little bump in a massdistribution, the better your resolution the higher your bump willstand on top of a flat background - so your discovery reach on theHiggs is a linear function of your energy resolution! In 1998 Iblessed the result shown ...

I found a small signal, about 90 events (you can see the excess ofred points over the blue histogram in the inset, shown as a functionof the reconstructed Z boson mass, and a background-subtracteddistribution in the main plot): not altogether so significant byitself, but an important assertion about the capability of hadroniccolliders to use that signal to learn how to measure b-jets andsearch for a light Higgs boson.

 Then, in 2005 Julien and I, together with other members of asmall, dedicated group were able to bless the "public-relation"plot shown here,

using four times more statistics and a much more efficienttriggering procedure from Run II data. The green Z decay signal shownhere amounts to about 3000 Z decays, but we did not use it in thedetermination of the b-jet energy scale, nor did we obtain a preciseestimate of the production cross section.

Last year it was D0's turn to show what they could do in thisdepartment. They found a signal of about 1200 events ... see plot...

but they did not venture in a full measurement of the energy scaleor the Z cross section either.

 Ok, that was the past. Now, in little less than twoweeks, my work of ten years will reach a significant milestone.The search for Z bosons in CDF Run II data will be made public by CDFvia what is called a formal "blessing" - an internal meeting wherethe results of a full-blown analysis are carefully scrutinizedand approved for public consumption. After that happens, you will seehere the new plots and the results we obtain. ...".

Howto extract a signal of Z decays to b-jets February 19, 2007,posted by Tommaso Dorigo:

"... In a recent post I promised I would discuss the analysis bywhich the CDF collaboration (that is, Julien and me, in this case!)extracts a signal of Z decays to b-quark-jet pairs… So here itgoes.

The process of Z production, followed by the decay of the weak Zboson to b-quark pairs, occurs roughly once every fifty million timesat the Tevatron. This fact would not be disheartening by itself -signals a thousand times rarer can and have been isolated - were itnot for the additional complication that the signature producedby two b-quarks emitted by the Z is really not verydistinctive: a pair of back-to-back jets which contain hints of thedecay of B-hadrons. The lack of tale-telling features capable ofdistinguishing the Z event from the 49,999,999 other collisions isthe real trouble in extracting a Z&endash;>bb decay!

The CDF online data selection is a set of hardware devices thatread in the detector output and decide in real time whether aproton-antiproton collision is worth writing to disk or not. Whilethe collision rate is 3 millions per second, the storage system canonly write to disk 100 events per second or so: a very selectivefilter, the trigger, takes care of picking the most interesting onesand discarding the rest. The search for Z->bb decays starts rightthere, at the trigger selection: to ensure that they get collected,the trigger must somehow recognize them.

An extraordinary device called the SVT takes care of preciselyreconstructing the trajectories of charged tracks contained in jetsin microseconds, allowing the trigger to select jet production eventswhere b-quarks are likely to be present.

Due to the need of reducing the huge rate of QCD processesyielding b-quark jets in the final state as well as light-quark orgluon jets mistaken by the SVT as coming from b-quarks, the triggerselection kills 95% of our beloved events. We have to live with this,and be happy of the 5% we have a chance to study. (Note: QCD (quantumchromo-dynamics) processes are those mediated by the stronginteraction. Most frequently, they consist in a high-energyscattering of two quarks or gluons off one another. The scatteredbodies hadronize in a pair of back-to-back jets not unlike those fromZ decay.)

The surviving events are reconstructed by an offline program, andthen a further filtering starts: we need to get rid of as muchbackground as we can, by relying on any conceivable characteristicsable to discriminating Z production from QCD events.

The selection starts off by requiring both jets to contain areconstructed secondary vertex. A secondary vertex is the originatingpoint of two or more charged tracks contained in the jet cones, andit is the result of the long lifetime of the b-quark. The hadroncontaining the b-quark leaves the primary interaction point at littleless than light speed, traveling as much as a few millimeters beforedisintegrating into the charged particles which allow to determinethe decay point.

After the selection of an almost pure b-quark pair productionfinal state, the signal-to-noise ratio is still very small: in thedata analyzed by CDF, about 10,000 Z->bb events are contained indataset of about a million events. But QCD events, producedby strong interactions, are different from Zproduction, due to weak interactions. The difference isthat the Z can only be produced by the collision of two quarks -one from the proton and one from the antiproton-, while QCDevents are mostly the result of a gluon-gluon collision -again,each coming from one of the two colliding bodies.

Gluons, the mediators of the strong force, have a higher colorcharge than quarks. They are thus more likely to radiate additionalgluons. This is a difference that results in a higher fraction ofthree-jet events in QCD processes than in Z production processes. Forthe same reason, the two leading jets emitted from the Z decay aremore likely to be back-to-back than jet pairs from QCD events.

A selection tuned to increase as much as possible the signalsignificance is thus performed on two kinematic variables: the anglebetween the two jets, which is required to be closeto 3.14, and the transverse energy carried by the third jetin the event - if there is any. This allows the signal to noiseratio to become roughly 2% - a Z event every 50 QCD events.

Once you have done all you can to increase the signalsignificance, you have one last weapon - the most powerful one,which you saved for your advertising plot. It is the invariantmass of the dijet system: when the Z boson decays to twob-jets, it imparts them with energy equal to its restmass. By measuring the energy of the two b-jets you can getback the 91 GeV of the Z mass, give or take some uncertainty due tothe imprecise determination of jet energy. QCD events, on theother hand, are not the result of a massive particle decay:the reconstructed mass of b-jets from QCD processes is a smoothlyfalling distribution, whose shape reflects the selection cuts(requiring jets to have energy in excess of 20 GeV, say) and thedecreasing probability of higher energy collsions.



The plot shown above is the final result of the analysis performedon early Run II data in 2005. The Z  contribution extracted by atwo-component fit is shown in green. You can see that it is a smallfraction of the total data! Yet, its measurement carries somesignificance, because by comparing the result with simulations welearn how well we measure the jet energy in our detector.

In four days, Julien will bless the result of our analysis of newdata - almost twice as much as the data shown in the plot above. Fromthe Z signal we have successfully extracted a measurement of ourb-jet energy scale, which carries an uncertainty as small as the bestdetermination obtained so far with the W->jj signal in top pairproduction events.

So what is the big deal ? Well, this determination of the b-jetenergy scale is the first ever obtained with b-quarks. The W signalallows to set the energy scale of light quark jets, which arepotentially quite different. With this new measurement, the futuretop quark mass measurements at the Tevatron will sizably decreasetheir systematic uncertainty.

I will say something more when I will have permission to bemore quantitative… After the blessing! ...".

Just Blessed!!!!!!!February 23, 2007. posted by Tommaso Dorigo

"... Julien blessed our new result of a Z->bb signal from RunII CDF data.

We measure a signal of 5674+1540-725 events, and a b-Jet energyscale factor of 0.974+0.020-0.018. The latter is the most precise andonly determination of this quantity so far. It basically tells us theamount of miscalibration of our jet energy corrections forb-jets.

Using the b-JES for the reduction of the top quark mass systematicuncertainty will require more work. ...

In the plot below, hot from the press, you see as blue points withtiny error bars the dijet mass distribution of experimental dataafter a tight selection aimed at increasing as much as possible theelusive Z signal. The black curve is the fit, the fittedZ signal is in red, and the QCD background is in green. In theinset you see the excess of data over the background alone, comparedto the fitted Z signal. ...



 Tony Smith's Home Page