SM Higgs limits: how well is CDF doing ? August 17, 2007Posted by dorigo in news, physics, science.
In a post about Higgs limit predictions last May I discussed how, with some assumptions and approximations, the combined limit then presented by the D0 collaboration on Standard Model Higgs boson production could have been used to verify how well their searches were doing with respect to predictions made as far back as 1999 (and then revised in 2003).
I will not repeat here the assumptions I used there, nor the method – which basically assumes a simple scaling of the limit reach with the square root of integrated luminosity, and the same performance by both CDF and D0 detectors on all Higgs searches.
These assumptions are only good to first order. But that is just fine for a hand-waving argument such as the one I am making. Also, check
there for the relevant prediction plots. (I am quite lazy today – must be the rain).
In a nutshell, the results of the analysis of the prediction curves of 1999 are the following: with 1.5/fb per experiment a 110 GeV Higgs should be excluded by a CDF-D0 combination; at 140 GeV, 10/fb would be needed for a 95% CL exclusion; at 160 GeV, 4.5/fb;
and at 200 GeV 80/fb per experiment.
Now take the new combined 95% CL limit results by CDF. They are shown in
the post I wrote yesterday, and they still do not include the analysis of one of the search channels (ZH associated production with a dilepton plus dineutrino final state).
The analyses have used samples of data varying between 1 and 1.9 inverse femtobarns,
and so we will assume the average value of 1.45 for all, erring a little here and a little there. Now, we take that luminosity and divide by two: the result becomes what CDF and D0
would find together by combining their results with half the luminosity. So we consider a base
luminosiy of 0.72/fb per experiment. The ratio of 0.72/fb to 1.5/fb, 10/fb, 4.5/fb, and 80/fb are factors which, once square roots are taken, represent the “times the SM” limit that a combination of CDF and D0 would be setting according to 1999 predictions. One gets 1.4, 3.7, 2.5, 10.5: these are the values where the expected CDF limit curve should lie today, at Higgs mass values of 110, 140, 160, and 200 GeV, respectively.
Confused ? Don’t be. Let us just read off the CDF combination limit plot what are the actual limits they obtain today. At 110 GeV, CDF expected to exclude cross sections above six times the SM Higgs production (they did worse in the actual limit, but that is a fluctuation
and we will not consider it here, since we are comparing predicted limits and limits expected with the actual data and analyses). At 140 GeV, CDF expected to exclude cross sections of 7 times the SM; at 160 GeV, 3.5xSM; and at 200 GeV, 10xSM. So the bottom line is that CDF is doing 4 times worse than expected in searching for a 110 GeV Higgs boson;
at 140 GeV, they are doing 2 times worse; at 160 GeV, 40% worse; and at 200 GeV they are doing just as they thought in 1999.
The plot below shows graphically and at a glance the result of the exercise discussed above. Overimposed to the new CDF limit plot, you can see blue stars at the points where the 1999 Tevatron SUSY-HIGGS working group predicted one Tevatron experiment with 1.5/fb of data would set 95% CL limits to Higgs boson production. I must say the predictions made 8 years ago were not that much off the mark!
Now, before I get flamed by my CDF colleagues and then denied authorship for the next hundred years, let me say that I do think – and you should too – that the current CDF results are not only really impressive, but they are actually much better than what one would gather by lazily looking at the picture:
— First, you must realize that not all analyses have been combined in the preliminary combined limit. Some will make little or no difference, others may make some.
— Second, it is quite normal that limits at low values of Higgs mass are worse than expected: these are extremely hard analyses which exploit technologies which improve as more data is collected. An efficient b-tagging, advanced methods to improve the dijet mass resolution, and the exploitation of leptons with high rapidity through sophisticated algorithms are fundamental ingredients to tag associated production of Higgs and vector boson, which are the golden mechanism to find the Higgs at low mass. I fully expect that in the next few years the performance of low-mass Higgs searches will at least double.
— Third, do not forget that the predicted limits are defined as “what the combination of CDF and D0 expects to set 50% of the times”. The variability of the actual result with respect to that average is quite large! CDF and D0 might get lucky, and actually discover the Higgs even if on average they would need five times more statistics to do so!
All in all, one learns that the game of excluding the existence of a Higgs boson is still on at the Tevatron, and that it is quite possible that a 160 GeV Higgs will be excluded – or found! – next year. For lower masses, the Tevatron experiments will need to be even more patient, and collect all the luminosity they can, working further at their analysis methodologies. But if the Higgs mass is 115 or 120 GeV, the LHC experiments will take a long time themselves to find it, so… You better find a comfortable chair!