jump to navigation

New CDF Combination of Higgs limits! February 11, 2009

Posted by dorigo in news, physics, science.
Tags: , , ,

A brand-new combination of Higgs boson cross-section limits has been recently produced by the CDF experiment for the 2009 winter conferences. The results are almost one month old, but I decided to wait a bit before posting them here, in order to avoid arising bad feelings in a few of my CDF colleagues, those who believe I have no right to post here published results in too timely a fashion, because they feel those results should first be presented at conferences by the real authors of the analyses.

Now I think it is due time to have the most relevant plots here, since they are all available from a public web page of CDF anyway; so here we go, with the most updated information. Mind you, these are CDF-only results: a sizable improvement in the limits will come when they get combined with the findings of DZERO. I seem to understand that the Tevatron combination group folks are dragging their feet this year, so we have better to just as well take the CDF results and comment them alone.

The first graph is the most important one of all: it describes the combination of CDF results, in the usual “95% CL limit on times the SM cross section“. It is shown below.

On the x axis is the Higgs boson mass, and on the y axis the cross-section limit. Different colors of the curves refer to different analyses, which target the various decay channels of the sought particle; hatched lines show expected limits, and full ones show instead the limits actually obtained by the analysis.

As you can see by examining the thick red curve at the bottom, CDF by itself cannot rule out the 170-GeV point, which last summer was excluded by the CDF+D0 combination. However, a sizable improvement can be observed across the board in the results. The red curve, for one thing, is considerably flatter than it used to be, a sign that the low-mass searches have started to pitch in with momentum. Another thing to note is that these results correspond to 3.0/fb of analyzed luminosity or less (2.4/fb at low mass): there is already at least twice as much data waiting to be analyzed, and results are thus expected to sizably improve their sensitivity.

The above summary brings me to mention another important point. By looking at the graph, you might run the risk of failing to appreciate the enormous effort which CDF is putting into these searches. In truth, the name of the game is not “wait more data and turn the crank”. Quite the opposite! The most important improvements in the discovery reach have been achieved in the course of the last five years by continuously improving the algorithms, the search methods, by refining tools, by finding new avenues of investigation, and new search channels neglected before. This is summarized masterfully in the two graphs shown below.

Above, you can see that for a Higgs mass of 115 GeV, the limit that CDF was able to set on its existence, in terms of cross section (well, “times the SM cross section” units to be precise: the ones shown on the y axis) has improved much more than what one would have expected by scaling down the limit with a simple square root law -the one that Poisson Statistics would dictate, for statistically-limited measurements. Quite the opposite: as time went by, the actual limits (colored points) have moved down almost vertically, a sign that the data has been used better and better! Above, if you took the extrapolation expected after the first limit was published (the one in green), you would expect that the limit today, with 2.4/fb analyzed, was at 7xSM, while it in fact is at 3xSM: this corresponds to a 2.3x improvement in the limit, which would have been granted by a 5.2 times larger analyzed dataset!!

Above, the same information is shown for the M_H = 160 GeV value. In this case, CDF is now expected to be able to set an exclusion alone with 9/fb of data, but we still expect to see some improvements in the data analyses, which should move the points well into the brown band. In this case, 7/fb of data might be enough.

The last two plots I wish to discuss are shown below. BEWARE: This is information that LHC scientists would really, really not like to see – so, if your life depends on the success of ATLAS or CMS, please stop reading now, take my advice.

OK. The plot below shows the probability that the Tevatron experiments, by combining their datasets and results, may observe a 2-sigma evidence for SM Higgs production, with 5/fb and 10/fb of data collected by each. If the analyses will not perform better than what they have so far, you get the full lines -red for 5/fb, blue for 10/fb. If they improve as much as it is reasonable to predict, you get the hatched lines.

What to gather from the plot ? Well: it seems that, regardless of the Higgs boson mass, the Tevatron has a sizable chance to be able to say something good, by the time CDF and D0 will have analyzed the datasets they already possess (which are in excess of 5/fb each: the delivered luminosity of the Tevatron is passing the 6/fb mark as we speak, and the typical live time of the experiments is above 80%).

Below, we see what is the chance of a 3-sigma evidence. Again, there is a sizable chance of that happening, although if no additional improvements occur in the analyses, it seems that the Tevatron will need to get lucky!

I remember that in 2005 I gave a talk in Corfù (Greece) where I ventured to speculate on the possible scenarios for Higgs searches at the Tevatron and the LHC. One of the scenarios saw the two experiments competing to find the particle with roughly equal reach, and eventually producing a combined observation. That possibility  does not seem too far-fetched any longer!

In the next few days I plan to discuss in some detail the most important analyses which contribute to the combination discussed above. Stay tuned…


1. Anonymous - February 11, 2009

Interesting that at 150 GeV, H -> WW observed is a tiny bit better than everything combined at that mass. So is removing the WWW analysis one of the foreseen improvements? 😉

2. onymous - February 11, 2009

How soon can we expect to see this combined with D0? It will be fun to watch the 170 GeV ruled-out point expand into a ruled-out interval….

3. Daniel de França MTd2 - February 11, 2009

What do you mean as the probability to observe an excess of something? Does it mean that there is a certain probability that certain process actualy shows Higgs is 1%? That means Higgs were found but people are not sure if they are really Higgs?

4. dorigo - February 12, 2009

Hi Anon,

that is precisely right: improvements in such a high-cuisine recipe as a combined limit may come from unexpected sources sometimes… If the WWW analysis insists in getting too many events observed over background predictions, it should not be averaged, because either (1) there is something not right in the estimates or (2) there is some additional process contributing. In the latter case, that process is not Higgs production, which -at the level it should contribute to the WWW data- would be utterly inconsistent with the WW limit. So one should draw the obvious conclusion that it should be removed from the average.

I do not think we are at that point yet. In any case yes, foreseen improvements include better background predictions…


5. dorigo - February 12, 2009


as I mentioned in the post, things are not working fast in the combination. I believe something might be ready for Moriond and La Thuile in mid-March, but it is more likely that we’ll have to wait for the summer.

Daniel, the probability has nothing to do with any particular event. It has to do with the chance that the Tevatron experiments, by combining their analyses, observe an excess of events that is consistent with Higgs production, and that excess has a significance of 2-sigma or 3-sigma. For instance, imagine that for a M=120 GeV search CDF sees 110 events in one search dataset, when 100+-10 are expected, and 30 in another where 25+-5 are expected; and that D0 sees the same thing. You may combine the four +1-sigma observations together with advanced statistical tools, and you will come up with roughly the same answer you could have computed on the back of your envelope: globally, a 2-sigma excess.


6. Daniel de França MTd2 - February 12, 2009

Hi Tommaso. What I said was confusing, but what you answered is what I asked… well, at least what I inteded to say… :O

7. Daniel de França MTd2 - February 12, 2009

I don’t get something else. The third graph excludes SM higgs at 160GeV with 10fm, while at the last one one it show evidence of events consistent with Higgs peaking at 160 with 3sigma at 10fm. I don’t get it, it seems a contradiction to me.

8. Daniel de França MTd2 - February 13, 2009

I still can’t figure out…

9. The 1999/2003 Higgs predictions compared with CDF 2009 results « A Quantum Diaries Survivor - February 13, 2009

[…] the Higgs reach of the Tevatron with a rather solid set of hard data: the CDF combination, which I briefly discussed two days ago, is based on analyzed sets of data ranging from 2 to 3 inverse femtobarns, and the comparisons do […]

10. dorigo - February 13, 2009

Daniel, these are two different questions. One is “if the higgs does not exist (ie it does not contribute to the datasets), how much data it takes to exclude its existence ? The other is “if the Higgs exists and has a mass M, how likely is it that I will see evidence (at 2-sigma, or 3-sigma) with this much data ?”


11. Thomas Larsson - February 13, 2009

Does this mean that the odds for a light Higgs are less than a year ago?

12. dorigo - February 13, 2009

Hi Thomas,

nope. The fact that we limit the cross section does not have a direct connection to the likelihood of the Higgs existence, until we get close to the SM xs with our 95% CL limit…

13. Daniel de França MTd2 - February 13, 2009

But the highest probability of Higgs mass is around 165, right?

14. dorigo - February 14, 2009

Nope Daniel, those graphs do no tell you what is the highest probability where the Higgs is. They tell you how likely it is that so much data at the Tevatron is enough to find the Higgs at 2- or 3-sigma, if its mass is such.

What you want is instead an electroweak fit, which I have discussed elsewhere. I can give you the coordinates if you wish, or just search for “LEP electroweak working group”.

If you instead insist in asking what is the most favoured Higgs mass from the data analyzed at the Tevatron, well – you can just look where the obtained xs limit is higher than the expected one. That is a most likely sign of a background fluctuation, but it might even be due to some Higgs bosons contributing there. The highest departure is at 130 GeV, but it is ridicule to use the graph that way, since the actual expected amount of Higgs bosons in the analyses that found more events than expected there is too small to have caused a difference.



15. Daniel de França MTd2 - February 15, 2009

Let me see if I get it. This is not a combination for bounds on Higgs mass but limits on how much data, or luminosity, is necessary to include or exclude higgs, using as possible plots the parameter of higgs mass, luminosity, its variations for different experiements and luminosity.

16. dorigo - February 15, 2009

Daniel, it depends on which graph you are looking at !!!

17. Daniel de França MTd2 - February 15, 2009

So, the fact that there is a higher confidence of finding higgs around 165 means that it is more likely to exclude it at these levels, which happened at 170, right?

18. dorigo - February 15, 2009

Yes, there is a higher sensitivity there. You can see the sensitivity for a 95%CL exclusion, a 3-sigma evidence, or a 5-sigma observation as a function of Higgs mass in a plot I posted in a more recent post, the one of February 13th.


19. Too Many Topics « Not Even Wrong - February 15, 2009

[…] the Tevatron will manage to see the Higgs or rule it out, see excellent postings by Tommaso Dorigo here and here. The bottom line is that by the time the LHC has enough data to start saying something […]

20. Ptrslv72 - February 17, 2009

Hi Tommaso,

from the webpage that you linked I see that the CDF analysis is based on the prediction of hep-ph/0306211 (supplemented with electroweak corrections) for the cross section for Higgs boson production in gluon-gluon fusion. In particular, the value used for the cross section at mH=170 GeV is 367.6 fb. However, two of the authors of hep-ph/0306211 have recently published an update of their analysis, 0901.2427, in which the implement all of the current state-of-the-art ingredients for the calculation of the cross section (including EW corrections and new PDFs). They find that the central value of cross section at the Tevatron varies by +- 9%, depending on the Higgs mass, w.r.t. their earlier prediction. In particular, for mH=170 GeV they find 349 pb, i.e. 5% LOWER than the value used in the CDF analysis. How bad is this news for the CDF/D0 exclusion limit?

Cheers, Ptrslv72

21. dorigo - February 17, 2009

Hi Ptrslv,

I guess it only makes things 5% worse 😉 However , I’ll have a look at the new analysis first. Thanks for pointing it out.


22. immenodig - March 20, 2009

very intresting

Sorry comments are closed for this entry

%d bloggers like this: