Tevatron Higgs Results – Evidence?
The bottom line is:
- 2.5 standard deviations for all channels
- 2.9 standard deviations for bb channel alone
These are global p-values. The local p-values are 3.0 s.d. and 3.2 s.d. respectively. Below are my notes and comments from the presentation today.
The Tevatron Higgs group (CDF+D0) just presented their update – and very nearly final – results for the search for the Higgs boson. Eric James gave the first half (CDF results) and Wade Fisher gave the second half (D0 results and Tevatron combination). Although the Tevatron has stopped taking data, and we have seen impressive results from the Tevatron last winter based on essentially all of their data, this update is important because the D0 analyses have been improved significantly, thereby improving the sensitivity of the Tevatron Higgs search. (Recall that they had an excess at about the 2.5σ level last winter.) The Tevatron data set corresponds to 10 fb-1 recorded data per experiment.
Eric started by reminding us of the impact of indirect constraints on the Higgs boson, valid within the standard model (SM). At present those indirect constraints on MH are consistent with the range not already excluded by LEP, the Tevatron and the LHC. One should not forget that the precision measurements of the top and W masses play a key role in those indirect constraints. Wade stated that the Tevatron searches took as their goal the ability to exclude a SM Higgs boson across the full range favored by the precision electroweak data: 100 – 150 GeV, roughly. They very nearly achieved this goal.
CDF and D0 both search for Higgs decays to γγ, WW, ZZ and bb, but the most sensitive channel by far is pp→VH with V→L or ν and H→bb. The sensitivity of each channel is a strong function of the Higgs mass, MH, but for the interesting region (120 < MH < 140 GeV), the γγ, WW and ZZ channels are still several times the SM signal because the production cross sections are really very small. The bb channel, on the other hand, has a sensitivity near 2×SM or even better. As Eric pointed out, this is actually better than what CMS and ATLAS have achieved so far in this channel.
Eric and Wade, and the entire Higgs search teams that they represent, recognize this channel (“Vbb”, or pp→VH + H→bb) as the main opportunity for the Tevatron to make a major contribution to understanding the Higgs, if there indeed is a Higgs with a mass near 125 GeV. (LHC will make a statement about that on Wednesday.) One should not underestimate what these teams can do, as Eric nicely illustrated with a plot of sensitivity versus integrated luminosity. To be plain, the Tevatron Higgs searches today exceed even the most optimistic projections from five years ago. Inspiring. Eric also showed that the significance of any excess around 125 GeV at the Tevatron in the Vbb channel is quite comparable to what ATLAS and CMS can achieve with other channels.
Wade explained succinctly that the D0 Higgs searches have all improved due to technical improvements and the addition of a little more data. The improvements are on the oder of 10% per channel, and some small improvements are still expected over the summer. One the main technical improvements comes from splitting backgrounds in any given channel into categories which are then suppressed individually. This bolsters the S/B ratio and was also done by CDF (and the LHC experiments).
Both CDF and D0 see a fairly broad excess in the MH range 120 – 140 GeV. The characteristics of this excess are very similar for CDF and D0, appearing most clearly in the Vbb channel, with only weak signals in γγ and ZZ, and none in WW. (Recall that the LHC results hint at an enhancement in γγ and a slight deficit in ZZ and WW, assuming SM signal strengths.) Here is the comparison of the exclusion limits from CDF (left) and D0 (right):
Naturally, since they are so similar, the combined exclusion limit will be about the same, only sharper in its features:
Since the observed limit curve lies well above the expected one, we know that we have an excess. This excess comes mainly in bb. Wade showed the following plot to quantify the excess: This plot shows the signal strength μ (σ×BF normalized to the SM calculation) for the three main channels. Clearly the WW channel indicates no signal, while the γγ has low precision (though it does not favor zero). The interesting bit is the bb channel, showing a signal strength of approximately μbb=2.0±0.7 – according to my ability to read the graph… The plot on the right shows the probability density for the bb channel; the most probable value is indeed μbb=2 and μbb=0 is unlikely, according to this picture.
The Tevatron Higgs group likes to show a likelihood ratio as a function of mass, and indeed the plot is informative:
The likelihood ratio (LLR) shows the separation of the signal hypothesis from the null hypothesis. For a given MH, there will be two Gaussians showing the possible outcomes if a signal is present or absent. If these Gaussians are well separated, then one has very good sensitivity to the signal. The data then pick one Gaussian or the other. Looking at the plot of LLR vs. MH, the dotted black line shows the expectation if the null hypothesis is correct, while the red dotted line shows the expectation of a signal is present. The data from the Tevatron, represented by the solid black line, twists and turns as a function of MH due to statistical fluctuations in the data. But one can clearly see a preference for the signal hypothesis for MH in the 115 to 135 GeV range, and a preference for the null hypothesis elsewhere. The plot on the right shows the ideal case if a Higgs signal is present with MH = 125 GeV and with a signal strength μ=1.5. The value μ=1.5 comes from the statistical analysis of the Tevatron data; basically the Vbb `signal’ is a bit stronger than one would expect with for the SM Higgs boson; I eye-ball it to be μ=1.4±0.6.
This result is impressive and exciting, but Wade and the Tevatron Higgs community made a rather cautious, sober statement: The significance of the excess is still under 3σ, so we are not making any announcements today. Still, the Tevatron results are getting interesting. I’ll say!
In my opinion, this result from the Tevatron is extremely important. While it does not constitute discovery of a Higgs boson, or even `evidence’ in a technical sense, it does illuminate the issue of a new state if indeed the LHC experiments have independent evidence for something at 125 GeV. And it cannot be over-emphasized that this result is completely independent of what goes on at the LHC: the beam energies and particle types are different, the final state is different, and the analysis teams are different.
If there is a new particle, some sort of Higgs boson, at 125 GeV, then I certainly believe it decays to bb with a fairly large branching fraction.
So the Tevatron did not `scoop’ the LHC but it will play an important part in the coming months and years to elucidate whatever Nature will give us.
PS: sorry for the poor quality of the images, this is the best I could get while sharing the live broadcast with more than 1200 other people. For more information, see the Fermilab press release.
PPS: Of course many excellent blogs have already written about the Tevatron results. I recommend that you visit viXra log, Tommaso Dorigo and Resonaances and keep an eye on Matt Strassler and Not Even Wrong, among others…
Entry filed under: Particle Physics.