In the previous post I explained that the CMS Higgs plots shown at Lepton-Photon were not as good as they could have been because the MVA analysis for the WW channel was not ready. In fact the earlier MVA based analysis for WW shown at Europhys is better in theory than the Cut-based analysis shown at LP. The best combination plot should use all the best data from each channel over all the experiments and that means using the MVA WW channel from EPS for CMS. So here is the best possible plot (in theory) using all the best LHC data and the latest Tevatron data. I have even added in the LEP result for the first time .
This is a pretty nice result. Everything is excluded at 95% confidence in the mass range from 145 GeV to 460 GeV, but there are small excesses over the range from 115 GeV to 145 GeV. A good thing to notice about this plot is that the expected CLs line is below the 95% confidence limit all the way up to 500 GeV. If there were no Higgs boson in that range they would expect to have excluded it, but they haven’t.
At this point I think that signal plots are more informative than the exclusion plots so here is how the signal changed between EPS and LP.
Remember that in these plots a level at zero indicates the absence of a Higgs Boson and a level at one indicates the presence of a Higgs boson. So the signal in the 115 GeV to 145 GeV is roughly the right strength for the Standard Model Higgs but it has dropped in the 140GeV area with the latest data as the error bands have tightened (though not as much as we previously thought). Now we should just wait for more data to resolve the low mass range, or should we?
I want to make the case that there is a big problem with these plots. Our expectation is that as more data comes in a sharp peak (or two) will emerge somewhere in the low mass region to reveal where the Higgs is. However, the plot is dominated by the WW channel over most of this range and the WW channel has low resolution. This is because it uses missing energy observations to construct the underlying mass of the events. The W’s decay into neutrinos which can never be detected directly. The result is that the Higgs appears as a broad excess in the WW channel and you can’t locate it well. The WW channel is great for excluding large ranges of the mass spectrum, but it is not good for pinpointing a low mass Higgs that has a narrow width.
Furthermore, the situation will not improve as more data is added. The WW channel will always remain low resolution and it will always dominate the combination plot. Sadly the Tevatron data has the same problem. It is dominated by WW and bb channels with neutrinos in each case. In fact the detectors themselves have poorer resolution and even the digamma and ZZ channels are only ever plotted at 5 GeV intervals for the Tevatron. So what should we do? if some data could be making the plot worse the best thing is to remove it and see what we get. So here are the signal plots without WW channels and without Tevatron data.
These plots use mostly the digamma and ZZ channels from the LHC. The LEP data is also retained because it is high-resolution too. How many Higgs bosons do you see now? Above 160 GeV the Higgs is strongly excluded by WW data in the earlier plots. Without WW, the ZZ channel dominates but the background is high above 155 GeV which accounts for the large error bands. So the bump centred on 165 GeV can be safely ignored.
What is particularly interesting now is the bump at 140 GeV. Some people said that this excess came mostly from the WW channel, yet when the WW channel is removed the bump is still there with nearly 2-sigma significance. The two bumps peaking at 118 GeV and 128 GeV are also the right size for a Higgs signal but error bands are still too big. Any of these bumps could be statistical fluctuations but it is very unlikely that they all are.
With current data available in the high-resolution channels it is not yet possible to draw robust conclusions, but I think I have demonstrated that this will be the best way to find the Higgs with future data. I hope the experimenters will take note and produce similar plots from the official data.
Updated results with 2.5/fb could appear within weeks and we will see where the three candidate bumps are heading. With 5/fb or more by the end of the year these plots will be showing strong signals and with 10/fb or more by the end of 2012 the Higgs discovery should be conclusive from these plots, unless it isn’t there.