Wednesday, June 08, 2016

ATLAS+CMS: excess of \(ttH\) production

...and a few other deviations...
Fifth force: First, off-topic, Natalie Wolchover wrote a helpful article about the Hungarian claims of a new 17 MeV boson, pointing out that the researchers in Debrecen (the town that gave the name to the popular Czech Debrecen ham) have "discovered" a dozen of similar bosons in recent years and they have viewed null results as a "failure", thus proving their lack of integrity.
ATLAS and CMS have combined their Higgs production-and-decay analyses and many numbers agree with the SM predictions within 2 sigma. However, numerous measured numbers don't agree so well.

Look at page 22, second paragraph from the end. The most interesting paragraph starts with "The \(p\)-value".

You are immediately shown a quantity that deviates from the Standard Model by 3 sigma:\[

\frac{\sigma_{ttH}/\sigma_{ggF}}{\text{the same ratio in SM}} = 3.3\pm 0.9

\]Note that the right hand side would be "one" if there were a "perfect" agreement of the theory with the observations. That doesn't mean that the cross sections are the same; it's one because the cross sections were divided by one another in the same way. However, the measured cross section ratio is 3.3 times higher.

So either the \(ttH\) production (top-top-Higgs) seems more frequent than predicted; or \(ggF\) production (gluon fusion) seems less frequent than predicted. We are told that "multi-lepton categories" contribute to this deviation.

That's not the only similar ratio that looks off. An analogous cross section ratio\[

\frac{\sigma_{ZH}/\sigma_{ggF}}{\text{the same ratio in SM}} = 3.2\pm 1.4

\] boasts some 2-sigmaish excess, mainly due to the \(ZH\to ZWW\) subchannel. Also, the ratio of branching ratios\[

\frac{B^{bb}/B^{ZZ}}{\text{the same ratio in SM}} = 0.19\pm 0.21

\] which they say to be a 2.5-sigma deficit relatively to the value one, i.e. the Standard Model. So the Higgs seems to decay to a bottom quark-antiquark pair less frequently or to the electroweak boson pairs more frequently. This deviation in the ratio of branching ratios seems very strong on Figure 9.

(I have always indicated the symmetric \(\pm\) error margins. But the paper always reports asymmetric ones and to simplify things, I have chosen the unified error margin from the "more relevant side", the side towards the theoretical prediction.)

Now, it is very interesting and such deviations may suggest a more complicated Higgs sector than the Standard Model Higgs sector. Maybe, there is least one other Higgs boson which "specializes" in those bottom decays and takes some work from the \(125.09\GeV\) Higgs. The supersymmetric standard models and their extensions could probably have an explanation if these decays were real.

Note that only the \(\sqrt{s}=7\TeV\), \(8\TeV\) data from 2011 (5/fb) and 2012 (20/fb) have been used.

One complaint is that they evaluated various ratios of cross sections and branching ratios. There are many ratios you may consider and some of them are bound to deviate more than others. If you take the cross section with the greatest excess and divide it by the cross section with the greatest deficit, you will increase the excess even more. So the numerous possible ratios increase the look-elsewhere effect and I am not sure whether this extra look-elsewhere effect has been used to "punish" the declared confidence levels.

Why do they talk about the ratios and not the individual cross sections and branching ratios themselves? It's because if one takes the ratios, the theoretical uncertainty is almost zero. I don't know why it's so important because some of these are quoted to be as low as 5 percent etc.

I wouldn't get carried away by similar 2-sigma and 3-sigma excesses, especially because they are deviations in rather artificial quantities we weren't thinking too much about previously. But as always, some of these 3-sigma and maybe even 2-sigma deviations may turn out to be real. The LHC detectors have already collected 4/fb (2015) plus 3/fb (2016) of the collisions at \(13\TeV\) which may strengthen or weaken all of these deviations.

No comments:

Post a Comment