Mysterious variance found by CMS in Run 1 stats not confirmed in 2016 data

Mysterious variance found by CMS in Run 1 stats not confirmed in 2016 data



Fig. one. The distribution of the selected events over the invariant mass of the muon pair in the signal region 1 (left) and 2 (on right). Black dots with errors – data of the CMS detector in the Run 1 session, colored histograms – contributions of background processes. Image from the article in question

Performing a routine search for New Physics effects in the channel of production of muon pairs and b-jets, the CMS collaboration found in the Run 1 data an unexpectedly strong deviation from the background at an invariant muon pair mass of 28 GeV. This could be a big revelation if Run 2 confirmed a deviation, but the 2016 data does not show such a deviation. In the absence of an understanding of how to map these two datasets, the CMS collaboration has so far refrained from making a verdict.

When building New Physics models, theorists usually supplement the Standard Model with new heavy particles with masses of the order of TeV and higher. When experimenters test the predictions of such models from the LHC data, they measure the cross section for the production of two or more particles (two photons, two leptons, or two hadronic jets) for different invariant masses and look for anomalous bursts against the smooth background of the Standard Model. In 2013–2015, when the Run 1 session data was intensively processed, such bursts did occur at times (we followed them on the Collider Riddles page). Then, under the onslaught of Run 2 data at a higher collision energy of 13 TeV, all these deviations, one by one, disappeared.

However, New physics can manifest itself in another way, in the form of relatively light particles, which are quite accessible to the collider in energy, but for some reason remain unnoticed. For example, they may very rarely be born at the LHC due to the weakness of their interaction with ordinary matter. Or they prefer to decay into such sets of particles in which the background of the Standard Model is very high. Therefore, every time experimenters accumulate a large amount of new data, they look for in them not only heavy, but also moderately light particles, with masses of several tens of GeV.

Recently, the CMS collaboration in its article Search for resonances in the mass spectrum of muon pairs produced in association with b quark jets in proton-proton collisions at sqrt (s) = 8 and 13 TeV (arXiv: 1808.01890) published the results of the search for new particles with masses from 12 to 70 GeV in the data of Run 1 and the first stage of Run 2, which was limited to 2016 (integrated luminosity 36 fb-one). In this work, the production of a b-quark-antiquark pair and a muon-anti-muon pair with sufficiently large transverse momenta was studied. Muons are well registered directly, and b-quarks are determined indirectly, by the presence of at least one b-jet (hadron jet containing a lovely hadron). The selection criteria for events took into account the momenta of all particles. Muons had to carry a transverse momentum of at least 25 GeV, their invariant mass had to exceed 12 GeV in order to cut off events with the creation and decay into muons of heavy hadrons. The jets should also have a significant transverse momentum, no less than 30 GeV. In this case, the b-jet had to hit the central part of the detector, and the second hadron jet either flew out in the direction pressed against the collision axis (signal region 1), or was also recorded in the central part of the detector (signal region 2). All these criteria were chosen to search for events with the production of a b-anti-b pair, which immediately emits a hypothetical light boson A, decaying into muons. Such particles are found in many models of New Physics, including multi-Higgs models or theories with new interactions and moderately heavy carrier particles.

The same CMS collaboration already looked for similar events in Run 1 data (see last year’s article Search for a light pseudoscalar Higgs boson produced in association with bottom quarks in pp collisions at sqrt (s) = 8 TeV (arXiv: 1707.07283), but nothing rushing I did not find it in my eyes then.In the distribution over the invariant mass of muons mμμ single bursts were observed, but due to the large residual background they did not arouse suspicion. Now the event selection algorithm has been improved, the background from the Standard Model has been significantly suppressed – and the data suddenly shows details that were not visible before.

In fig. 1 shows the distribution over mμμ in two signal areas. Almost everywhere the data agree well with the background of the Standard Model (colored histograms). However, in the 25–30 GeV region, there is a clearly noticeable difference in both plots, which affects not one point, but a small range of values. The local statistical significance of the difference in the first region is 4.2σ, and even after adjusting for the multiplicity of the sample, it remains at a solid 3σ level. In the second area, the local significance of the deviation is more modest and reaches only 2.9σ. However, the fact that it falls on exactly the same mass region emphasizes the unusualness of the deviation.

This message could be a big statement, capable of sparking a theoretical stream if Run 2 data confirmed a rejection. But alas, in the statistics of 2016 (the bottom pair of graphs in Fig. 2) the eye has nothing to cling to at all. If in signal area 1 in this place, at least some hint of a surge appears, then in area 2 there is a lack of events, and not an excess at all.


Fig.  2. The same distribution as in fig.  1, for Run 1 data (top row) and Run 2 data for 2016 (bottom row)

Fig. 2. The same distribution as in Fig. 1, for data Run 1 (top row) and Run 2 data for 2016 (bottom row). Dotted line shows the result of fitting (see Fitness approximation) to the data with only one background, solid curve – the best approximation taking into account the resonance. Image from the article in question

Does 2016 data cover the burst of the Run 1 session? The CMS collaboration adheres to very careful wording in its article. On the one hand, the authors have repeatedly checked all known sources of systematic errors and cannot attribute the surge to any of them. Could it be caused by some exotic physical process that weakens with increasing energy of proton collisions? The authors note that the Run 1 signal certainly cannot be the extra Higgs boson predicted by simple multi-Higgs models, since then it manifested itself much more strongly in other decay channels. But if it is not clear what could have caused the burst in the Run 1 session, then for the time being, one should refrain from directly comparing the two data sets at energies of 8 and 13 TeV. The collaboration is limited only to the general remark that new data and a fresh view of theorists on this situation are needed. Until the community comes to a consensus, the situation remains suspended.

Addition. In the comments to this news, the direct authors of the analysis once again emphasize that in the current situation, without a detailed understanding of what process could have caused the signal detected at 8 TeV, it is still impossible to make an unambiguous conclusion whether the run at 13 TeV confirms the Run 1 data or not. It can only be argued that the new data did not close the signal. In addition, the collision conditions at 13 TeV differ from the Run 1 session, and the selection algorithm has not been optimized for these changes. Therefore, the authors urge to be patient and wait for the CMS data update, a similar result from ATLAS, and listen to what the theorists have to offer.

Igor Ivanov

Recommended Articles

Leave a Reply

Your email address will not be published. Required fields are marked *