This week's graph is a reminder that particle physicists are, deep in their bones, bump hunters. Sure, some of my colleagues could best be described as detector builders; others as software wizards; still others as statistical gurus. But what excites us the most is to go hunting for a bump in a mass histogram. 
That's because spikes in mass histograms have been the way to produce particle discoveries ever since particle physics became a discipline, some eighty years ago. What is more exciting than findind a new resonance, which maybe forces a complete overhaul of our understanding of matter at the subnuclear scale ? It could be the next Nobel price; why not? It was so for the J/Psi meson, for the Upsilon, for the W and Z bosons, for the Higgs. Oh, wait - the Higgs did not grant a Nobel prize to experimentalists, but still.

So what can be more entertaining today for particle physicists, as well as for laypersons who happen to have an interest in the science we do, than to examine a brand new mass spectrum, obtained with all the data so far collected by the ATLAS experiment at the unprecedented collision energy of 13 TeV ? There might be a new particle hiding here. Or maybe not. Check it out for yourselves.

The spectrum shows the ATLAS data in bins of reconstructed dijet invariant mass. If the two jets originate from the decay of a new resonance X, as in X->jj, then we expect that some data will pile up at a specific point in the abscissa. Once added to the large background coming from non-resonant, regular QCD dijet production, such a X signal might be invisible. Because of that, even if physicists do not see a bump they avoid concluding that there is none; all they do is to place an upper limit on the rate of the X production process, reasoning that if the rate was higher than a certain value, the bump would have been evident - while they did not see one.

The ATLAS graph is a bit too busy for my taste - it includes a lot of information on the fit, new physics models for excited quarks and black holes, systematic uncertainties, the data minus fit residuals, the bump-hunter program result, etcetera. We may just concentrate on the latter here: bump-hunter is a simple program that scans the distribution in search for the most significant upper fluctuation of data in neighboring bins. In the case at hand the two bins at mass of about 1.5 TeV turn out to make the most significant fluke, but once one factors in the look-elsewhere effect this is indeed not a very interesting feature.

ATLAS does much more than searching for bumps in the above spectrum in the paper they just released. I have no chance to comment on the results they obtain, but if you are interested you should give it a look. And if you are interested in the general topic of dijet bump hunts, why not checking also the CMS article on the same topic, which came out just a few days ago ? By the way, neither CMS sees a bump in its data...