3 ms·

Official ATLAS and CMS plots: https://atlas.web.cern.ch/Atlas/GROUPS/PHYSICS/CONFNOTES/ATLAS-CONF-2015-081/ https://atlas.web.cern.ch/Atlas/GROUPS/PHYSICS/CONF

• cshimmin 11y ago
  > While exciting, there are countless examples in history of such discoveries disappearing as more data is collected. As someone who works on ATLAS and has been watching this result develop over the last several weeks (without knowing what was going on at CMS), my excitement has been very tempered by this fact. Especially given that the global p-value is quite modest (see also: Look Elsewhere Effect [1]). For sure it's too early to know anything. But I have to say, when I heard the CMS announcement yesterday that they were also seeing it, in almost exactly the same place, it's much more exciting! [1] https://en.wikipedia.org/wiki/Look-elsewhere_effect https://en.wikipedia.org/wiki/Look-elsewhere_effect
  • idlewords 11y ago
    A domain expert on an Internet forum? Prepare to be corrected!
  • jessriedel 11y ago
    On the optimistic side: has anyone done an estimate of the global significance taking into account the fact that CMS and ATLAS both saw the excess at the same place (~740 GeV)? (Clearly, if the situation was a simple as each bin being "excess" or "no excess", then you would only correct for the look-elsewhere effect for one of the experiments. When you turned your attention to the second experiment, that bin would already be special. This is muddied by the fact that things are not so clean, and the small blip of CMS was just the largest of several blips, etc.) On the pessimistic side: has anyone done a "fully global" significance test, taking into account all the other bump hunts of similar a-priori interest?
    • cshimmin 11y ago
      > has anyone done an estimate of the global significance taking into account the fact that CMS and ATLAS both saw the excess at the same place (~740 GeV)? Not yet, and it's actually a bit tricky to think about how this should be interpreted. Since neither experiment alone provides a super strong motivation to look at 750 GeV (since their global significance is low), it's hard to argue that one experiment can remove the LEE from the other by fixing the search mass to 750 GeV. But IMO it doesn't even make sense to do this thing post-hoc; otherwise one could always avoid the LEE by "fixing your attention" on a single hypothesis mass where you had previously observed a local excess. Now, suppose you had a (probably contrived) theory which specifically predicts a particle at this mass in a way that is totally non-negotiable, so that the only meaningful question to ask about the theory is "is there a particle X at 750 GeV?". In this case you can do away with the LEE and use the stronger local p0 at that mass. Theorists are already flooding the arXiv with such articles (and even have been since a day before the announcement... oops!). But in the general case, really the best way to understand the combined significance is to combine the datasets from CMS and ATLAS in a fully correlated way that handles systematics etc, and then make the diphoton mass plot again with the bigger dataset and evaluate the global significance of the resulting bump, still accounting for the LEE. But this is a really big effort, since the two detectors are substantially similar/different in many ways and it takes great care to handle correlations etc between their resulting data. Something like this was done for the Higgs discovery [1], in a paper which you might recall made news [2] for having more authors than any paper in history! [1] http://arxiv.org/abs/1503.07589 http://arxiv.org/abs/1503.07589 [2] http://www.nature.com/news/physics-paper-sets-record-with-more-than-5-000-authors-1.17567 http://www.nature.com/news/physics-paper-sets-record-with-mo...