The Plot Of The Day
Starting today, I am going to make an attempt at creating interest in this blog by posting a figure from current research in HEP, with minimal commentary. This does not substitute other material that I wish to post (or see posted by my collaborators!), but rather should be seen as a small addition -an embellishment of this endeavour.
So for today’s plot-of-the-day: It is a figure from the recent ATLAS search for Supersymmetry in the missing Et plus jet final state. I discussed this analysis in the previous post (see below -am too lazy to link it!), and placed the same discussion in my other blog, A Quantum Diaries Survivor.
In the comments thread of the post in the QDS blog a reader actually corrected a mistake (a wrong figure from the ATLAS paper), and explained that my argument about ATLAS having been lucky to observe fewer events in their SUSY signal boxes than predicted by Standard Model backgrounds was bogus: ATLAS observed an excess in one of the four search boxes, and used only one box at a time in extracting their exclusion in the CMSSM parameter plane; so my argument about ATLAS being “lucky” and CMS being “unlucky” did not hold.
I countered by showing the following figure, of which I suspect the reader was already aware. This shows which of the four search boxes used by ATLAS (A, B, C, D, which differ by different cuts on the kinematical quantities observed in the selected events) is the one picked by the limit-setting method as the “a-priori most sensitive” in each point of the scanned parameter space.
As you can see, the region where the limit is sitting is all circled by D’s, and a few C’s. Now let us look at the number ofe vents seen by ATLAS in the four regions:
Region A: observed 87, expected 118 (with 40-ish error bar)
Region B: observed 11, expected 10.0 (+- 6, roughly)
Region C: observed 66, expected 88 (+- 30, roughly)
Region D: observed 2, expected 2.5 (+- 1.5).
This clearly means that the one upward fluctuation occurred in one of the two regions which are actually irrelevant as far as the boundary at high m0:m_1/2 is concerned. Which is obvious, since the observed limit is above the expected one throughout the plot!
That said, I will stop arguing that ATLAS was lucky! This is a very uninformative issue. The only informative issue is that ATLAS used a better search method for SUSY, and that ATLAS used a rather obsolete (albeit revamped) method for limit setting, one which I think will be abandoned again in the future.