Gunning for the goddamned: ATLAS results explained

Here are some of the photos from the CERN webcast yesterday (July 4, Wednesday), with an adjoining explanation of the data presented in each one and what it signifies.

This first image shows the data accumulated post-analysis of the diphoton decay mode of the Higgs boson. In simpler terms, physicists first put together all the data they had that resulted from previously known processes. This constituted what’s called the background. Then, they looked for signs of any particle that seemed to decay into two energetic photons, or gamma rays, in a specific energy window; in this case, 100-160 GeV.

Finally, knowing how the number of events would vary in a scenario without the Higgs boson, a curve was plotted that fit the data perfectly: the number of events at each energy level v. the energy level at which it was tracked. This way, a bump in the curve during measurement would mean there was a particle previously unaccounted for that was causing an excess of diphoton decay events at a particular energy.

This is the plot of the mass of the particle being looked for (x-axis) versus the confidence level with which it has (or has not, depending n how you look at it) been excluded as an event to focus on. The dotted horizontal line, corresponding to 1μ, marks off a 95% exclusion limit: any events registered above the line can be claimed as having been observed with “more than 95% confidence” (colloquial usage).

Toward the top-right corner of the image are some numbers. 7 TeV and 8 TeV are the values of the total energy going into each collision before and after March, 2012, respectively. The beam energy was driven up to increase the incidence of decay events corresponding to Higgs-boson-like particles, which, given the extremely high energy at which they exist, are viciously short-lived. In experiments that were run between March and July, physicists at CERN reported an increase of almost 25-30% of such events.

The two other numbers indicate the particle accelerator’s integrated luminosity. In particle physics, luminosity is measured as the number of particles that can pass detected through a unit of area per second. The integrated luminosity is the same value but measured over a period of time. In the case of the LHC, after the collision energy was vamped up, the luminosity, too, had to be increased: from about 4.7 fb-1 to 5.8 fb-1. You’ll want to Wiki the unit of area called barn. Some lighthearted physics talk there.

In this plot, the y-axis on the left shows the chances of error, and the corresponding statistical significance on the right. When the chances of an error stand at 1, the results are not statistically significant at all because every observation is an error! But wait a minute, does that make sense? How can all results be errors? Well, when looking for one particular type of event, any event that is not this event is an error.

Thus, as we move toward the ~125 GeV mark, the number of statistically significant results shoot up drastically. Looking closer, we see two results registered just beyond the 5-sigma mark, where the chances of error are 1 in 3.5 million. This means that if the physicists created just those conditions that resulted in this >5σ (five-sigma) observation 3.5 million times, only once will a random fluctuation play impostor.

Also, notice how the differences between each level of statistical significance increases with increasing significance? For chances of errors: 5σ – 4σ > 4σ – 3σ > … > 1σ – 0σ. This means that the closer physicists get to a discovery, the exponentially more precise they must be!

OK, this is a graph showing the mass-distribution for the four-lepton decay mode, referred to as a channel by those working on the ATLAS and CMS collaborations (because there are separate channels of data-taking for each decay-mode). The plotting parameters are the same as in the first plot in this post except for the scale of the x-axis, which goes all the way from 0 to 250 GeV. Now, between 120 GeV and 130 GeV, there is an excess of events (light blue). Physicists know it is an excess and not at par with expectations because theoretical calculations made after discounting a Higgs-boson-like decay event show that, in that 10 GeV, only around 5.3 events are to be expected, as opposed to the 13 that turned up.