Today, all over America (land of science enthusiasts) fireworks will be set off in remembrance of the many particle collisions that it took to bring our observations of the decay products of the Higgs up above a five sigma confidence interval. If you need a refresher or just haven’t followed all this very closely, you’ll definitely want to watch PhD Comics‘ animated explanation below (you’ll want to see it full screen).
The best print coverage I’ve seen so far today is, unsurprisingly, at Ars Technica. What I kind of love best about all this is that this isn’t just a triumph of theoretical physics or engineering. It’s also a big day for statistics. When I worked in a genetics lab, there were biostatisticians working overtime trying to figure out how to quantify all our assumptions about high thoroughput shRNA screening and decide whether tests were more about identifying plausible candidates for further testing or trying pass fewer targets that were more likely to work. (Remember sensitivity and specificity?).
We’re a long way from the practice stat problems you did in high school where the teacher tells you how big the standard deviations are (the ‘sigma’ in five sigma). Ars does a beautiful job giving you a conceptual sense of how complicated it is for statisticians to come up with a rigorous way of gauging the data:
But the huge number of collisions created its own problems. At times, up to 30 collisions were taking place nearly simultaneously, and the computer systems had to reconstruct which signals came from what collisions and trigger the system to save the data if something looked interesting—all within a fraction of a second. According to the talks, the software triggers were improved, the code reconstructed events faster, and the computing grid was given more sophisticated analysis tools to identify events that could come from a Higgs decay. The net result was today’s announcement (and yesterday’s accidental pre-announcement).
Where do we now stand? There are a lot of ways to look at it. One is basically the probability of finding the Higgs at a specific mass. If we assume the Higgs is 125GeV, we see a signal that’s a specific sigma above background. But there’s no particular reason to assume 125GeV and not, say, 135GeV, and the statistics need to compensate for this (called the “look elsewhere effect”). Then there are multiple channels thanks to the different decay pathways, and two different detectors. So, for the CMS detector, the two-photon channel produces a local Higgs signal that’s 4.5 sigma, but that drops to 2.5 sigma when the look elsewhere effect is considered. It’s only by combining all its channels that CMS reaches a 4.9 sigma, and the data from both detectors had to be combined to be able to declare discovery.
So huzzah for theorists, engineers, and professional heuristic makers! Go off and caper! Jump up and down and celebrate as your atoms, affected by the Higgs field, pull you back to earth. And tonight, enjoy the fireworks all the more after you watch the video below.
(The fourth of July after I read an article on fireworks chemistry in Muse magazine, I had a great time yelling out the names of the elements that produced the colors we’d just seen. Copper! Charcoal! Strontium! Fun times.)