Monday, September 3, 2012

Higgs: A view from the moment of discovery!

[Note from Shaun: When Mikko wrote us a guest post about the Higgs discovery he also gave me a short note he had written on the day CMS first opened their 2012 box and looked at the Higgs-relevant data. I decided to save that note for a rainy day. Today, is that rainy day (literally, in Helsinki). What follows is more or less exactly what Mikko wrote down the evening that he and about 100 other people first learned that they really had discovered an entirely new fundamental (probably) particle. The rest of us couldn't be there in that room, but we can read about it now!]

**** Do not open before July 9 *****

Recollections of a Higgs discovery

It's not official yet, and will not be for another three weeks, but you could say Higgs was finally discovered today, on Friday, 15th of June, 2012. More than fifty years of searching, and there it is, at 125 GeV, just like the first hints last December indicated.

The big occasion was the unblinding of the 2012 data set at a Higgs meeting held at CERN at 15:00 hours on Friday evening. The meeting venue, the non-descript Building 222 better known as the Filtration Plant, was stacked with CMS physicists, with half of the crowd sitting on the floor or leaning against the back wall. The air was dense from expectation, and immensely hot from the mass of people and failing ventilation.

Everybody was appropriately informed of the formal proceedings of the day: the slides would not be posted on the web, no recordings of the video meeting would be allowed (except an official one by the CMS Outreach Team), and nothing shall be leaked outside the collaboration after the meeting. Only the highest level of CMS management had seen all the results before, at a special preview held at 11am in the morning.

For the uninitiated, I should probably explain what the unblinding is all about. Scientists are intimately aware of unconcious biases in analysis, when the stakes are high and the statistics are low. Therefore, it is considered good practice to not look into the signal region before fixing the analysis procedure and cuts. The expected background in the signal region is estimated using side bands, and the analysis only proceeds to look in the signal region, the "box", when those side bands are found to be sufficiently well understood.

The Higgs group had agreed that nobody would look into the signal region of 2012 data before today (or yesterday evening really, to allow some time for analyzers to prepare their talks). The previous week was spent by review committees scrutinizing each of the analyses and making sure all the systematics were thought of and no obvious mistakes would remain. Only the analyses given official green light would be allowed to open the box, and the whole collaboration was invited to join the event.

A significant fraction of the three thousand collaborators apparently did indeed join, most of them remotely. From the first few minutes it was clear that the video meeting system was creaking and was barely holding the traffic. The outside world could hear the audio, and we could hear some of them (despite frequent reminders to mute), but the video feed was apparently stalling. With no slides posted, the people in the videoland were more or less blind.

All the more reason to feel privileged to be at CERN to listen to the talks in person.

The first three talks were strategically ordered to go from the channel with the worst mass resolution and lowest expected sensitivity to the one with best resolution and expected sensitivity. The HWW (Higgs decaying into two W bosons) analysis got the honor to be the first messenger.

After a bit of a jumpy start with switching lights on and off for better contrast on the video projector, trying to transmit slides outside CERN and accidentally dropping the network connection, the talk finally got up to speed. Several slides showing impressive agreement between data and simulation covered the sidebands before moving on to signal regions, with quite visible excesses. The bottom line: a little more than a three sigma excess with combined 2011(5/fb)+2012(3/fb) data, precisely in agreement with the standard model expectation for a 125 GeV Higgs. Hey, this starts to look quite promising!

After a few more minutes of more and less technical questions from the collaborators we turned to the Hgammagamma (Higgs decaying into two photons) channel. The talk was given by a young Chinese graduate student from MIT, who'd obviously absorbed the American style of putting a bit of drama into the talks. With skill she had the collaboration holding their breath waiting to see the new limit plots... with a gigantic peak and a local excess of more than 4 sigma at 125 GeV when combined with 2011 data.

At that point I had to fight a bit breaking into tears. Those two channels alone meant that we'd have to be above the 5 sigma discovery limit already. It would mean we had discovered the Higgs. After 50 years of searching. Us, here.

Ok, back to sobering up a bit. The signal was much stronger than expected from standard model, which means we had either got very lucky, or that this could be a non-standard-model Higgs. All the better, we might have more to discover later in the year. The measurements from different subcategories of photons pairs and from 2011 and 2012 looked all perfectly consistent so there was no hint of a measurement error.

The last of the big three talks was ZZ4l (Higgs decaying into two Z bosons, which in turn decay into four leptons). This is the ultimate channel with very little background so you could even claim with good probability that some individual events are from a Higgs boson decay, unlike in the background dominated HWW and Hgammagamma channels. The expectations were already high from the two previous talks, and the results certainly did not disappoint. Around half a dozen nicely clustered events right at 125 GeV, just like the standard model predicted.

It's interesting to note that improvements to the analysis, like Particle Flow based lepton isolation and recovery of photons radiated off the Z bosons, had both improved sensitivity and caused the secondary peak seen at 119 GeV in 2011 to disappear. The updated results combined with 2012 statistics made a very convincing case, racking up another 3 sigma or so.

The main trio was followed by a fourth talk on VH (Higgs produced in association with a vector boson, i.e. Z or W), which however had not yet been granted green light to open the box. Nevertheless, the analysts had made nice improvements to the analysis, gaining 50% more sensitivity out of the 2011 data, and showing a small excess consistent with standard model Higgs. A planned fifth talk on Higgs decaying into two tau leptons was postponed pending more checks, as was appropriate. The background checks before opening the box were clearly taken seriously.

Overall it was quite a tour-de-force, with all channels lining up in unison. This is still not all, because the analyses used only the first 3.9/fb of 2012 data collected until June 8, and in most cases even less. With 5.6/fb already in the can today and three more days to go to reach above 6/fb, the analyses will likely have about 50% more integrated luminosity for ICHEP. This might be enough to take some channels already above 5 sigma by themselves.

In the short summary the Higgs conveners reminded everybody that this is really a result by everybody in the collaboration, not just the Higgs group: thousands of people had contributed in building, maintaining and running the detectors, writing reconstruction software, calibrating the detectors, checking the data etc. The final analysis was only the tip of a large iceberg. And the work was not yet over, there was still plenty to do before presenting the results in Melbourne, Australia on July 9.

A final warning was given before people departed the room: smiles should be subdued and no champagne bottles should be popped in the cafeteria; there were filming crews outside that had not been allowed in the meeting room, and they had vowed to film the expressions on the people as they came out. We should not let the world know just yet ;)

At CERN in Geneva, Switzerland
June 15,

1 comment:

  1. If I had been in that room, when leaving, my face would have been a 5-sigma fluctuation.

    I'm a really bad poker player...