Or maybe it should be called the ICHEP Crunch. We are one week out now. And from an experimentalist at-a-large-collider-experiment’s point-of-view, ICHEP is almost settled. The almost-final versions of all the plots are prepared, the supporting text sanitized of jargon is being fine-tuned, and the large collaborations are getting their last review in of the results. Heck, even some early results are staring to trickle out (CDF has released a two mass Higgs searchs: WH and ZH associated production (no excess!!) and a search for a standard model higgs decaying to taus (find all CDF results here), and DZERO has a search for a t’ quark (find all DZERO results here).
It feels a little bit like the ‘night before Christmas. It is the calm that is so spooky… Which is in direct contrast, of course, to what is going on outside the experiments in the papers and other blogs…
I’m fortunate enough to be on two of the experiments, ATLAS and DZERO, and on each experiment I’m participating in two sides of the shuffle. On DZERO I have a student working on getting one of the results ready. I bet that the “ICHEP deadline” was first talked about almost 6 months ago. That is about the time that the pressure started building for the folks doing the real work in the experiments. Ever since then each time someone wants to add something new, or spend the time to better understand and reduce a systematic error, they have to ask themselves “does this mean not releasing it for ICHEP?”*
In ATLAS I help run one of the groups producing results. This is management, not physics I’m doing there. The goal for people doing this task is to make sure that all the results are high enough quality to enter the review process. Sometimes this means convincing people to drop a plot, or add an extra plot. By the time someone like me gets heavily involved in the day-to-day of the analyses being prepared we are getting very close to the internal review process.
Back on DZERO I’m also helping out with one of the reviews of an analysis. Once the analyzers have put the finishing touches on their analysis, and the people running their physics group give the ok, the analysis is handed off to an internal review group. These people are meant to be independent reviewers of the analysis, and are supposed to go over it with a fine-toothed comb. Unlike external reviewers, they have access to all the internal information of the experiment. The review is based on an “analysis note” and supporting documentation. Just the note can be 100’s of pages long in case of a complex analysis (Like the Higgs searches from CDF I mentioned above). If no problems are found this review probably takes about a week. But a fresh set of eyes always turns up new problems. It is very intense time for the reviewers and the analyzers: the reviewers send questions, and the analyzers respond as fast as they can so that the review doesn’t get stuck and miss ICHEP. BTW, if you watch the experiments you’ll often see one or two or three analyses come out just after ICHEP – these are the ones where an issue was raised and couldn’t be addressed in time to make it for the conference.
Finally, with that done, the ICHEP shuffle enters its last phase: collaboration review. The now perfected analysis along with documentation suitable for everyone to read (which may be a journal paper draft) is thrown up on an internal web page and a message is broadcast to the complete collaboration (all 600 for DZERO or 3000 for ATLAS): “Our experiment is going to release this work – this week is the last chance to raise issues before it is made public!”
At the same time this review is on going everyone is putting together their ICHEP talks, running practice talks to make sure they are high quality, and answering questions that come in from the collaboration. A popular analysis can get 100’s of comments, for example (an unpopular one might get just a few). Each comment must be reviewed and answered by the analyzers and the answers cross-checked by the internal review. At the same time public web pages are being prepared with high quality versions of all the plots and links to the supporting text and documents.
And then ICHEP starts!
So… you might ask… what defines the length of the ICHEP Shuffle? Which is about 6 months? At the two ends of the year there are two large sets of conferences. ICHEP this year is in the summer, and in the winter is the Morriond series. I’ll bet you good money that I’ll see emails with the title “Morriond Analysis Planning” hitting my Inbox the week after ICHEP is over.
But, for us in these large collider experiments, ICHEP is the breather between the dances. A chance to relax, look around, see what everyone else is doing, get some new ideas, and maybe even explore Pairs!
Thanks for the insight! It's fascinating. A probably dumb question from an IT engineer in the industry: if confidentiality and priority in presenting the results are so important in this field, how can you manage to be on two competing experiments? If DZERO discovers the Higgs, will your DZERO-brain-side let your ATLAS-brain-side know well in advance or it will keep the secret and let maybe the ATLAS part look in the wrong direction? ;)
ReplyDeleteExcellent question. Several things help me out. First, DZERO and ATLAS aren't going to compete for something like the Higgs for many years. It will take the LHC a while to accumulate enough data to be a threat. So I will avoid exactly that conflict of interest.
ReplyDeleteCDF and DZERO, the two experiments at the Tevatron, specifically do not allow you to be a member of the both experiments at the same time (people switch all the time when they change jobs). I think ATLAS and CMS, the two big competitors at the LHC, have the same rule - and it is for this reason you mention.
Eventually, however, it will become a problem - especially if the Tevatron runs for a few more years. Then the secrets will be based on the honor system - which it is mostly how it works now. The repercusions for a leak are mainly a good talking-to.
It might seem like secrecy is a priority in this field right now. But that is because of the mess caused by this rumor that T gave life to. So right now you are getting a biased view. The problem is that this stuff gets into the news, so now any mistake has a huge visibility. It really puts everyone in a tough spot - there is a reason these results aren't released yet: they aren't done yet! Normally, however, the press spotlight isn't on us, and we talk to each other all the time. Normally we don't share plots or data, but you also know exactly what the important things are not to talk about and what you can talk about without much worry. Also, some people do actually keep rumors they hear to themselves. :-)
Thanks! The honor system isn't something that would work as easily in the 'real world', but I guess in your field the feeling of a 'higher common goal' and the thirst for pure Knowledge make personal ambitions more controllable.
ReplyDeleteOr... do they? ;)
Thanks again, I'll definitely keep following.
But doesn't it? I mean - you don't go talk to your competitor. Your company doesn't do anything to prevent you doing it - you just know there will be some consequences - your peers will think you are dirty, perhaps you get a repremand, or in an extreme case (or you work for Apple) you get fired.
ReplyDeleteIt is pretty much the same system here.
And there is no way things are kept tight and zipped up. Rumors and gossip make their way around (obviously!).