Quantum Diaries

Now that summer is fully here, are you feeling that old wanderlust, the desire to hit the open road? Well then, there are a lot of interesting places to go on the physics conference circuit between now and Labor Day. There are many fabulous locations on the menu, and who knows, you might get to hear the first public presentation of an exciting new physics result. While it’s true that what many would consider the most glamorous stuff from the LHC has already been pushed out (at the highest priority), you can be assured that scientists are hard at work on new results, and of course there are many other particle-physics experiments that are doing important work. So, find your frequent-flyer card and make sure you’ve changed the oil, and let’s see where you might be headed this summer:

2013 Meeting of the American Physical Society Division of Particles and Fields, Santa Cruz, CA, August 13-17. Like the EPS conference, DPF also meets in odd-numbered years and is a chance for the US particle physics community to gather. It’s one of my favorite conferences, with a broad program of particle physics and neither too big or too small. It is especially friendly to younger people presenting their own work. Measurements that weren’t ready for the earlier conferences could still get a good audience here. Yes, you might have gone to nearby San Francisco in June, but Santa Cruz has a totally different feel, and you can study the hydrodynamics that power the redwood trees that are all over the campus.

And you might ask, where am I going this summer? I’d love to get to all of these, but I have another destination this summer — I will be moving my family to Geneva for a sabbatical year at CERN in July. It’s a little disappointing to be missing some of the action in the US, but I’m looking forward to an exciting year. I will be returning to the US for the Snowmass workshop, where I’m co-leading a working group, but that’s about it for conferences for me this summer. That will still be plenty exciting, and I’ll do my best to report all the news about it here.

Note to readers: this is my best attempt to describe some issues in accelerator operations; I welcome comments from people more expert than me if you think I don’t have things quite right.

The operators of the Large Hadron Collider seek to collide as many protons as possible. The experimenters who study these collisions seek to observe as many proton collisions as possible. Everyone can agree on the goal of maximizing the number of collisions that can be used to make discoveries. But where the accelerator physicists and particle physicists might part ways over just how those collisions might best be delivered.

Let’s remember that the proton beams that circulate in the LHC are not a continuous current like you might imagine running through your electric appliances. Instead, the beam is bunched — about 10¹¹ protons are gathered in a formation that is about as long as a sewing needle, and each proton beam is made up of 1380 such bunches. As the bunches travel around the LHC ring, they are separated by 50 nanoseconds in time. This bunching is necessary for the operation of the experiments — it ensures that collisions occur only at certain spots along the ring (where the detectors are) and the experiments can know exactly when the collisions are occurring and synchronize the response of the detector to that time. Note that because there are so many protons in each beam, there can be multiple collisions each time two bunches pass by each other. At the end of the last LHC run, there were typically 30 collisions that occurred per bunch crossing.

There are several ways to maximize the number of collisions that occur. Increasing the number of protons in each bunch crossing will certainly increase the number of collisions. Or, one could imagine increasing the total number of bunches per beam, and thus the number of bunch crossings. The collision rate increases like the square of the number of particles per bunch, but only linearly with the number of bunches. On the face of it, then, it would make more sense to add more particles to each bunch rather than to increase the number of bunches if one wanted to maximize the total number of collisions.

But the issue is slightly more subtle than that. The more collisions that occur per beam crossing, the harder the collisions are to interpret. With 30 collisions happening at the same time, one must contend with hundreds, if not thousands, of charged particle tracks that cross each other and are harder to reconstruct, which means more computing time to process the event. With more stuff going on each event, the most important parts of the event are increasingly obscured by everything else that is going on, degrading the energy and momentum resolution that are needed to help identify the decay products of particles like the Higgs boson. So from the perspective of an experimenter at the LHC, one wants to maximize the number of collisions while having as few collisions per bunch crossing as possible, to keep the interpretation of each bunch crossing simple. This argument favors increasing the number of bunches, even if this might ultimately mean having fewer total collisions than could be obtained by increasing the number of protons per bunch. It’s not very useful to record collisions that you can’t interpret because the events are just too busy.

This is the dilemma that the LHC and the experiments will face as we get ready to run in 2015. In the current jargon, the question is whether to run with 50 ns between collisions, as we did in 2010-12, or 25 ns between collisions. For the reasons given above, the experiments generally prefer to run with a 25 ns spacing. At peak collision rates, the number of collisions per crossing is expected to be about 25, a number that we know we can handle on the basis of previous experience. In contrast, the LHC operators generally to prefer the 50 ns spacing, for a variety of operational reasons, including being able to focus the beams better. The total number of collisions delivered per year could be about twice as large with 50 ns spacing…but with many more collisions per bunch crossing, perhaps by a factor of three. This is possibly more than the experiments could handle, and it could well be necessary to limit the peak beam intensities, and thus the total number of collisions, to allow the experiment to operate.

So how will the LHC operate in 2015 — at 25 ns or 50 ns spacing? One factor in this is that the machine has only done test runs at 25 ns spacing, to understand what issues might be faced. The LHC operators will re-commission the machine with 50 ns spacing, with the intention of switching to 25 ns spacing later, as soon as a couple of months later if all goes well. But then imagine that 50 ns running works very well outset. Would the collision pileup issues motivate the LHC to change the bunch spacing? Or would the machine operators just like to keep going with a machine that is operating well?

In ancient history I worked on the CDF experiment at the Tevatron, which was preparing to start running again in 2001 after some major reconfigurations. It was anticipated that the Tevatron was going to start out with a 396 ns bunch spacing and then eventually switch over to 132 ns, just like we’re imagining for the LHC in 2015. We designed all of the experiment’s electronics to be able to function in either mode. But in the end, 132 ns running never happened; increases in collision rates were achieved by increasing beam currents. This was less of an issue at the Tevatron, as the overall collision rate was much smaller, but the detectors still ended up operating with numbers of collisions per bunch crossing much larger than they were designed for.

In light of that, I find myself asking — will the LHC ever operate in 25 ns mode? What do you think? If anyone would like to make an informal wager (as much as is permitted by law) on the matter, let me know. We’ll pay out at the start of the next long shutdown at the end of 2017.

I must apologize for being a bad blogger; it has been too long since I have found the time to write. Sometimes it is hard to understand where the time goes, but I know that I have been busy with helping to get results out for the ski conferences, preparing for various reviews (of both my department and the US CMS operations program), and of course the usual day-to-day activities like teaching.

The LHC has been shut down for about two months now, but that really hasn’t made anyone less busy. It is true that we don’t have to run the detector now, but the CMS operations crew is now busy taking it apart for various refurbishing and maintenance tasks. There is a detailed schedule for what needs to be done in the next two years, and it has to be observed pretty carefully; there is a lot of coordination required to make sure that the necessary parts of the detector are accessible as needed, and of course to make sure that everyone is working in a safe environment (always our top priority).

A lot of my effort on CMS goes into computing, and over in that sector things in many ways aren’t all that different from how they were during the run. We still have to keep the computing facilities operating all the time. Data analysis continues, and we continue to set records for the level of activity from physicists who are preparing measurements and searches for new phenomena. We are also in the midst of a major reprocessing of all the data that we recorded during 2012, making use of our best knowledge of the detector and how it responds to particle collisions. This started shortly after the LHC run finished, and will probably take another couple of months.

There is also some data that we are processing for the very first time. Knowing that we had a two-year shutdown ahead of us, we recorded extra events last year that we didn’t have the computing capacity to process in real time, but could save for later analysis during the shutdown. This ended up essentially doubling the number of events we recorded during the last few months of 2012, which gives us a lot to do. Fortunately, we caught a break on this — our friends at the San Diego Supercomputer Center offered us some time on their facility. We had to scramble a bit to figure out how to include it into the CMS computing system, but now things are happily churning away with 5000 processors in use.

The shutdown also gives us a chance to make relatively invasive changes to how we organize the computing without potentially disrupting critical operations. Our big goal during this period is to make all of the computing facilities more flexible and generic. For the past few years, particular tasks have often been bound to particular facilities, in particular those that host large tape archives. But that can lead to inefficiencies; you don’t want to let computers remain idle at one site just while another site is backed up because it has particular features that are in demand. For instance, since we are reprocessing all of the data events from 2012, we also need to reprocess all of the simulated events, so that they match the real data. This has typically been done at the Tier-1 centers, where the simulated events are archived on tape. But recently we have shifted this work to the Tier-2 centers; the input datasets are still at the Tier 1′s, but we read them over the Internet using the “Any Data, Anytime, Anywhere” technology that I’ve discussed before. That lets us use the Tier 2′s effectively when they might have been otherwise idle.

Indeed, we’re trying to figure out how to use any available computing resource out there effectively. Some of these resources may only be available to us on an opportunistic basis, and taken away from us quickly when they are needed by their owner, on the timescale of perhaps a few minutes. This is different from our usual paradigm, in which we assume that we will be able to compute for many hours at a time. Making use of short-lived resources requires figuring out how to break up our computing work into smaller chunks that can be easily cleaned up when we have to evacuate a site.

But computing resources include both processors and disks, and we’re trying to find ways to use our disk space more efficiently too. This problem is a bit harder — with a processor, when a computing job is done with it, the processor is freed up for someone else to use, but with disk space, someone needs to actively go and delete files that aren’t being used anymore. And people are paranoid about cleaning up their files, in fear of deleting something they might need at an arbitrary time in the future! We’re going to be trying to convince people that many files on disk aren’t getting accessed, and it’s in our interest to automatically clean them up to make room for data that is of greater interest, with the understanding that the deleted data can be restored if necessary.

In short, there is a lot to do in computing before the LHC starts running again in 24 months, especially if you consider that we really want to have it done in 12 months, so that we have time to fully commission new systems and let people get used to them. Just like the detector, the computing has to be ready to make discoveries on the first day of the run!

Just a quick note here on something that isn’t really mine: CMS has now released some video footage from the internal meetings where the results of the search for a Higgs particle decaying to a pair of photons (light) were first presented to the collaboration. I think it’s pretty interesting to watch this little bit of science history.

Here is the context: the Higgs searches were done “blind”, in that every little bit of the analysis was done without examining the actual detector data sample where the Higgs might be observed. Using simulations and data samples that are similar to, but not actually, the data in question are used to design the Higgs search, to understand what the experimental uncertainties are, and to give an expectation of what would be seen in the data if there were a Higgs (or not). Everyone avoids looking at the key data samples to avoid biasing the search based on what is seen. (The fear is that if you see a small signal, you might start to make changes in the analysis to enhance the signal…which could turn out to be a statistical fluctuation in the end.)

Then, in the late stages of the data analysis, we finally take a look at the “signal” sample. Of course, someone has to be the first person to do this, which means that for a brief moment, that’s the only person in the world who knows a new scientific fact. And then there is a lot of suspense for everyone else! In the video, someone who was among the first to see the result, just hours beforehand, is presenting it to the rest of the collaboration. You can certainly see how excited the presenters are about that moment.

I suppose there isn’t any suspense now, since we know what the answer is, but try to put yourself in the mindset of the audience in the room…and enjoy!

On Monday at 6 AM CERN time, the LHC ended its collisions of protons for 2012, and in fact until 2015, when the “long shutdown” for energy upgrades is completed. [There will be heavy-ion collisions in early 2013, but the details are beyond my expertise.] Here’s what appeared on the LHC status screen:

It’s the end of an era for the LHC, the end of what we might someday call “Run 1″, the period of first beams at (relatively) low energies, when we got our first glimpse of a new energy scale, and gathered just enough data to see the first glimmerings of (maybe) the Higgs boson. Considering how long we waited for the start of Run 1 — nearly twenty years from the first concepts for the the LHC and its detectors — it is rather amazing that we’re now at the end of the run, after a mere three years.

Still, it’s been a great three years. Here is the plot that captures the whole story:

This is the integrated luminosity recorded by CMS, essentially the number of collisions that the experiment observed, year by year. Remember all the excitement of the first data in 2010? That turned out to be a tiny amount of data compared to what we have recorded since; while we made very good use of it at the time, it was just hinting at the future success of the LHC. And even after the great advances in 2011, by the start of June 2012 we had recorded more data this year than we had in all of last year. Once again, all the experimenters thank the LHC team for the excellent performance of this still new machine.

The last few days of the proton run were spent looking towards the future. Since there won’t be any more proton collisions for two years, it’s important to do some tests that can guide our thinking about how to operate the LHC in 2015. So far, the LHC has run with “50 ns bunch spacing”; that is, the minimum time between bunch crossing is 50 nanoseconds. (Remember, the LHC beam is not continuous, but “bunched”, with a large number of protons close together in the beam, followed by a 50-foot gap before the next bunch in the beam.) This week, the LHC experimented with 25 ns bunch spacing, and even allowed the experiments to take a little bit of data in this mode on Saturday night and Sunday morning. Obviously, with the shorter bunch spacing, you can have beam collisions happening twice as often, and that means that you could potentially achieve the same total number of collisions with fewer protons per bunch. That’s good for the experiments, as each event that we record will have fewer collisions in it, making it easier for us to reconstruct what went on. With 25 ns spacing, we’d probably need less computing capacity and calibration and the like would be easier. But from the accelerator perspective, it is easier to operate the LHC with 50 ns spacing, and the machine operators can’t guarantee that they could provide as much integrated luminosity at 25 ns spacing as the could with 50 ns. Thus, it was important to take some time to understand how to operate the LHC this way. It’s ultimately up to the LHC managers to decide what the best mode for operations is. From the experiment side, it would be easier for us to have 25 ns spacing, but we wouldn’t want to do that at the cost of less data, and perhaps missing a chance of a discovery as a results.

Meanwhile, what does a 3000-member collaboration do with itself when there is no data to record? (Besides sending and reading email.) Quite a lot. First, there are a number of upgrades, repairs and improvements to be made on the detector in the next two years. There is a carefully choreographed dance to be performed in the collision hall, where the CMS detector must be opened up for access to the different components, and the schedule for all the work to be done could be pretty tight. There are also preparations to be made for how we analyze the data in 2015. The environment will be a lot like in 2010: we’ll be at a new beam energy, and in a physics environment that we’ve never seen before, so we’ll have to be ready for anything that might appear in the data. And we will continue our studies of the fabulous three years of data already recorded. During the past three years, the collaborations have released multiple papers on particular topics, with increases in the amounts of data analyzed each time and improvements in analysis techniques. But the next round of papers will use the full dataset, and there won’t be any “next” papers. The analysis techniques then must be the best possible; there won’t be another shot for improvements, as the next word will be the final word, at least until 2015. This too will take a lot of effort from the scientists.

Congratulations to everyone on a successful Run 1, and let’s look ahead to a busy shutdown and an exciting Run 2 beyond!

I’m just back from a trip to CERN, which was mostly for a week of meetings about how well the computing for the CMS experiment is doing and how it could be done better. But meanwhile, the collaboration was also working through the scrutiny of new measurements that are targeted to be released for the Hadron Collider Physics conference that starts tomorrow (I guess today, given the time zone) in Tokyo. Obviously I can’t discuss these results yet. So instead I’ll spend a little time on philosophy, which I admit makes this a much less interesting post. But bear with me.

In case you’ve been hiding under a rock for the last week, you should know that Nate Silver of the FiveThirtyEight blog made another successful prediction of a presidential election outcome, state by state. I’ve written about Nate Silver’s work here before, because I admire his adoption of what I think is a particle-physics kind of approach to making predictions.

So, being in a Nate Silver kind of mood as I headed off on my trip, I bought a e-copy of his new book, “The Signal and the Noise,” to read on the plane. I’m not sure that I’d call it a great work of literature, but Silver does have some very interesting things to say about how to make predictions. In one section he reminds us of the classic Isaiah Berlin essay, “The Hedgehog and the Fox.” And in case you have been hiding under a different rock, that refers to a quote from Archilochus, who observed that “the fox knows many things, but the hedgehog knows one big thing.” Hedgehogs view the world through the prism of a single big conceptual framework, while foxes don’t believe that’s possible and are willing to be more flexible in their approaches. Silver asserts that it’s the foxes of the world who make better predictions. You have to be willing to try many different approaches and integrate many different tactics to make a good prediction, and, perhaps most importantly, to be prepared to adapt to new information and to change your ideas when your current framework isn’t working,

This got me thinking: are particle physicists foxes or hedgehogs? I would say some of both. Our hedgehog-ness is in our belief in physical law. That’s a big idea that is unavoidable. It is true that our knowledge of physics is always subject to revision in the face of new information, but we believe that in circumstances that have already been well-explored through experiment, physical laws hold without question. Certainly the much-revered standard model of particle physics is taken as a given in regimes where it has been thoroughly tested. And at the very least, we believe in physical law as a big, consistent framework at least as an ideal, if not something that we can truly realize.

But in terms of our approaches to experiment, we have to be foxes. I can say this about some of the results that will be shown at the HCP conference — these are hard measurements, and to get them done, we’ve had to use every trick in the book. A huge variety of techniques have been brought to bear to wring every last bit of useful information out of the data, and it has taken a gargantuan effort from a large team of people. I always come out of the detailed presentations of a measurement somewhat stunned by its complexity. We’re also always on the lookout for new and better tricks. No matter how good an idea sounds on paper, if it isn’t effective in making a measurement, or is less effective than other ideas, then you abandon it and find something better. It’s this flexibility and willingness to evolve and change that helps us do this work.

As a younger person, I saw myself as at least an aspiring hedgehog, hoping to find the one big idea that would pull everything together and give me a complete grasp of the world. But I’ve come to realize that life, and science, is more complicated than that, and you have to be a fox just to get through it all.

It’s Labor Day weekend. You’ve probably been on vacation sometime in the past month. So have a lot of physicists (and a lot of bloggers). But the LHC hasn’t been on vacation, by any means. As we last saw, the accelerator was running flat out to produce enough collisions to give us a shot at a Higgs-boson observation. Now we know that it worked; the Higgs results that the CMS and ATLAS experiments released had made use of data that had been collected through the middle of June, and that was enough to claim a discovery. But here’s what’s happened since:

(This plot is a direct link from the LHC luminosity page; it will get updated to the latest version if you read this post sometime later.) Since a technical stop in late June, the LHC has roughly doubled the number of collisions provided. In fact, about a week ago the LHC achieved its highest collision rate ever, which should allow us to accelerate the production of integrated luminosity. We have three months left in the proton-proton run (before a one-month heavy-ion run, and then a two-year shutdown), and we can be optimistic about how much more data we might record in that time.

What are the implications of doubling the size of the dataset? Here is a totally unofficial, totally back-of-the-envelope estimate. In a perfect world, uncertainties due to counting statistics, i.e. the amount of data you are using to make a measurement, fall like the square root of the number of events. Thus, a doubling of the data should result in reducing your uncertainties by a factor of the square root of two, or about 1.4. In our imperfect world, it’s a lot more complicated than that — the uncertainty on a measurement comes from more than just counting statistics. But let us suspend disbelief for a moment and consider the case of the measurement of the probabilities for the presumed Higgs boson to decay into its various final states. Here are the CMS measurements, based on the data that was in hand for the publications that have been submitted:

A value of one on the x axis would constitute a measurement in agreement with the predictions of the standard model. With the uncertainties shown here, you would say that at the very least the measurements do not disagree with the predictions. But now imagine those same error bars reduced by 40%, while the central values stay in the same place. Then the probability to decay to photons would start to look discrepantly large, while that for taus would look discrepantly small, and that would start getting really interesting — perhaps the “Higgs” we’re observing is not the standard-model Higgs.

Of course, this is pure speculation — when the additional data is incorporated into the analysis, all the points might start moving closer to the expected values. But the additional data we are recording will help make the picture more clear no matter what. 2012 has already been an exhilarating year, but as we head into the final one-third of it, we can imagine that it will get more exciting still, thanks to the excellent operation of the Large Hadron Collider.

landmark (n): 1. An object or feature of a landscape or town that enables someone to establish their location. 2. An event, discovery, or change marking an important stage or turning point in something.

Several times in his ICHEP 2012 closing presentation today, CERN Director General Rolf Heuer referred to ICHEP 2012 as a “landmark” conference. This inspired me to take a look at the dictionary definition of “landmark.”

Indeed, the conference seems to fulfill both definitions of the word. First, we have very much defined the location of the field of particle physics. The new particle observed this week, which at the moment appears to have the properties of the Higgs boson, now establishes that the standard model of electroweak symmetry breaking, the main working theory of particle physics for forty years, is correct. Many speakers at the conference who were not explicitly talking about the Higgs still needed to talk about the model, and they all noted with some relief that what they wrote down about the Higgs potential and so forth is now known to be true. (Implicit in all this is the usual scientific caveat that “true” only means “not yet shown to be false,” but please bear with me.) Of course we have long assumed and hoped that it was true, given how much supporting evidence there was for the model all along. But now we know our location: we have a theory that works.

(I should note that while all the attention lately has been on the Higgs, there has been another major advance this year, and that is the measurement of the neutrino mixing angle θ₁₃ at the Daya Bay nuclear reactor experiment in China, with confirmation in other experiments. This establishes that all elements of the neutrino-mixing matrix are non-zero, allowing for the possibility of CP violation in that system. So our colleagues in the neutrino sector also have a landmark measurement to show for themselves.)

And most definitely this observation, and the conference where we first learned about it, is also a landmark in the second sense. We now turn our attention to what this discovery means. What are the next set of questions to ask, and how do we go about asking them? Our first job is to do our very best to characterize the “Higgs boson.” Once its mass is established, the standard model fully specifies all of its other properties. Does it decay to the right final states as often as it is supposed to? Various extensions to the standard model predict different values for these “branching ratios.” Does it have the right spin? We know that the spin must be an even integer, but we don’t know if it is zero as predicted. The LHC will continue taking data into February of next year, before a two-year long shutdown. We’re going to need every ounce of data we can get, and all the cleverness we can muster, to answer these questions as accurately as we can. It will keep us busy during the data drought that is about to come.

I have been spending a lot of time this week thinking about what comes next, especially in terms of future facilities. The LHC is going to be the workhorse of Higgs physics for some time, but it will have its limits. Alas, there are some bad-news landmarks from this conference too: so far, there is no evidence at all for any other new particles beyond the Higgs at this mass scale — no supersymmetry, no exotic fermions or bosons, nothing. We can’t exclude the possibility that even the 13 TeV LHC that will start to run at the end of 2014 might not have enough energy to produce any other new particles. Heuer talked a little about the “high-energy LHC,” which he also noted is really a completely new machine in the same LHC tunnel, but it’s hard to imagine that this happens before 2030. Can we do something else in the meantime? Are there revolutionary, game-changing ideas that we can bring to the table? And, as a writer from the United States, I have to ask — can we build this machine at home?

Big questions for what we hope is a big new era of particle physics! This ICHEP, the first in the Southern Hemisphere, will truly be remembered as a landmark. I would like to take this opportunity to thank our hosts for all of their work in hosting this remarkable conference in an equally remarkable location.