It has been over a year since my last post, and since then, I have started developing my own blog. (I was just a guest author here.) It is called Coffeeshop Physics, and it is a relaxed presentation of physics topics that I think are interesting. The name was inspired by my experiences with Cafe Scientifique, a series of coffeeshop presentations about the sciences— I want to replicate that kind of atmosphere online.

Whereas the Everything Seminar was intended for mathematical audiences, Coffeeshop Physics is for general audiences, much like Cafe Sci. Therefore, I don’t assume that the readers know what a derivative is (relevant for yesterday’s article), but you might find it interesting anyway. Many of the topics that I’m writing about are things that I struggled to understand as an undergrad and even a grad student, the intuition behind the mathematical formalism.

For instance, my favorite article so far is about curved surfaces and gravitation. When I studied general relativity, I could push Christoffel symbols around, but I was frustrated by the fact that I couldn’t visualize the problems that we were working on.

I got a better appreciation for curved surfaces by learning to sew, and after making about a dozen little models, the picture came into focus. Here is a photo of a model of space-time at the surface of the earth, in which we can see that a freefall is a shorter path through space-time than just standing on the ground. It doesn’t have Minkowski structure, so it is not quantitatively accurate (it should open up at the top, not the bottom), but it is a picture to keep at the back of one’s mind.

I’ve also turned the spectrum of resonances in electron-positron collisions into a sound, so that we can hear what it sounds like when the collide, and found a nice demonstration of entropy in a story about a leprechaun tying ribbons on trees in a forest. If you enjoyed the What Killed Madame Curie? detective serial that I started on this blog, I am expanding it into a novel, with links on the site.

Cheers,
— Jim

The new result is not a search for a new type of particle or a decay chain that would indicate beyond-the-Standard Model physics; it’s a statistical correlation in the events that would be considered uninteresting backgrounds in most searches.  When you collide protons, most of the time (by many orders of magnitude in rate), they interact via the strong force according to the rules of Quantum ChromoDynamics (QCD).  This has only been demonstrated in certain limits of QCD, because the theory is highly coupled: quarks are attracted to other quarks by exchanging a gluon, in analogy with electromagnetism (replace “charged particles” for “quarks” and “photons” for “gluons”), but gluons are attracted to other gluons by exchanging yet more gluons: the result is a sticky mess.  I would put QCD/the strong force in the same category as turbulent flows of water: we believe that the underlying physics is understood, but exact results are incalculable because of complexity.  The result that has just been announced is a qualitatively new feature in the angular correlations of QCD interactions.

QCD interactions all take place on very short distance scales, about meters, so the products of the reaction all emerge from a single point. A useful way to look at them is in the projective space of azimuthal () and polar () angles relative to the beamline, with the polar angle usually represented as pseudorapidity ().  The angles are illustrated on a typical QCD event in the CMS detector below (the detector has a rough cylindrical symmetry around the beamline; the beamline points into the page in the figure on the right).

QCD interactions produce a lot of particles, as you can see in the event display above (the yellow lines are tracks from charged particles; the pink and green bars are energy deposits in the calorimeters).  We can quantify the correlation between pairs of particles as a function of angular differences and and plot what we see (see the paper, Eq. 1, for an exact definition of the correlation function).  The plot below is the observed pairwise correlation (vertical axis) versus each angular difference (horizontal axes).

The sharp peak at , is expected: a quark or gluon (collectively called “partons“) emerging from the collision must make a collimated “jet” of particles from partons begetting more partons and finally forming the final-state hadrons that we see.  (“Hadron” collectively refers to any particles made of quarks and held together by gluons, such as protons.)  At , there’s a second peak from the fact that partons usually emerge from the collision in pairs.  Momentum conservation forces the two jets to be back-to-back in , so when you select one final-state hadron from one jet and another from the other jet, they’ll be close to .  The fact that they don’t have close to zero— in other words, momentum conservation in the component parallel to the beamline— is a feature of hadron collisions: most of this component of momentum is carried by the proton’s two out of three quarks that didn’t collide (and usually aren’t observed because their deflection angles are too small to reach the detector).

Now here’s the new feature: in events with large numbers of particles () with moderate momenta ( GeV/c), there’s an excess of final-state hadrons correlated with , .

These are particles that aren’t close enough in polar angle to be in the same jet, but they would nearly line up in an azimuthal view of the collision.  That’s a strange feature, and none of the existing simulations of QCD (all approximations to the full theory) reproduce it.

What makes it especially interesting is its passing resemblance to similar correlations seen in heavy ion collisions.  The LHC is currently colliding single protons, but other experiments such as RHIC collide whole nuclei of heavy elements.  The figure on the right shows the same particle-pair correlation function from gold-gold collisions, and you can see a , ridge like the one observed at the LHC.

In heavy ion collisions, this correlation is interpreted as part of the evidence that high-energy, high-density collisions produce a new state of matter called a quark-gluon plasma, in which the nuclei literally melt into a dense, zero-viscosity fluid of constituent quarks and gluons.  The high density is key: a gold-gold collision, in which each of the gold nuclei has 197 protons and neutrons, is like a bloodbath of QCD interactions.  Partons produced in one interaction can’t get away from the partons produced by the other interactions, so they interact many times, like water molecules bumping into each other enough to become a fluid with an equilibrium temperature.  This is what makes it possible to apply thermodynamic terms to the process: concepts such as “fluid” and “temperature” only make sense in the context of a large number of particles undergoing many interactions.

In proton-proton collisions like those at the LHC, the usual picture is that one parton from each proton interacts and the others get away with only a minor deflection (called “spectators”).  It would be surprising if the collision products interact with themselves enough to make any kind of thermodynamic fluid.  The reason this interpretation is conceivable is that the LHC collision energies are the highest ever explored, and multiplicity (number of particles emerging from the collision) rises with collision energy.  Could 7 TeV be high enough to start melting single protons, the way that large nuclei are melted at, say, 0.2 TeV?

The onset of a quark-gluon plasma in proton-proton collisions is not the only interpretation, and in fact, it would be a highly contraversial interpretation.  Remember that the QCD-like simulations that tell us what to expect are only approximations to the full QCD theory; it may be possible to explain this new feature without invoking quark-gluon plasmas.  I’ve heard people talking about specific effects in single-interaction QCD that might account for it.  The experimental paper is careful to describe only the observed effect and leave the interpretation to theorists— I’m only describing the quark-gluon plasma interpretation here to explain why the result could be exciting.  At minimum, if this only amounts to a tuning of the QCD-like simulations, I don’t think anyone expected having to add a qualitatively new feature to the simulations, rather than just tweaking the numerical values of their parameters.  In that sense, it is an unexpected discovery however you look at it.

(Remember that the CMS public page has more details.  It’s also worth noting that the LHC will begin a 7 TeV heavy ion run later this year, which ought to produce quark-gluon plasmas well above the phase transition.)

The plan at first is just to let the beams collide without focusing them, so the luminosity will be low, and the rate at which new particles could be produced would be correspondingly low.  As time goes on, the beams will be focused and the intensity will be raised, which increases the rate of collisions and therefore the probability of seeing new stuff.  This is the beginning of an 18–24 month period of continuous data-taking and open-ended exploration.

Tonight I’ll be following this from the Fermilab control room (the LHC is in Switzerland— this is a remote control room).  I’ll post any interesting updates as comments to this article (they won’t come up in RSS feeds).  Here are other sources of information, all more direct than this blog (I mostly try to avoid repeating them):

• CERN twitter (from the LHC control room)
• ATLAS control room blog
• CMS e-commentary
• LHC page 1: live update of machine status.  When I last looked, the energy was 3.5 TeV and the beam intensities were 1.5e10 (higher than the past few weeks).  The red and blue lines are intensities of the clockwise and counter-clockwise beams versus time.
• CMS data aquisition: live update of CMS data collection.  The main plot is data accumulated versus time; it’s a constant slope for cosmic rays (no LHC beam), but could jump up if we get a lot of events from the beams.
• CMS event display: pictures of the events as we see them (in three projections: face-on, side view, and 3D).  Yellow lines are particle trajectories, red and blue bars are calorimeter energy deposits.  If a yellow line goes beyond the calorimeters, it’s a muon!  Right now, I think it’s re-playing events from the low-energy collisions of 2009; that will change sometime tonight.

In my timezone, the sun is setting. Happy Passover!

The last entry is for yesterday’s CDMS paper, which shows two candidate events surviving all analysis cuts, set prior to looking at the result (unblinding).  The probability for background fluctuating up to account for these two events is 20-23%, so no one is calling it a signal.  Both are close to the edges of the analysis cuts, so even if the observed events had significantly exceeded the background estimates, there would be room for doubt.  This may be the tip of the iceberg for direct dark matter detection, but then again, it may not.

Also, in case you haven’t heard, there have been a lot of rumors that the Cryogenic Dark Matter Search (CDMS) has discovered something interesting.  They’ll be presenting whatever it is tomorrow with a paper on the arXiv, a Fermilab presentation at 4:00 PM Central U.S. (webcast here), and a SLAC presentation at the same time, 2:00 PM Pacific (webcast here).  It might be the direct detection of dark matter particles, which would be incredibly exciting.

In the past month of LHC running, we’ve seen evidence or hints of the following particles:

* The two with asterisks require qualification: see below.

The muon is an easy one: as soon as the tracking detectors were turned on, they saw muons raining down from cosmic rays. CMS collected hundreds of millions of muons in a month-long campaign in 2008, the basis of 23 detector-commissioning papers submitted to JINST (a personal point for me, since I edited one of those papers).  Muons originating from proton collisions are more rare, but were observed.

The neutral pion (π⁰) was seen in the first 900 GeV LHC collisions this November.  Most of the charged particles produced in proton collisions are also pions (π⁺ and π^–), and the tracking detectors saw plenty of tracks originating from the collision point as well.  But the first LHC run required the experiments’ magnetic fields to be turned off to avoid complicating the orbits of the proton beams, and this meant that all of the tracks from charged collision products were straight lines, providing little information about their momenta.  The energy of the two photons (γγ) from neutral pion decays (π⁰→ γγ), measured by calorimeters, gives us a handle on the mass of the parent particle, and therefore confirm it definitively as a π⁰.

The December run was conducted with full magnetic fields, allowing for some precision tracking.  Two absolutely beautiful resonance peaks came out of that: K_S → π⁺π^– and Λ⁰→ π⁺p^–/Λ⁰→ π^–p⁺.  (These are the ones that I know have been approved by the collaboration so far: there’s a nice article on them in the CMS Times.)

Much like the π⁰ peak, the distributions above are the mass of the particle from which the pair of charged pions (top) or proton-pion pair (bottom) were assumed to originate.  The calculation is pretty simple: in special relativity, the relationship between mass (), energy (), and momentum is

so

.

The distributions above are histograms of , calculated for pairs of observed particles (1 and 2).  More charged pion pairs have an invariant mass of 0.497 GeV than would be expected from random combinations, so we see a K-short peak (K_S) on top of a low, flat background.  Similarly, pairs of protons and π^–, and of antiprotons and π⁺, pile up at 1.116 GeV, the neutral Lambda (Λ⁰) mass.  Red lines are fits to the distributions and the blue lines are the masses measured from experiments before the LHC.

When I flew home from CERN yesterday, I couldn’t resist and brought a reduced sample from the dataset with me on the plane.  Poking around, finding vertices where pairs of charged particle tracks intersect and calculating their masses, I saw our two friends K_S and Λ⁰ and tried looking for more.  It reminded me of why I became an experimental physicist: these things really are there!  The guy next to me on the plane asked if I was programming, and I had to say, “not exactly,” because even though it looks like computer work, it’s reaching beyond the computer to something physical, if not tangible, that was happening inside a beryllium beampipe in France.  The beam quality was better in some runs than others, and you could see that in the backgrounds.

Quarks and gluons (collectively called “partons”) have a weird history in that they were considered computational devices before the physics community begrudgingly, then whole-heartedly, considered them real particles.  The three “colors” of quarks and three anticolors of antiquarks were a physicist’s mneumonic for the algebra of the Lie group SU(3), with the 8 two-colored gluons being the group generators.  The problem with their interpretation as particles was that single quarks and single gluons were never seen in isolation, a phenomenon today known as confinement: a single quark can’t get away from other quarks without creating more quarks in the processes, and so a high-energy quark or gluon fleeing the proton collision “hadronizes” into a pack of hadronic particles.  It’s important to therefore be able to identify groups of particles originating from the same quark or gluon, called jets.  Here’s a nice candidate for a two-jet event:

The wireframe cylinder shows where the tracking detector is, with the yellow lines being tracks of charged particles from the collision.  On top of that, red and blue bars show where the calorimeters (surrounding the tracking detector) registered energy.  The tracks and calorimeter energy are clustered into two apparent jets, indicated by the yellow cones.  This is as much of a quark or gluon as nature will ever allow us to see.

On Monday, the LHC gave the experiments a few hours of record-breaking 2.36 TeV collisions.  At high collision energies, the production rate of more massive particles increases.  One intriguing event from this run contains not just one muon, but two.  Moreover, the invariant mass of this pair is 3.03 GeV, consistent with J/ψ→μμ, where the J/ψ mass is 3.097 GeV.  This event alone is not a “J/ψ observation” because other processes yield muon pairs— imagine one of the invariant mass plots above with a single event in it.  That event has the right mass to be in the peak of the distribution, though.

This display shows three views of the event, including the muon detector measurements that identify the two long, red tracks as muons.  Of all the stable charged particles that originate in proton collisions, only muons pass through enough steel to reach the muon detectors.  Thus, seeing anything at all in these detectors, matched to a track in the central detector, is a pretty clean muon identification.

Some of the (older) professors I worked with in grad school told stories about the November Revolution, the 1974 discovery of the J/ψ that changed particle physics overnight.  Up to that point, all of the major ideas of the Standard Model had been expressed in one form or another, but had not jelled into the single picture we know today.  One of these ideas was that the strange-flavored quark should have a charm-flavored counterpart— a patch on the quark theory to avoid neutral flavor-changing decays through Z bosons that were not observed (the GIM mechanism).  The dramatic J/ψ resonance discovered months later (thousands of events with little background) could only be explained as a charm-anticharm bound state, which lent a lot of credibility to the quark model for making such a prediction, and made W and Z bosons concievable, as long as there’s also a Higgs boson to generate their masses— one by one, the pieces of the Standard Model fell into place.  According to James Bjorken, the whole theory was complete by 1976, though people tell me that they weren’t convinced until the early 80’s when W, Z, and gluon jets were observed.  It turned the field from a collection of puzzling observations into a Theory of Almost Everything, and a search for hints of physics beyond the Standard Model.

Hopefully, we’ll get back into the business of puzzling observations soon enough.

The LHC has officially become the world’s highest-energy collider, by colliding protons at 2.36 TeV (above the Fermilab Tevatron’s record of 1.96 TeV),

I misunderstood a point in the press release that wasn’t heavily stressed.  The LHC has become the world’s highest-energy accelerator, reaching counter-rotating energies of 1.18 TeV each, but the beams were not collided at this high energy yet.  Last week, they were collided at low energy, and this week, they have been accelerated to high energy, but not collided.  The two beams passed by each other in the interaction region, held apart by electric fields.  A few protons on the fuzzy outer edge of the distribution might have collided, but the big collisions are yet to come.

Small steps, yet very fast from one step to the next.

• the first π⁰ particles have been reconstructed from their decay products, shown at the public LHC week 1 conference (by CMS and LHCb);
• the LHC has officially become the world’s highest-energy collider, by colliding protons at 2.36 TeV (above the Fermilab Tevatron’s record of 1.96 TeV);
• the first paper based on LHC data has been submitted to the arXiv (by ALICE).

The π⁰ observation represents the first step in “rediscovering the Standard Model” as part of the detector commissioning.  It’s like a walk through history, where this step is at about 1950, when the π⁰ was first discovered in cyclotrons and cosmic rays.

The above plot shows invariant mass distributions of pairs of photons observed in CMS and LHCb.  From every pair of photons, you assume that they came from the decay of a particle and plot what the mass of that particle must have been.  For many pairs, the assumption is false, so you get a combinatoric background of random photons, but for photons that actually came from π⁰ → γγ, you get a peak at the π⁰ mass.  The combined distribution is a peak on a smooth background.  Most particles in the Standard Model are known only through their decay products, and this is the first example to be seen at the LHC.

Since we already know a lot about π⁰s, we now use them to calibrate the photon detectors.

Needless to say, these algorithms are not just being developed now— they’ve been in the works for years.  That explains how ALICE was able to put together and internally approve a paper based on the first collisions in one week.  For a brand new analysis, that would take many months at least!

See the CMS e-commentary for hourly updates and more information.  (That’s how I know which results are public. :))

The yellow boxes are silicon strips that detected the passage of particles (most likely pions in this case) and the green lines radiating from the center are tracks reconstructed from those hits.  They’re not constrained to meet at the center: that’s an indication that these particles actually originated where the beams collide.  Beyond that, the red and blue bars show how much energy was collected in the electromagnetic calorimeter (electrons, photons, and hadrons) and the hadronic calorimeter (hadrons only), respectively.  No activity can be seen in the muon detector (red boxes).

This is all consistent with what one should expect from the collision of two protons— a strong (QCD) interaction between the quarks and gluons producing a handful of strongly-interacting hadrons, rather than photons, electrons, muons, or taus, which are insensitive to the strong force.  An electroweak interaction between the quarks and gluons, producing possible Higgs bosons or any of a number of other exciting possibilities * * * * * * * *…,  are more rare, and will require collecting and sifting through huge numbers of collisions.

The CERN twitter site says that all four experiments saw collision-like events.  It’s finally happening!

Update: now it’s 500 times around the ring (about 0.05 seconds).  Last year’s record was about 9 minutes of continuous beam.

Update: up to 9 seconds, 30 seconds (50k events seen by CMS)…

Update: and now a beam in the other direction has made a full orbit.  (All you gotta do is smack ’em together!)

Update (Nov 21): on Saturday, we got hours of stable beam (single beams, not colliding).  This has never been done before with the LHC: from here on, it’s all new territory.  Now I’ve got to get to work on the offline data, which should be great for detector alignment…

“Beam-splashes” are when a beam is threaded part-way through the LHC ring, then deliberately collided with an absorbing block of tungsten to stop it, upstream of a detector. Many particles are created in this collision, most of them are absorbed, with the exception of the muons and neutrinos. CMS can detect muons, and what it sees is a huge splash of activity, shown in this event display from September, 2008.

The blue bars indicate huge deposits of energy in the calorimeters. They seem to project from the center of the detector, but this is an artifact of the software, which was designed to visualize collisions from the center. The calorimeter cells measure energy, not direction, so when it sees energy coming from a flood of particles arriving from the right, it draws them as though they came from the center.

You can also see little parallel lines surrounding the central burst like a school of fish. These are individual muons seen by the barrel muon detectors, which do measure direction.

Update: here it is, the first CMS beam-splash of 2009 (from the e-commentary page)!

The little red lines are reconstructed muon tracks, blue dots are raw hits, and the yellow/blue starburst in the center is the calorimeter energy.  You can tell that the beam is coming from the right-hand side of the detector (“LHC beam-1”, the clockwise direction around the ring).