Discovering a Kink in the System

The phrase “whatever is not forbidden is compulsory” is commonly attributed to physics Nobel laureate Murray Gell-Mann and taken to mean that any imaginable process is possible unless something explicitly prevents it from occuring. Why does a rock thrown upwards not fly into space and hit the moon? Because it doesn’t have sufficient velocity to overcome the Earth’s gravitational potential. Why does an iceberg not sink? Because the density of ice is less than that of water, and buoyant forces keep it afloat.

In particle physics, we ask a parallel question: If neutrinos and charged leptons (which include electrons, muons, and taus) are part of the same lepton “family” in the Standard Model (SM) of particle physics, and neutrinos spontaneously change from one type into another through oscillations, why don’t muons spontaneously become electrons?

I don’t know. In fact, no one does. This is a hypothesized phenomenon known as charged lepton flavor violation (CLFV). It’s an open mystery of particle physics.

When I started a physics major, “SM” and “CLFV” were just combinations of letters to me. My heart wasn’t even set on sticking with physics. What I did know, however, was that I needed a job to fund my bubble tea consumption. I approached my introductory physics professor, Paul Tipton, about getting a job and ended up working for CERN’s ATLAS experiment. And it was only by participating in research, making decisions, and posing my own questions to investigate that I really understood the appeal of the physics path. I have never looked back.

While I do have many stories about ATLAS, this story is about my work in Professor Sarah Demers’ Mu2e group. Mu2e is an upcoming experiment at Fermilab, a particle physics lab in Illinois, designed to look for evidence of CLFV. In particular, we’re looking for the spontaneous conversion of a muon to an electron without the emission of a neutrino (“mu to e” is where the experiment gets its name). This process is not explicitly forbidden in the equations that define our understanding of SM physics. The uncharged partners to muons and electrons—the muon neutrino and the electron
neutrino—go back and forth all the time as they travel through space. You would expect similar behavior from muons, yet it has never been observed.

We do see muon-to-electron conversions that involve a muon neutrino and an electron antineutrino in, for example, atomospheric muons. This “normal” (neutrinoful?) muon-to-electron conversion happens countless times a day, but the Standard Model predicts the conversion without neutrinos to happen 10-50 times less frequently—a level that is experimentally undetectable. Many extensions to the Standard Model predict that the neutrinoless conversion ought to be more frequent, but it has been lurking at sensitivities below what Mu2e’s predecessors could observe. It might be just within reach of modern experiments like Mu2e.

Since the target process is so rare, Mu2e must observe a lot of muon decays, hoping that one of them might be the one. As you might expect, observing every decay and writing it to a hard drive would require an impossible amount of storage space, which is why the experiment employs an online trigger—software designed to look at each event as it’s happening and decide if it might be interesting. This places two key constraints on the trigger:
• The trigger must run really fast
• The trigger must maintain a high efficiency to not miss any of the already-rare target process

When I joined Mu2e, the trigger was working really well most of the time, but certain events would lead to significant delays, and no one knew why. I was tasked with figuring it out.

By then, I had already been working in particle physics research for two summers, so I thought I was reasonably competent. This illusion was shattered when I was immersed in an entirely new ecosystem of code written in C++, which I had not worked in before. With the help of the amazing mentors in my research group, we embarked on the long process of learning how the system worked so I could eventually help make it better.

When I was up to speed, I analyzed each of the individual trigger modules that work together for particle detection to determine which ones were delayed. Once I narrowed down the cause to one, I timed the performance of each step in that module’s code, edging closer and closer to the answer. Ultimately, I realized we could run a step of the pattern recognition algorithm fewer times, since some of the incoming data corresponded to protons—particles the detector was not supposed to be looking for—and could therefore be filtered out by the characteristic energy deposited. This change improved the speed of the entire trigger sequence by 15% at no cost in efficiency, mainly by speeding up many of the events that used to take a long time to process.

In the paragraph above, I described a year of research, which spanned a summer working at Fermilab and my senior thesis. And implementing the change was the easy part. Diagnosing the problem and coming up with a solution was where the time really went. Looking back, I’m tempted to feel a little dumb—if the problem can be summarized so succinctly, why had it taken me a year to solve? But such is the physics experience. We needed to try lots of things that didn’t end up working before we finally ironed out this tiny kink, which is only one of many.

And the analogy is broader. The Standard Model is incredibly successful, but it also has some kinks. Sometimes it feels like, since the discovery of the Higgs boson, we haven’t found any new solutions. But we keep looking. Mu2e might find evidence of CLFV or it might not. But through knowing what doesn’t happen we can zero in on what does. There are only so many places for the answers to keep hiding.