Neutrino Oscillations at NOVA

Seamus Sweeney Saint Henry District High School

Katheryn Slattery Oak Hills HS

Dr Alex Sousa Mentor

During my QuarkNet internship at UC, I chose to work on the neutrino experiment under
Dr. Sousa. The purpose of my work was to run analysis on neutrino oscillations (namely
identifying the best parameters for the muon neutrino survival function) by comparing simulated
Monte Carlo data to observed Far Detector data. My first task was to develop selection criteria in
order to select signal from the MC and Far Detector data and get rid of as much background as
possible. I decided to make a cut on the variable event length because I had noticed that signal
events tended to be longer than background events; I focused on data from each set with an event
length greater than 15 meters. For the MC data, this yielded an efficiency of 83% and a purity of
93%. Once I had signal data I could use, I experimented with applying the probability function
(which determines the likelihood of a muon neutrino remaining a muon neutrino, rather than
oscillating to another type of neutrino) to the MC data for different parameter values. The
parameters for the probability function are sin ​2​(2θ) and Δm ​2​, where 0.5 ≤ sin ​2​(2θ) ≤ 1.0 and
0.001 ≤ Δm ​2​ ≤ 0.004 at a constant length of 735 km. The probability function is a function of
true neutrino energy (which is different from the reconstructed MC energy I was analyzing), so
in order to apply the probability function to the MC data, I created a 2-D histogram of true
neutrino energy vs. MC reconstructed energy, input true energy to the probability function bin by
bin, and multiplied each bin by the probability yielded. This operation gave me oscillated MC
data (for arbitrary parameter values) that I could compare to Far Detector data. My goal,
however, was to determine the optimal parameters to match the MC data to the Far Detector data
as accurately as possible, so I wrote code that would go over 100 different values each of sin ​2​(2θ)
and Δm ​2​ within the acceptable ranges, which gave 10,000 distinct probability functions, and
therefore 10,000 unique MC data sets to compare to the Far Detector data set.