Daily Summary for 2026-11-13

daily

Transcript

Introduction to the show: ident: High Energy Physics: Experiment. The day in high-energy experiment: detectors, searches and limits.

Kai: It's the thirteenth of November, twenty twenty-six, and this is the day's research.

Mira: 2 new papers came out today.

Kai: I'm Kai.

Mira: And I'm Mira, here with Lev.

Kai: We'll take the day in one pass, then pull out the papers we're staying with.

Mira: Start us off with Silent coverage failures in rare-event searches and a degeneracy index that predicts them.

The day's research: Kai: We are using machine learning to search for nucleon decay, which offers insights beyond the Standard Model.

Mira: The focus is on the 16O to 13Cpi+pi+e+ decay mode in Super-Kamiokande.

Lev: This mode has theoretical predictions suggesting a partial lifetime near one thousand years.

Kai: The study sets a new limit of four point two times ten to the power of thirty-two years for this search.

Mira: A convolutional neural network with an attention mechanism was employed for event classification in Super-Kamiokande data.

Lev: This model projects the data into a three dimensional space using a Mercator projection.

Kai: It was trained on Monte Carlo simulations including Fermi motion and NEUT cascade models for pion interactions.

Mira: The analysis showed two distinct peaks separating the signal from the background in the network output.

Lev: This allowed determining an optimal cut value across five operational periods for consistent signal selection efficiency.

Kai: The method achieved a lower limit on the trinucleon decay lifetime greater than four point two times ten to the power of thirty-two years.

Mira: The paper introduces a Poisson–Fisher degeneracy index called IPF equals beta gamma.

Lev: This index helps diagnose potential coverage failures by measuring signed signal bias and residual mismatch.

Kai: It is useful for screening model errors before toy calibration, though its use varies by model deformation type.

Mira: The paper Search for the 16 O to 13 C pi+ pi+ e+ Decay Mode in Super-Kamiokande Using Machine Learning Techniques focuses on trinucleon decay mode searches.

Lev: Silent coverage failures in rare-event searches and a degeneracy index that predicts them is a paper introducing an index for signal coverage issues.

Lucky paper: 2608.09203: Kai: So we're looking at Silent coverage failures in rare-event searches and a degeneracy index that predicts them today.

Mira: This paper by Davide Pagano introduces the Poisson–Fisher degeneracy index, IPF equals beta gamma.

Lev: That index pairs the signed standardized signal bias with the residual left outside the complete fitted tangent space, which is called beta.

Kai: And gamma is described as the Poisson–Fisher norm of that residual after all fitted directions have been profiled.

Mira: So beta measures how much a model error can bias the fitted signal, and gamma shows how much mismatch there is in the goodness-of-fit test.

Lev: The authors define a controlled model misspecification as a deformation of the nominal expected-count vector nu star equals nu zero plus delta nu, where s is the severity from zero to one.

Kai: They also specifically study deformations like an unmodelled background tail scaled exponentially with severity s, and they test shifted dark-matter halos and overestimated signal efficiency.

Mira: The index is used in a five-step procedure to guide model checks and uncertainty treatment before toy calibration using only expected bin counts.

Lev: The sign of beta indicates whether a positive bias threatens the lower endpoint, which means false rejection of mu equals zero, while negative bias can push the upper endpoint under.

Kai: So if you see a large beta and small gamma, that plausible deformation might evade diagnosis and needs to be treated with auxiliary information or included in a defensible envelope.

Mira: The paper compares various inference methods including Feldman–Cousins, CLs, profile-likelihood, and Bayesian intervals for counting studies.

Lev: For spectral studies, they use profile likelihood, LF2I, and split-conformal posterior scores to estimate coverage.

Kai: Coverage estimation is done by estimating effective coverage from nominal or misspecified datasets as the fraction of regions containing theta.

More episodes

← Home