21–26 Jun 2026
University of California, Irvine
US/Pacific timezone

PROfit: A PROfessional, PROficient Framework for Robust Neutrino Parameter Estimation and Optimization

Not scheduled
20m
Conference Center (University of California, Irvine)

Conference Center

University of California, Irvine

Poster Sterile Neutrinos Poster session

Speaker

Brandon Weiss (Columbia)

Description

We present PROfit, a new open-source C++ framework designed for efficient and robust estimation of neutrino oscillation parameters, systematic uncertainties, and beyond-the-Standard-Model (BSM) physics, with particular emphasis on reliable global optimization in high-dimensional parameter spaces and rigorous treatment of systematics. Initially developed for the Short-Baseline Neutrino (SBN) programme, PROfit is seeing extensive use at ICARUS, SBND, MicroBooNE, DUNE, and beyond.

Through its novel combined approach of nuisance-parameter splines and covariance matrices, the construction of robust Feldman–Cousins confidence intervals can be achieved in an extremely timely manner. Key systematics for a given analysis are treated as continuous nuisance-parameter splines, while smaller, Gaussian-like systematics are handled as computationally simpler covariance matrices, enabling efficient treatment of 1000’s of systematics. PROfit streamlines the process of end-to-end sensitivity estimation, fake data studies, brazil-band analysis and frequentist interval construction.

PROfit is designed to be fully user-configurable at run-time for any number of detectors, beam running modes, search channels, sidebands, and physics models. Default physics models include a range of short-baseline sterile neutrino scenarios (such as 3+1, 3+2, and 3+1+decay) and BSM template fits, while inclusion of additional user-defined models has been made straightforward.

This poster will introduce the toolkit, provide detailed usage examples, and discuss CPU-hour scaling behaviour as the number of physics parameters, bins, and systematic complexity are expanded. We will also discuss ongoing efforts to enable direct side-by-side Bayesian and Feldman–Cousins intervals by default, as well as integration into standard HEP workflows and widespread HPC facilities.

Authors

Presentation materials