AMBER: Algorithm for Multiplexing spectrometer Background Estimation with Rotation-independence

TL;DR

AMBER is a segmentation algorithm that efficiently separates foreground and background signals in neutron spectrometry data, leveraging rotation-independence to reduce expert input and systematic errors.

Contribution

The paper introduces AMBER, a novel, model-agnostic segmentation algorithm that improves background estimation in neutron spectrometry by exploiting rotational independence.

Findings

Reduces need for expert input in data analysis

Enhances full dataset utilization

Minimizes systematic errors

Abstract

State-of-the art neutron spectrometers enable simultaneous measurements of high-dimensional datasets, allowing for a large collection rate of dynamic material properties. In this paper, we present the Algorithm for Multiplexing spectrometer Background Estimation with Rotation-independence (AMBER), which is a segmentation algorithm designed to decompose measured neutron scattering data into model-agnostic foreground and background contributions. The method takes advantage of the fact that background and foreground signals are measured simultaneously during the data collection process, relying on rotational independence of background contributions. The algorithm, initially developed for multiplexing neutron spectrometers, aims to strongly reduce time consuming expert input, therefore promoting full data set usage while minimizing the source of systematic errors.