Bootstrapping the statistical uncertainties of NN scattering data

TL;DR

This paper evaluates the Monte Carlo bootstrap method for estimating uncertainties in nucleon-nucleon scattering data and compares it with the traditional covariance matrix approach, finding consistent results and proposing alternative error propagation strategies.

Contribution

It demonstrates that the bootstrap method yields comparable uncertainty estimates to covariance matrices for NN scattering data, offering a viable alternative for nuclear physics applications.

Findings

Bootstrap and covariance methods produce similar scattering observables.

No significant differences in phase shifts between methods.

Supports alternative error propagation strategies in nuclear force calculations.

Abstract

We use the Monte Carlo bootstrap as a method to simulate pp and np scattering data below pion production threshold from an initial set of over 6700 experimental mutually $3\sigma$ consistent data. We compare the results of the bootstrap, with 1020 statistically generated samples of the full database, with the standard covariance matrix method of error propagation. No significant differences in scattering observables and phase shifts are found. This suggests alternative strategies for propagating errors of nuclear forces in nuclear structure calculations.