Saved in:
Bibliographic Details
Main Author: Larsen, Andreas Haahr
Format: Preprint
Published: 2023
Subjects:
Online Access:https://arxiv.org/abs/2311.06408
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910973391011840
author Larsen, Andreas Haahr
author_facet Larsen, Andreas Haahr
contents Small-angle X-ray and neutron scattering (SAXS and SANS) are powerful techniques in material science and soft matter. In this study, it was addressed how multiple SAXS or SANS datasets are best weighted when doing simultaneous fitting. Three weighting schemes were tested: (1) equal weighting of all datapoints, (2) equal weighting of each dataset through normalization with the number of datapoints, (3) weighting proportional to the information content. The weighing schemes were assessed by model refinement against synthetic data under numerous conditions. The first weighting scheme led to the most accurate parameter estimation, especially when one dataset substantially outnumbered the other(s). Furthermore, it was demonstrated that inclusion of Gaussian priors significantly improved the accuracy of the refined parameters, as compared to common practice, where each parameter is constrained uniformly within an allowed interval.
format Preprint
id arxiv_https___arxiv_org_abs_2311_06408
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Optimal weights and priors in simultaneous fitting of multiple small-angle scattering datasets
Larsen, Andreas Haahr
Soft Condensed Matter
Biological Physics
Small-angle X-ray and neutron scattering (SAXS and SANS) are powerful techniques in material science and soft matter. In this study, it was addressed how multiple SAXS or SANS datasets are best weighted when doing simultaneous fitting. Three weighting schemes were tested: (1) equal weighting of all datapoints, (2) equal weighting of each dataset through normalization with the number of datapoints, (3) weighting proportional to the information content. The weighing schemes were assessed by model refinement against synthetic data under numerous conditions. The first weighting scheme led to the most accurate parameter estimation, especially when one dataset substantially outnumbered the other(s). Furthermore, it was demonstrated that inclusion of Gaussian priors significantly improved the accuracy of the refined parameters, as compared to common practice, where each parameter is constrained uniformly within an allowed interval.
title Optimal weights and priors in simultaneous fitting of multiple small-angle scattering datasets
topic Soft Condensed Matter
Biological Physics
url https://arxiv.org/abs/2311.06408