Fitting multiple small-angle scattering datasets simultaneously: on the optimal use of priors and weights

arXiv (Cornell University)(2023)

引用 0|浏览3
暂无评分
摘要
Small-angle X-ray and neutron scattering (SAXS and SANS) are powerful techniques for elucidating the structure of diverse particles and materials. This study address the challenge of effectively combining SAXS and SANS data for accurate structural parameter determination. Surprisingly, our results demonstrate that equally weighting all data points leads to the most accurate parameter estimation, even when SAXS data significantly outnumber SANS data. We compared this approach with weighting schemes normalized by the number of points and by the derived information content. Furthermore, we assessed the impact of prior knowledge by incorporating Gaussian priors for model parameters. Our findings indicate that Gaussian priors improve the accuracy of refined parameter values compared to uniform priors. When using a minimum and a maximum values for model parameters, which is common practice, uniform priors are implicitely applied. Finally, we show that utilizing information content aids in determining the degrees of freedom, enabling accurate calculation of the goodness of fit. In conclusion, this research provides valuable insights into the optimal combination of SAXS and SANS data, emphasizing the importance of weighting schemes and prior knowledge for enhanced accuracy in structural parameter determination.
更多
查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要