view article
Figure 1
A benchmark for size-distribution inversion of polydisperse spheres. (a) shows the true and inverted size distributions; the truth is composed of two analytical parts, a Gaussian on the left and a Boltzmann on the right, with the radius ranging from 400 to 800 Å and discretized by 500 points. (b) shows the true and fitted intensity curves, with q ranging from 10−3 to 1 Å−1 and discretized by 200 points in logarithmic scale; we add a 20–30% error to the intensity observation (we use 3σ for the error bars in this plot). To obtain the SasView solution, we need to create a user-defined model combining a Gaussian and a Boltzmann distribution, and set their initial peaks close enough to the truth. For Irena (MaxEnt), we need to decrease the observational error to 1% to achieve an accuracy comparable to that of the other three solutions. The metrics are summarized in Table 1[link].

Journal logoJOURNAL OF
APPLIED
CRYSTALLOGRAPHY
ISSN: 1600-5767
Volume 55| Part 4| August 2022| Pages 966-977
https://doi.org/10.1107/S1600576722006379
Follow J. Appl. Cryst.
Sign up for e-alerts
E-alerts
Follow J. Appl. Cryst. on Twitter
Twitter
Follow us on facebook
Facebook
Sign up for RSS feeds
RSS