Freiburg 2019 – wissenschaftliches Programm

Sitzungen | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

FM: Fall Meeting

FM 65: Poster: Quantum & Information Science

FM 65.4: Poster

Mittwoch, 25. September 2019, 16:30–18:30, Tents

Applications of Neural Networks on Small-Angle X-Ray Scattering Data — •Thomas Stielow, Paula Respondek, and Stefan Scheel — Institut für Physik, Universität Rostock, Albert-Einstein-Straße 23, 18059 Rostock

Modern phase retrieval algorithms allow for a detailed reconstruction of two dimensional electronic densities from small-angle scattering patterns obtained in FEL experiments. However, despite major improvements, such algorithms still suffer from convergence to local minima and perform best if structural information such as the object’s silhouette is provided [1]. Deep learning algorithms have recently been employed in the reconstruction of wide-angle scattering patterns [2]. Here we demonstrate how such a procedure can be adopted to the small-angle regime. In particular, we show that deep learning models can both be used in orientation and density reconstruction. The observed reconstruction results are highly stable due to the possibility of including known propeerties of the observed system into the neural network.

[1] T. Ekeberg et al., PRL 114, 098102 (2015).
[2] T. Stielow et al., arXiv:1906.06883 (2019).