Compositional neural networks for multimodal X-ray and neutron scattering inversion
Abstract
X-ray and neutron scattering are indispensable for structural characterization across condensed-matter physics and materials science. Yet, because scattering experiments generally record intensities without phase information, distinct structures can yield similar scattering patterns, rendering the corresponding inverse problems non-unique. This ambiguity can be reduced by combining complementary measurements, but their integration into machine-learning frameworks remains challenging. Here we introduce a modular framework for probabilistic multi-measurement analysis based on prior-amortized flow matching. Single-observation posterior estimators are trained separately for individual observation types or physical probes and combined only at inference time through a compositional sampler. This late fusion approach allows pretrained single-observation models to be reused without dedicated joint retraining. As a case in point, we consider reflectometry, which provides depth-resolved, nanometer-scale information about thin films and interfaces. Our single-observation demonstrations span X-ray, time-of-flight neutron and polarized neutron reflectometry. In separate co-refinement studies, task-matched estimators enable homogeneous composition across multiple neutron contrasts and heterogeneous multimodal composition of X-ray and neutron measurements. Together, these results establish compositional flow matching as a modular approach to uncertainty-aware inversion and co-refinement, with potential applicability to other scattering problems in which multiple observations constrain a common physical structure while retaining measurement-specific parameters.
Related
- Approaching the Inverse Neutron Scattering Problem with Neural Networks
- Nonuniform iterative phasing framework and sampling requirements for 3D dynamical inversion from coherent surface scattering imaging
- Island of Inversion in neutron-rich cobalt isotopes revealed from mass measurements
- Reduced vortex descriptors linking polycrystallinity in magnetic nanoparticles with polarized magnetic small-angle neutron scattering
- Fabrication and characterization of multi-shell nested fully annular quasi-ellipsoidal focusing mirrors with m=3 supermirrors for small-angle neutron scattering
BibTeX
@misc{munteanu2026compositional,
title = {{Compositional neural networks for multimodal X-ray and neutron scattering inversion}},
author = {Valentin Munteanu and Alexander Hinderhofer and Dmitry Lapkin and Vladimir Starostin and Frank Schreiber},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.29362},
eprint = {2610.29362},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.29362},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Compositional neural networks for multimodal X-ray and neutron scattering inversion
AU - Valentin Munteanu
AU - Alexander Hinderhofer
AU - Dmitry Lapkin
AU - Vladimir Starostin
AU - Frank Schreiber
PY - 2026
DA - 2026/10/21
PB - arXiv
DO - 10.48550/arxiv.2610.29362
UR - https://arxiv.org/abs/2610.29362
AB - X-ray and neutron scattering are indispensable for structural characterization across condensed-matter physics and materials science. Yet, because scattering experiments generally record intensities without phase information, distinct structures can yield similar scattering patterns, rendering the corresponding inverse problems non-unique. This ambiguity can be reduced by combining complementary measurements, but their integration into machine-learning frameworks remains challenging. Here we introduce a modular framework for probabilistic multi-measurement analysis based on prior-amortized flow matching. Single-observation posterior estimators are trained separately for individual observation types or physical probes and combined only at inference time through a compositional sampler. This late fusion approach allows pretrained single-observation models to be reused without dedicated joint retraining. As a case in point, we consider reflectometry, which provides depth-resolved, nanometer-scale information about thin films and interfaces. Our single-observation demonstrations span X-ray, time-of-flight neutron and polarized neutron reflectometry. In separate co-refinement studies, task-matched estimators enable homogeneous composition across multiple neutron contrasts and heterogeneous multimodal composition of X-ray and neutron measurements. Together, these results establish compositional flow matching as a modular approach to uncertainty-aware inversion and co-refinement, with potential applicability to other scattering problems in which multiple observations constrain a common physical structure while retaining measurement-specific parameters.
ER -