Real-Time Steering in Neutron Scattering Experiments
Author Information
Author(s): Junqi Yin, Viktor Reshniak, Siyan Liu, Guannan Zhang, Xiaoping Wang, Zhongcan Xiao, Zachary Morgan, Sylwia Pawledzio, Thomas Proffen, Christina Hoffmann, Huibo Cao, Bryan C. Chakoumakos, Yaohua Liu
Primary Institution: Oak Ridge National Laboratory
Hypothesis
Can an integrated edge-to-exascale workflow enhance real-time decision-making in neutron scattering experiments?
Conclusion
The developed framework significantly reduces data processing time and improves experimental accuracy, enabling real-time adjustments during neutron scattering experiments.
Supporting Evidence
- The framework reduces data processing time from hours to minutes.
- It allows for dynamic adjustments to experimental parameters based on real-time data.
- The Temporal Fusion Transformer model provides voxel-level precision in predicting neutron scattering patterns.
- Integration of edge computing and exascale resources enhances experiment efficiency.
- The system achieved a relative error of 2.8% in real-time predictions.
Takeaway
This study shows how computers can help scientists make quick decisions while doing experiments with neutrons, making the process faster and more accurate.
Methodology
The study utilized a transformer-based model for real-time data processing and decision-making in neutron scattering experiments, integrating edge and exascale computing.
Limitations
The real-time experiment steering is still in the proof-of-concept stage and has not been fully deployed in live experimental settings.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website