Fabian Unruh

Future nuclear disarmament treaties will likely include the verification of complete and correct declarations of items related to nuclear weapons programs. Nuclear archaeology is a field of research aiming at reconstructing the operational history of facilities producing fissile material which is an integral component of nuclear weapons. The methods of nuclear archaeology can provide estimates for past fissile material production making them a useful asset for assessing the completeness of fissile material declarations.
This research project, located within the Research Group Science for Nuclear Diplomacy of the Cluster for Natural and Technical Science Arms Control Research (CNTR), intends to advance existing techniques by including novel data sources as well as sophisticated statistical data analysis tools. It is planned to improve forensic measurement analysis, which is an important tool in nuclear archaeology, by systematically focusing on the most important information as well as complementing the measurements with data from archives documenting the historical operation of nuclear reactors. A particular focus of the project lays on implementing machine learning techniques, including recently established methods in the field of deep learning, for proper analysis of large data sets and reliable statistical statements. Finally, the integration of statistical results in a political verification regime is addressed.