Nuclear Material Inference as an Inverse Problem

• Activity Number: 27 - SDNS Speed Session
• Type: Contributed
• Date/Time: Sunday, August 8, 2021 : 1:30 PM to 3:20 PM
• Sponsor: Section on Statistics in Defense and National Security
• Author(s): Emily Casleton* and Paul Mendoza and Dave Osthus and Dan Archer and James Ghawaly and Irakli Garishvili and Andrew Rowe and Ian Stewart and Michael Willis
• Companies: Los Alamos National Laboratory and Los Alamos National Laboratory and Los Alamos National Laboratory and Oak Ridge National Laboratory and Oak Ridge National Laboratory and Oak Ridge National Laboratory and Cadre5 and University of Tennessee and Oak Ridge National Laboratory
• Keywords: nuclear material transfer; inverse problem; emulator; nonproliferation; nuclear material identification

Abstract

Identification of the nuclear material moving around a facility is a particularly challenging problem of high interest to the nuclear non-proliferation community. We will present a new approach to isotope identification by solving a series of inverse problems to perform inference on a separate variable of interest for a project called MINOS: Multi-Informatics for Nuclear Operations Scenarios. Radioactive materials exhibit distinct gamma signatures, but those signatures can be obscured by many confounding variables. First, we select a collection of nuclear materials of interest. For each material, we perform simulations using a transport code for gamma radiation with various values of distance between the sensor and source and scattering coefficient. Next, we build an emulator for each material. Our observed gamma spectra are measured by a network of sodium-iodide detectors. Given the data from the sensors, we solve the inverse problem for each material, estimating the distance and scattering coefficient. We put forward the radioactive material, or set of materials, that produced the most plausible solution to the inverse problem as the most likely material being transferred.