Estimating Attributes of Nuclear Weapon and Other Fissile Material Configuration Using Features Of Nuclear Materials Identification Signatures

This brief describes a strategy that, when implemented, will allow the attributes, i.e., the physical properties, of nuclear weapon and other configurations of fissile material to be estimated from Nuclear Material Identification System (NMIS) signatures for arms control, treaty verification, and transparency purposes. Attributes are estimated by condensing measured NMIS signatures into ''features'' that approximately represent physical characteristics of the measurement such as gamma-ray transmission, induced fission, etc. The features are obtained from NMIS signatures to estimate quantities related to gamma and neutron transmission through the inspected item and gamma and neutron scattering and production via induced fission within the inspected item. Multivariate, i.e., multiple-feature, linear models have been successfully employed to estimate attributes, and multivariate nonlinear models are currently under investigation. Attributes estimated employing this strategy can then be examined to test the supposition that the inspected item is in fact a nuclear weapon