Radiation source reconstruction based on a covariance matrix adaptation evolution strategy and the adjoint flux and forward source inner product evaluation method

•A radiation source reconstruction algorithm using gamma measurements is proposed.•The algorithm uses an evolution strategy to iterate on possible solutions.•To evaluate a given possible solution, an adjoint flux based method is used.•A test case shows the algorithm works well when scattered photons are included.•It is shown that the algorithm is very sensitive to small detector fluctuations. A novel approach to solve inverse photon transport problems is presented and used to characterize a shielded source based on gamma measurements taken outside the shielding. The iterations on source characteristics are controlled by a covariance matrix volution strategy and the evaluation of each of the trial sources calculates the inner product of the adjoint flux and the trial source, for each detector and relevant energy group. This approach uses a combination of cross sections evaluated for both coarse energy groups and discrete energies. The impact of including information from photon scattering on reconstruction accuracy is studied. Using a 2D Cartesian system, we show that a successful characterization is possible and that the inclusion of scattered photons can significantly improve results, especially for determining the source strength and shielding thickness at the same time. It is further demonstrated that the source reconstruction is sensitive to any discrepancies between the source evaluation deterministic method and the simulated detector measurements used for validation.