Application of dynamic mode decomposition to exponential experiment for spatial decay constant determination

Abstract

When measuring the fundamental mode decay constant γ in an exponential experiment, it is difficult to extract the fundamental mode from a measured neutron flux distribution that is contaminated by many higher harmonics. Dynamic mode decomposition (DMD), which is intended for time-varying system analyses, is applied to the eigendecomposition of a three-dimensional flux distribution by viewing a series of two-dimensional flux profiles at horizontal planes as snapshots of a time-varying system. Through DMD, a fundamental mode can be extracted accurately from limited measured data that are largely contaminated by higher harmonics. The DMD yields a biased γ-eigenvalue if the statistical fluctuation of measured data is considered, and the bias is adjusted through a numerical simulation of an exponential experiment. Bootstrapping is employed for resampling the measured data from an exponential experiment. Ultimately, several measurements for an exponential experiment are closely simulated by bootstrap samples from a limited number of measurements.