The Bayesian Inference Engine (BIE): a computational statistical inference framework for deceleration-phase Rayleigh-Taylor instability studies

The Bayesian Inference Enging (BIE) has been utilized for the analysis of radiographic images capturing the dynamic evolution of deceleration-phase R.-T. modes during laser-driven implosions of cylindrical targets at Omega. Within the BIE, an analyst may construct a parameterized physical model representing the object being imaged, produce synthetic data, and optimize model parameters to obtain a maximum a posteriori solution that considers both weighted statistical likelihood and prior information. 2D implosions are modeled so as to infer the growth rate and evolution of cylindrical modes in a driven Al marker layer, comprehensively accounting for blur, alignment and illumination effects to achieve unprecedented accuracy for comparison to hydrodynamic simulation. The BIE also allows uncertainties to be quantified in a rigorous manner through response surface methodologies, establishing sensible error bars and guiding the refinement of experimental techniques.