Bayesian inference for the potential in PT-symmetric optical systems

This paper introduces a Bayesian statistical framework for quantifying uncertainty in the complex potential of PT-symmetric optical systems. We capitalize on parity and continuity to formulate an informative prior density function, thereby enhancing the statistical robustness of Bayesian inference. In addition, we propose the Modified-Landweber-Metropolis-Hastings method, integrating the modified Landweber iterative method with the Metropolis-Hastings method to calculate the posterior distribution. Extensive numerical experiments reveal that the Modified-Landweber-Metropolis-Hastings method achieves higher precision over other modern sampling techniques. This contributes to a powerful tool for the selection of nonlinear optical materials.