Bayesian identifiability of loss strength and barrier width in complex absorbing barrier tunneling under sparse observation protocols

Abstract Effective loss in quantum tunneling is often inferred from attenuated transmission, but similar attenuation can also result from changes in barrier geometry. We examine this ambiguity using a stationary one dimensional scattering model with a complex absorbing square barrier. With fixed incident energy and real barrier height, the absorptive loss strength γ and barrier width d are inferred from sparse probability density observations, with or without the transmission proxy T = |t| 2 . Bayesian grid posteriors show that scale normalized density observations can produce a negatively correlated ridge in the (γ, d) plane, where weaker loss is partly compensated by a wider barrier and stronger loss by a narrower barrier. Sensitivity and Fisher information analyses suggest that this ambiguity is associated with the near collinearity of the sparse forward responses to γ and d. When the incident amplitude scale is retained, the posterior spread and correlation are reduced, indicating that the inferred loss and geometry parameters depend strongly on the observation protocol. Entropy based sensor screening further suggests that incident region interference may contain useful information for separating the two effects. These results indicate that the apparent separation between loss and geometry in absorbing barrier tunneling is not determined by the forward model alone, but also by the information content of the observation protocol.