Statistical Physics-based Framework and Bayesian Inference for Model Selection and Uncertainty Quantification

A major challenge with the proliferation of atomically resolved images on material systems is how best to construct models that can explain the observations. One way of approaching this problem is to consider that images of atomic configurations and distortions represent the actual microstates of a thermodynamic system. Probing their fluctuations then provides a means to determine the system's response to thermodynamic perturbations.