When dealing withimperfect models and data, it is beneficial to integratethe Bayesian approach with physics-informed learningfor uncertainty quantification, such as Bayesian PINNs(B- PINNs) 107 .