Stochastic Physics
Wherever we have errors in a model (initialization errors, sub-grid parameterization errors, or model structural errors), it may be possible to inject an element of noise and randomness, a 'stochastic' approach, and obtain additional outcomes and computational accuracy. For example, adding stochastic perturbations to a model may ensure it does not remain in one particular regime.