Model selection in continuous time

The foundations of a theory of model selection for continuous-time autoregressive systems is outlined. The authors define the predictive stochastic complexity for continuous-time systems and investigate its asymptotic properties. An almost sure asymptotic result is presented.<
>