Order Reductions of "Predictive Dynamical Systems"

It has been recently pointed out that dynamical systems depending on future values of the unknowns may be useful in different areas of knowledge. We explore in this context the extension of the concept of order reduction that has been useful with singular and delay differential equations in electrodynamics and general relativity. We discuss some general properties of order reductions in this new context and explore a method of successive approximations, which among other things is used to check and improve the ``extrapolate prediction'' and ``fixed rate prediction'' methods.