**Links to**: [[Concept]], [[Attractor]], [[Narrative]], [[Basin of attraction]], [[Cannalization]], [[Gravity]], [[Representation]], [[Autosemeiosis]], [[Model]] *This entry is unfinished.* In the mathematics (i.e., the measuring and formalizing) of random dynamical systems, an attractor is the set of values toward which a system advances. An attracting _set_ are the points in the system’s phase space, which all preceding states tend to approach in the linear observation of spatiotemporal evolution. We know that random dynamical systems = random (at least: _what we call random_). So, subtle details in initial conditions can lead to unexpected outcomes. The pullback attractor can^[Like the [[Path integral]] for particles?] reveal the upper and lower bounds of possibility in the projection of futures for (autonomous) random dynamical systems (in the case of active inference: based on Bayesian induction, hence the divergence often mentioned by Friston). An attractor remains a demarcated entity because it is the smallest set which cannot be further subdivided into more attracting sets. This restriction is necessary since a dynamical system may have multiple attractors. **Modelling attracting sets** A graph, is like a table, is a set of relations, and nothing more. Mapping those relations and generalizing them is modelling. Modelling is like setting up a film plan: the idea is there, but what happens is in the filming (and its reception). ### Footnotes