Some Two-Steps Discrete-Time Anticipatory Models With ‘Boiling’ Multivaluedness
Alexander S. Makarenko · AIP conference proceedings · 2006
This paper describes and investigates a class of models and the concept which can make a universal methodological background for difficult social, economic and public systems concerning different spatial and time scales and hierarchical levels. These are nonlinear models of difficult processes with foresight expectations. In the review some existing models with foresight expectations is presented, and the new nonlinear model with the behavior similar to models of neural network is proposed. This model has further differences from the existing one. At first the model is anticipatory, that is passing on two steps ahead, and secondly the nonlinearity in the equations has a piecewise‐linear character, and looks like the activation function of neurons. The condition for multivaludness had been found. Such multivaluedness is a special type of ‘boiling tank’ when the multiplicity had created at the restricted region of space. The suggested concept and principles allow the development of some practical applications of models.