Premium
Neural field equations with neuron‐dependent Heaviside‐type activation function and spatial‐dependent delay
Author(s) -
Burlakov Evgenii,
Zhukovskiy Evgeny,
Verkhlyutov Vitaly
Publication year - 2021
Publication title -
mathematical methods in the applied sciences
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.719
H-Index - 65
eISSN - 1099-1476
pISSN - 0170-4214
DOI - 10.1002/mma.6661
Subject(s) - heaviside step function , mathematics , nonlinear system , field (mathematics) , activation function , mathematical analysis , type (biology) , volterra integral equation , artificial neural network , integral equation , pure mathematics , computer science , physics , quantum mechanics , machine learning , ecology , biology
We introduce a neural field equation with a neuron‐dependent Heaviside‐type activation function and spatial‐dependent delay. The basic object of the study is represented by a Volterra–Hammerstein integral equation involving a discontinuous nonlinearity with respect to the state variable that is both time and space dependent. We replace the discontinuous nonlinearity by its multivalued convexification and obtain the corresponding Volterra–Hammerstein integral inclusion. We investigate the solvability of this inclusion using the properties of upper semicontinuous multivalued mappings with convex closed values. Based on these results, we study the solvability of an initial‐prehistory problem for the former neural field equation with the Heaviside‐type activation function. The application of multivalued analysis techniques allowed us to avoid some restrictive assumptions standardly used in the investigations of the solutions to neural field equations involving Heaviside‐type activation functions.