A gradient flow formulation for the stochastic Amari neural field model

Area of Science:

• Computational Neuroscience
• Mathematical Biology
• Dynamical Systems

Background:

• Neural field equations model large-scale brain activity.
• Amari-type models are mean-field approximations of neuronal networks.
• Previous analyses were limited by the lack of a rigorous gradient flow formulation.

Purpose of the Study:

• To establish a rigorous gradient flow formulation for stochastic Amari-type neural field equations.
• To enable the application of advanced gradient flow analysis techniques to these neural models.
• To investigate the well-posedness and regularity properties of the neural field model.

Main Methods:

• Analysis of stochastic Amari-type neural field equations.
• Proof of gradient flow structure in a nonlocal Hilbert space.
• Demonstration of well-posedness and spacetime regularity for correlated noise.
• Investigation of invariant measures and ergodic properties.

Main Results:

• The neural field model is shown to be a gradient flow in a nonlocal Hilbert space.
• The model is proven to be well-posed, with solutions remaining in the space.
• Spacetime regularity results are established for spatially correlated noise.
• Uniqueness of invariant measures and ergodic properties are discussed.

Conclusions:

• The gradient flow formulation provides a powerful new framework for analyzing neural field models.
• This work opens avenues for applying established gradient flow techniques to neuroscience.
• The findings contribute to a deeper mathematical understanding of neural activity dynamics.