Artificial Intelligence and Electrical & Electronics Engineering: AIEEE Open Access
Neural Network Formation via NavierâStokes Dynamics and Spectral Encoding by Riemann Zeta Zeros
Abstract
Chur Chin
We propose a theoretical framework in which neural network formation is modeled as an emergent process governed by Navier–Stokes–type dynamics on a high-dimensional functional manifold. In this framework, neural connectivity patterns arise from the evolution of a continuous flow field representing synaptic density and signal propagation. We further hypothesize that the long-term stable modes of this flow admit a spectral decomposition whose critical frequencies correspond to the non-trivial zeros of the Riemann zeta function. Under this conjecture, neural computation can be interpreted as the selection and interaction of zeta-zero–indexed eigenmodes, providing a novel bridge between fluid dynamics, number theory, and learning systems.