Complete Mean-Field Theory for Dynamics of Binary Recurrent Networks

Area of Science:

• Statistical physics
• Network science
• Theoretical neuroscience

Background:

• Macroscopic dynamics of recurrent interactions in binary units are complex.
• Existing mean-field theories often fail for systems with finite size and interactions.
• Understanding nonequilibrium fluctuations is crucial for complex systems.

Purpose of the Study:

• To develop a unified mathematical theory for recurrent interactions in arbitrary network architectures.
• To provide a complete description of nonequilibrium fluctuations in finite-sized networks.
• To investigate systems where deterministic mean-field theories are inadequate.

Main Methods:

• Utilizing martingale theory for mathematical analysis.
• Developing a unified theoretical framework for recurrent binary unit networks.
• Analyzing networks with finite size and finite interaction degrees.

Main Results:

• A complete description of nonequilibrium fluctuations in finite networks was achieved.
• The theory successfully addresses systems where mean-field approaches fail.
• A novel dynamic state was uncovered in statistically inhomogeneous networks.

Conclusions:

• The developed theory offers a comprehensive framework for analyzing complex recurrent networks.
• The novel dynamic state demonstrates collective nontrivial stochastic fluctuations in the thermodynamical limit.
• This approach advances the understanding of nonequilibrium phenomena in network science.