Qjbtiger / Equilibrium-RNN

An optimization-based equilibrium measure describes non-equilibrium steady state dynamics: application to edge of chaos

Code for "An optimization-based equilibrium measure describes non-equilibrium steady state dynamics: application to edge of chaos" (Arxiv: 2401.10009v1). Here, we propose an optimization-based pseudo energy function to characterize recurrent neural network dynamics. Under the replica calculation in both finite and zero temperature cases, we found two kinds of order parameters describing the order-chaos transition. For more detail, please review our article for free.

Requirements

Software

• Python 3.11

• Numpy 1.22.4

• Pytorch 2.1 with Cuda 11.8

Citation

This code is the product of work carried out by the group of PMI lab, Sun Yat-sen University. If the code helps, consider giving us a shout-out in your publications.