WilsonCowanramverket

WilsonCowanramverket is a modular computational framework designed to simulate Wilson-Cowan neural population models. It provides an abstraction for modeling interactions between excitatory and inhibitory neuronal populations using rate-based dynamics, capturing how activity in one population modulates the other through recurrent connections.

At its core, the framework implements the standard Wilson-Cowan equations: dE/dt = -E + S_E(W_EE E - W_EI I + P_E) and dI/dt = -I + S_I(W_IE E - W_II I + P_I), where E and I denote mean firing rates, W terms are synaptic weights, P denotes external input, and S is a transfer function (often a sigmoid). WilsonCowanramverket extends this with modular components for connectivity patterns, time constants, and optional spatial extensions via kernels, enabling both single-population and neural-field simulations.

Key features include configurable populations for excitatory and inhibitory cells, a range of transfer functions, adjustable weights and time constants, and multiple numerical solvers for time integration. The framework supports parameter sweeps, sensitivity analyses, and visualization tools for time series, phase-space plots, and bifurcation diagrams. For spatially extended models, it offers grid-based lattices or kernel-driven connectivity to study wave propagation and pattern formation.

Typical use involves defining network architecture, setting initial conditions and parameters, running simulations, and analyzing outputs to examine oscillations, stability, and responses to external stimuli. WilsonCowanramverket emphasizes reproducibility and extensibility, facilitating experimentation with alternative transfer functions, coupling schemes, and integration schemes, while remaining faithful to the core Wilson-Cowan dynamics.