Module wilson_cowan 

Batch parity with WilsonCowanUnit.step in src/sc_neurocore/neurons/models/wilson_cowan.py (Wilson & Cowan 1972, Biophys. J. 12:1–24).

Per step: dE/dt = (−E + sigmoid(w_ee · E − w_ei · I + ext)) / τ_e dI/dt = (−I + sigmoid(w_ie · E − w_ii · I)) / τ_i (E, I) advance through one fixed-step RK4 update.

where sigmoid(x) = 1 / (1 + exp(−a·(x − θ))).

The model is deterministic (no noise), so bit-exact parity with the Python primary requires only matching arithmetic — no pre-drawn RNG buffer needed.

Functions

derivatives 🔒

logistic 🔒

sigmoid 🔒

simulate

Simulate ext_input.len() Wilson-Cowan iterations, writing per-step E and I traces into caller-allocated buffers. Returns final (E, I) for convenience.