Spiking LIF (SLIF)
An event-driven spiking neural network model for the Diffrax backend, enabling differentiable simulation with JAX. Adapted from snnax.
python
from livn.models.slif import SLIF
model = SLIF()Dynamics
Spike intensity follows:
intensity(v) = exp(β · min(v - v_th, 10)) / τ_s| Parameter | Value |
|---|---|
| β | 5 |
| v_th | 1 |
| v_reset | 1.2 |
| τ_s | 1 |
| α | 3×10⁻² |
Features
- Event-driven spike detection with Marcus lift for precise spike time resolution
- Optional diffusion (Brownian motion)
- Interleaved spike tracking for temporal resolution
- Output: spike times, spike neuron indices, state trajectories