Struct AdaptiveThresholdMoENeuron 

pub struct AdaptiveThresholdMoENeuron {
    pub v: f64,
    pub v_th: f64,
    pub k: f64,
    mean_abs_x: f64,
    ema_alpha: f64,
}

Adaptive threshold spiking neuron matching the SpikingBrain architecture.

Converts activations into integer spike counts via data-dependent threshold, enabling addition-based event-driven computation with ~69% sparsity.

Exact equations from arXiv:2509.05276v2 (SpikingBrain Technical Report):

V_th(x) = (1/k) · mean(|x|) (adaptive threshold) v[t+1] = v[t] - V_th · s[t] + x[t+1] (membrane with soft reset) s_INT = round(v_T / V_th) (integer spike count)

In time-collapsed mode: v_T = x, s_INT = round(x / V_th). Parameter k controls the firing rate / sparsity trade-off.

Reference: SpikingBrain-1.0, arXiv:2509.05276v2, September 2025.

Fields

v: f64

Membrane potential.

v_th: f64

Current adaptive threshold.

k: f64

Firing rate control parameter (higher k → lower threshold → more spikes).

mean_abs_x: f64

Running EMA of |input| for threshold computation.

ema_alpha: f64

EMA decay for mean estimation.

Implementations

impl AdaptiveThresholdMoENeuron

pub fn new() -> Self

pub fn with_k(k: f64) -> Self

pub fn step(&mut self, current: f64) -> i32

Returns integer spike count (0 or more) — not binary.

Implements: V_th = (1/k)·mean(|x|), s = round(v/V_th), soft reset v -= V_th·s.

pub fn step_collapsed(&mut self, activation: f64) -> i32

Time-collapsed single-step mode: s_INT = round(x / V_th).

pub fn sparsity(&self) -> f64

Current activation sparsity estimate (1 if below threshold, 0 if firing).

pub fn reset(&mut self)

Trait Implementations

impl Clone for AdaptiveThresholdMoENeuron

fn clone(&self) -> AdaptiveThresholdMoENeuron

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for AdaptiveThresholdMoENeuron

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for AdaptiveThresholdMoENeuron

fn default() -> Self

Returns the “default value” for a type.