Table of Contents
Fetching ...

Threshold Dynamics of Voter Radicalization on the Probability Simplex

Alexander Omelchenko

Abstract

We analyse two coupled ODE models of political competition on invariant probability simplices with a conserved electorate. The baseline three-group model tracks left-radical, centrist, and right-radical voter shares. We characterise the unique interior equilibrium by a Perron--Frobenius threshold, establish global asymptotic stability in the symmetric and asymmetric cases, and exclude periodic orbits unconditionally via the Dulac criterion. A structural consequence is that the baseline model cannot produce irreversible centrist decline, history-dependent long-run floors, or multiple attractors. We then extend the model with a disengaged voter compartment and distinguish pure state shocks from permanent structural parameter shifts. The post-shock dynamics are governed by the same spectral threshold: below it the centrist state is globally asymptotically stable; above it every trajectory with a nonzero radical seed converges to the unique radicalised equilibrium. Cumulative sub-threshold structural shifts can cross the threshold and produce staircase dynamics absent from the baseline; the symmetric reduction yields closed-form expressions for the critical shock amplitude and the radicalization window.

Threshold Dynamics of Voter Radicalization on the Probability Simplex

Abstract

We analyse two coupled ODE models of political competition on invariant probability simplices with a conserved electorate. The baseline three-group model tracks left-radical, centrist, and right-radical voter shares. We characterise the unique interior equilibrium by a Perron--Frobenius threshold, establish global asymptotic stability in the symmetric and asymmetric cases, and exclude periodic orbits unconditionally via the Dulac criterion. A structural consequence is that the baseline model cannot produce irreversible centrist decline, history-dependent long-run floors, or multiple attractors. We then extend the model with a disengaged voter compartment and distinguish pure state shocks from permanent structural parameter shifts. The post-shock dynamics are governed by the same spectral threshold: below it the centrist state is globally asymptotically stable; above it every trajectory with a nonzero radical seed converges to the unique radicalised equilibrium. Cumulative sub-threshold structural shifts can cross the threshold and produce staircase dynamics absent from the baseline; the symmetric reduction yields closed-form expressions for the critical shock amplitude and the radicalization window.
Paper Structure (52 sections, 27 theorems, 137 equations, 7 figures, 3 tables)

This paper contains 52 sections, 27 theorems, 137 equations, 7 figures, 3 tables.

Key Result

Proposition 3.1

For any $(L_0,R_0)\in\Sigma_2$ and any positive parameters, the solution $(L(t),R(t))$ of eq:model remains in $\Sigma_2$ for all $t\ge0$.

Figures (7)

  • Figure 1: Time series for three representative scenarios across the two dynamical regimes of Theorem \ref{['thm:global']}. Left ($\beta=0.22<\mu=0.25$): subcritical case. Both radical wings decay; the centrist share approaches 1. Centre ($\beta=0.40>\mu=0.20$): supercritical case, moderate polarisation. All groups persist at $P^*=0.25$, $C^*=0.50$. Right ($\beta=0.75>\mu=0.20$): supercritical case, high polarisation. Strong reactive polarisation shifts the equilibrium to $P^*=0.367$, $C^*=0.267$. Both right-panel cases belong to the same qualitative regime ($\beta>\mu$); they differ only in the degree of polarisation at equilibrium. Dashed lines show analytical predictions; numerical and analytical values agree within tolerance $10^{-8}$.
  • Figure 1: The three shock regimes of the four-group model. Left: Small shock ($\Delta=0.10 < \Delta_c=0.25$): no surge, monotone recovery to $C^\infty=1$. Centre: Large state shock ($\Delta=0.55>\Delta_c$, $\Delta\beta=0$): transient populist surge followed by full recovery. The apolitical pool $A$ (dotted) drives the surge and is depleted on timescale $1/\rho=10$. Right: Structural shock ($\Delta=0.55$, $\Delta\beta=0.15$): $\beta'=0.45>\mu=0.40$; the system converges to a new equilibrium with $C^\infty=\mu/\beta'\approx0.89<1$. The state component $\Delta$ determines the transient; the structural component $\Delta\beta$ alone determines $C^\infty$. Parameters: $\beta_0=0.30$, $\mu=0.40$, $\delta=0.70$, $\rho=0.10$.
  • Figure 2: Phase portraits in the feasible simplex $\Sigma_2$. Stars mark attracting equilibria. All trajectories converge to the same fixed point, confirming global stability. In Regime I the attractor is $E_0=(0,0)$; in Regimes II and III it is the interior equilibrium $E_1=(P^*,P^*)$. No limit cycles are present, consistent with the Dulac criterion (Corollary \ref{['cor:dulac']}).
  • Figure 2: Staircase dynamics under four sequential shocks. Each shock has state component $\Delta_k=0.40$ and structural component $\Delta\beta_k=0.04$. The cumulative structural shift $B_k$ crosses $\mu=0.40$ at shock $k^*=3$ (since $\beta_0+3\times0.04=0.42>\mu$). Before $k^*$: the centrist share recovers (partially) after each shock. After $k^*$: each additional shock permanently lowers the long-run centrist share, producing the staircase pattern. Parameters: $\beta_0=0.30$, $\mu=0.40$, $\delta=0.70$, $\rho=0.10$.
  • Figure 3: Equilibrium radical share $P^*$ (left) and centrist share $C^*$ (right) as functions of $\gamma$ for $\alpha\in\{0.05,0.10,0.15\}$ and $\mu=0.20$. The bifurcation threshold $\gamma^*=\mu-\alpha>0$ is clearly visible as the left endpoint of each curve. The hyperbolic decay $C^*=\mu/(\alpha+\gamma)$ (right panel) approaches but never reaches zero over the plotted parameter range.
  • ...and 2 more figures

Theorems & Definitions (77)

  • Definition 2.1: The baseline model
  • Proposition 3.1: Forward invariance of $\Sigma_2$
  • Proof 1
  • Remark 3.2
  • Definition 3.3: Symmetric case
  • Theorem 3.4: Global dynamics of the symmetric model
  • Proof 2
  • Corollary 3.5: Centrist share at equilibrium
  • Proof 3
  • Definition 3.6: Interior equilibrium
  • ...and 67 more