Non-Volterra Quadratic Stochastic Operators

• Non-Volterra QSOs are quadratic mappings on a probability simplex that allow offspring probabilities to mix beyond the parental traits, breaking the traditional Volterra constraint.
• Operators employing permutation parameters, convex combinations, and kernel methods demonstrate unique invariant sets, fixed points, and periodic behaviors.
• The use of Lyapunov functions and cycle-sum invariants facilitates analysis of their ergodic convergence and stability, making them pivotal in modeling genetic and statistical dynamics.

A non-Volterra quadratic stochastic operator (QSO) is a quadratic mapping of a finite-dimensional simplex that violates the Volterra heredity constraint, i.e., it allows positive offspring probabilities in coordinates not inherited from the parent types. These operators arise naturally in models of population dynamics, genetics, and statistical mechanics. Their study has revealed a diversity of long-term behaviors distinct from the classical Volterra case, influenced by algebraic properties, permutation structures, and stochastic kernels.

1. Mathematical Definition and Principal Classes

Let $$S^{m-1} = \{x = (x_1,\dots,x_m)\in\mathbb R^m: x_i\ge0,\,\sum_{i=1}^m x_i = 1\}$$, the $(m-1)$-dimensional probability simplex. A quadratic stochastic operator is a map $V: S^{m-1} \to S^{m-1}$ of the form

$$(Vx)_k = \sum_{i,j=1}^m p_{ij,k}\,x_i\,x_j,\qquad k=1,\dots,m,$$

where the coefficients $p_{ij,k}$ satisfy

$$p_{ij,k} \ge 0,\qquad p_{ij,k}=p_{ji,k},\qquad \sum_{k=1}^m p_{ij,k}=1.$$

A QSO is called Volterra if $p_{ij,k}=0$ for all $k\notin\{i,j\}$. Otherwise, it is non-Volterra, a more general class allowing “mixing” among all or subsets of coordinates beyond the parental ones.

Principal non-Volterra examples include:

• Permutation-parameterized operators: where offspring distributions are shuffled by a permutation $\pi$ of the first $m-1$ coordinates (Jamilov, 2022).
• Convex combinations of non-Volterra QSOs, which interpolate between maximal mixing and permutation-driven regimes (Jamilov et al., 24 Jan 2026).
• Centred kernel QSOs: defined on measure spaces, incorporating perturbation kernels $G$ (Bartoszek et al., 2015).
• Low-dimensional explicit families: such as quasi-strictly non-Volterra operators on $S^2$ (Hardin et al., 2018) and $\xi^{(s)}$-QSOs (Mukhamedov et al., 2013).

2. Permutation-Parameterized Non-Volterra Families

A canonical construction is given by the permutation-parameterized family (Jamilov, 2022): $$V_\alpha: S^{m-1}\to S^{m-1},\qquad \begin{cases} x'_k = 2x_m\big(\alpha x_k+(1-\alpha)x_{\pi(k)}\big), &k=1,\dots,m-1, \ x'_m = x_m^2+\left(\sum_{i=1}^{m-1} x_i\right)^2, \end{cases}$$ with $\pi$ a permutation on $\{1,\dots,m-1\}$ and parameter $\alpha\in[0,1]$.

Special regimes include:

• Volterra-type ($\alpha=1$): reduces to a Volterra operator.
• Purely non-Volterra ($\alpha=0$): cycles are driven by $\pi$; no restriction that offspring match either parent.
• Intermediate $\alpha$: convex blends of Volterra and non-Volterra inheritance.

Distinct combinatorial supports and cycle structures in $\pi$ yield nontrivial invariant sets and finite periodicities.

3. Dynamical Behavior, Limit Sets, and Fixed Points

The dynamical typology of non-Volterra QSOs is strongly determined by parameters and structural choices.

• Permutation-parameterized families (Jamilov, 2022):
  • The simplex vertex $e_m$ is always a fixed point.
  • Further fixed points are characterized by $x_m=1/2$, with remaining coordinates constant on each cycle of $\pi$, and sum constraints.
  • For $0<\alpha<1$, every nontrivial trajectory (outside a measure-zero set) converges to a single point in $X$ (a subset of the simplex with cycle-constant coordinates), yielding global regularity.
  • In the pure non-Volterra limit ($\alpha=0$), trajectories fall into an $s$-cycle (where $s$ is the least common multiple of the cycle lengths of $\pi$).
  • For $\alpha=1$ (Volterra-type), trajectories globally contract to a unique fixed point in $X$.
• Convex combination families (Jamilov et al., 24 Jan 2026):
  • For $0\leq \alpha<1$, all orbits globally converge to a unique symmetric fixed point in the interior of the simplex.
  • For $\alpha=1$, the system admits finite periodic limit cycles determined by the permutation structure, with period equal to the l.c.m. of the cycles of $\pi$.
• Quasi-strictly non-Volterra families on $S^2$ (Hardin et al., 2018):
  • A unique interior fixed point exists for generic parameters, with bifurcations producing two-cycles for smaller values of a structural parameter $e$.
  • Stability (attractor, saddle, or repeller) and the appearance of 2-cycles are resolved by inequalities involving $e$ and the coordinates at the fixed point.
• $\xi^{(s)}$-QSOs (Mukhamedov et al., 2013):
  • Out of 20 non-conjugate classes in the $S^2$ setting, all but two are (permuted) non-Volterra.
  • Dynamics often reduce to monotone maps on lower-dimensional faces, showing bifurcations between fixed-point regimes and invariant continua or cycles.

4. Lyapunov Functions, Invariant Sets, and Ergodicity

Sharp control of the asymptotic behavior is afforded by explicit Lyapunov functions and invariant algebraic sets.

• Invariant coordinate faces: If $k\notin$ the support of $\pi$, the face $\{x_k=0\}$ is invariant under $V_\alpha$ (Jamilov, 2022).
• Cycle-sum invariants: For cycles $\tau_i$ of $\pi$, sums $\sum_{k\in S_i}x_k$ are invariant in the fixed-point sets, and level sets of these sums are invariant for $x_m=1/2$.
• Lyapunov functions: For $x_m\geq1/2$, coordinate sums and cycle-sums are non-decreasing along trajectories, implying strict monotonicity and precluding complex (e.g., chaotic) behavior.
• Ergodicity: Regularity and finite limit sets (fixed points or periodic cycles) imply ergodicity, i.e., the Cesàro means

$$\lim_{N\to\infty} \frac{1}{N} \sum_{n=0}^{N-1} V^n(x)$$

converge for all $x\in S^{m-1}$ (Jamilov, 2022).

A key consequence is that in contrast to Volterra QSOs, which may display non-ergodic phenomena such as infinite $\omega$-limit sets or boundary attractors, these classes of non-Volterra QSOs are uniformly ergodic in generic parameter regimes.

5. Explicit Non-Volterra QSOs: Kernel and Finite-Dimensional Examples

• Centred kernel QSOs (Bartoszek et al., 2015):
  • Defined on probability measures: $Q_G(F,F) = F^\cdot * F^\cdot * G$, with $F^\cdot(A) = F(2A)$ and $G$ a perturbation law.
  • Dynamics of the iterates are determined by weighted sums of independent random variables, enabling probabilistic limit theorems.
  • Weak convergence of iterates to distributions $H^\infty$ under finite-variance or power-tail kernel conditions; no universal CLT emerges due to kernel sensitivity.
  • Simulation algorithms are built upon this sum structure, with explicit error-control via truncating weighted sum tails.
• Low-dimensional explicit families:
  • The quasi-strictly non-Volterra $S^2$ operators, and the various $\xi^{(s)}$ classes, provide concrete regimes where the stability, bifurcation, and fixed point/cycle structure can be solved exactly (Hardin et al., 2018, Mukhamedov et al., 2013).
  • Dynamics are often reducible to one-dimensional nonlinear maps, allowing full classifications of attracting points, periodic cycles, and bifurcation phenomena.

6. Comparison to Volterra Operators and Phenomenological Implications

A summary comparison of generic dynamical outcomes:

Operator Type	Generic $\omega$-limit	Ergodicity
Volterra (with cycles)	Infinite set, possible chaos	Non-ergodic
Non-Volterra (permutation mix)	Single fixed point or finite cycle	Ergodic
Centred kernel (measure space)	Weak limit law (often unique)	Weakly mixing

These findings indicate that non-Volterra structure does not polymorphically increase dynamical complexity; indeed, convex mixing and certain purely non-Volterra constructions yield more regular behavior (global attractors or finite periodic orbits) than archetypal Volterra QSOs. This suggests that averaging in non-Volterra operators may generically suppress boundary effects and irregularity (Jamilov, 2022, Jamilov et al., 24 Jan 2026). A plausible implication is that further exploration of the non-Volterra field can yield tractable, ergodic operators with controlled convergence, useful for modeling global mixing or recombination in biological and statistical settings.