Age-Structured Jump Model

• Age-structured jump models combine continuous aging with stochastic reset events to capture dynamic changes in populations and systems.
• They utilize renewal equations and piecewise-deterministic Markov processes to analyze age distributions, trait selection, and spatial subdiffusion.
• The models provide insights into stable age profiles, ergodicity, and scaling limits, with applications spanning biology, neuroscience, and anomalous transport.

An age-structured jump model is a mathematical framework for describing the evolution of populations or systems in which individuals (or particles) possess an age variable that increases deterministically in time, punctuated by stochastic jumps or resets due to events such as birth, death, division, mutation, or aging. These models integrate the dynamics of both the continuous progression of age and discrete stochastic “jump” dynamics, resulting in renewal-type equations or piecewise-deterministic Markov processes. This class of models forms the foundational structure in population biology, subdiffusive transport, neuroscience, cell division, and other fields, allowing rigorous analysis of time-evolving age distributions, trait selection, spatial subdiffusion, and population ergodicity.

1. Core Mathematical Structure

The prototypical deterministic age-structured jump model describes the density $n(t,a)$ at time $t$ of entities (cells, particles, individuals) with age $a$ using the partial differential equation

$$\partial_t n(t,a) + \partial_a n(t,a) + \beta(a) n(t,a) = 0, \quad t>0,a>0$$

with a nonlocal boundary (renewal) condition at $a=0$

$n(t,0) = \int_0^{\infty} \beta(a') n(t,a') da'$

where $\beta(a)$ is the jump (reset, division, or death) rate at age $a$ (Altus et al., 2022, Berry et al., 2015). This system represents deterministic aging (via $\partial_a n$), removal by jumps ($\beta(a) n$), and the instantaneous reintroduction or birth at age zero (boundary condition). In generalizations, $\beta(a)$ can depend additionally on other structuring variables (traits, spatial position, or population density).

Stochastic and kinetic extensions describe the evolution of the full probability distribution over population sizes and age-configurations, using a hierarchy of master equations or kinetic equations (see §5).

2. Extensions: Multivariate, Trait-Structured, and Jump Models

In more elaborate contexts, the state variable is extended to $(a,x)$ where $x$ is a phenotypic trait or spatial coordinate, leading to a density $n(t,a,x)$ evolving under an age-structured PDE with nonlocal jump boundary conditions (Nordmann et al., 2020, Méléard et al., 2010): $$\partial_t n(t,a,x) + \partial_a (A(a,x) n) + [\mu(a,x) + \rho(t)] n = 0$$

$$A(0,x)\,n(t,0,x) = \int_{\mathbb{R}^n} \int_0^\infty K(y \to x)\,b(a',y)\,n(t,a',y)\,da'\,dy$$

Here, $A(a,x)$ is the age progression rate, $\mu(a,x)$ is mortality, $\rho(t)$ is aggregate competition, and $K$ defines the jump kernel (mutation, spatial jumps). These models capture trait-dependent selection, rare mutations via nonlocal renewal, and strong nonlinearities due to competition or resource constraints (Nordmann et al., 2020).

Biological aging with instantaneous jumps models (rejuvenation, premature aging) are described by piecewise-deterministic processes where the age variable experiences deterministic drift and multiplicative jumps or resets at random Poisson events (Demongeot et al., 2023).

3. Analytical Properties: Asymptotics, Stability, and Concentration

Key analytical questions include the existence, uniqueness, and long-time behavior of solutions to the renewal PDE:

• Stable Age Distributions: Under general conditions, the solution converges (in rescaled variables) to an exponentially growing or stationary profile determined by a principal eigenvalue (Perron–Frobenius theory) and associated eigenfunction (Altus et al., 2022).
• Trait Concentration and Evolutionary Dynamics: When structured by age and trait, and mutations are rare, the solution concentrates on traits maximizing long-term growth rates as characterized by a variational principal eigenvalue problem. This leads to concentration phenomena and effective trait selection governed by a constrained Hamilton–Jacobi equation (Nordmann et al., 2020).
• Self-similarity in Subdiffusive Systems: For heavy-tailed waiting times, the age distribution converges toward a self-similar profile (Dynkin–Lamperti law), with explicit rates established via entropy methods (Berry et al., 2015).
• Moment and Distributional Properties: Moment equations reveal dichotomies between bounded and diverging moments, corresponding to heavy-tailed distributions in models with nontrivial jump structure (Demongeot et al., 2023).
• Exponential Ergodicity in Branching Models: Under Lyapunov and coupling conditions, age-structured branching jump processes converge exponentially toward a stationary ergodic distribution in trait–age space, even with unbounded and singular jump kernels (Olayé et al., 2024).

4. Scaling Limits and Emergent Fractional Dynamics

Age-structured jump models are instrumental in deriving macroscopic transport and evolutionary limits:

• Fractional Subdiffusion: In spatial models where the inter-jump waiting time has infinite mean (power-law tails in $\beta(a)$), a suitable rescaling yields a time-fractional diffusion equation in the macroscopic limit (Berry et al., 13 Jan 2026). The Caputo derivative emerges from the heavy-tail waiting time via a renewal equation, rigorously connecting individual-based age-structured models with anomalous diffusion PDEs.
• Nonlinear Superprocess Limits: In large-population limits with multiple scales (fast aging, slow trait evolution), separation-of-timescale analyses show that trait marginals converge to nonlinear superprocesses with density-dependent drift and branching, while age-distributions rapidly equilibrate conditional on trait (Méléard et al., 2010).

5. Stochastic and Kinetic Theories

Age-structured jump models support rigorous stochastic formulations essential for finite populations, correlations, and demographic noise:

• Kinetic Theory: The evolution of the $n$-particle age distribution $P_n(a_1,\ldots,a_n;t)$ is governed by master (Liouville) equations incorporating birth, death, jump, and fission terms. These equations admit a BBGKY-type hierarchy for marginals, including higher-order correlation functions (Chou et al., 2015, Greenman et al., 2015).
• Non-Factorizability and Moment Hierarchies: The full hierarchy does not factorize except in mean-field limits, and thus accounts for stochastic fluctuations and population size effects (carrying capacity, density dependence, randomness in small populations).
• Deterministic Mean-Field Limit: In the large-number, weak-correlation regime, factorial-moment hierarchies close to the classical deterministic McKendrick–von Foerster equation.
• Microscopic Derivations and Propagation of Chaos: Particle-level jump processes converge to deterministic or stochastic age-structured equations under mean-field and propagation-of-chaos limits (Quiñinao, 2015).

6. Applications and Numerical Analysis

Age-structured jump models provide the mathematical underpinning for a diverse set of applications:

• Cell Cycle and Demographic Modeling: Classical renewal theory for age-structured populations, cell division models, and the analysis of asynchronous exponential growth (Altus et al., 2022).
• Evolutionary Dynamics and Adaptive Selection: Trait-age-structured models elucidate selection for optimal traits under competition and rare mutations (Nordmann et al., 2020).
• Subdiffusive Transport in Biophysics: Age-residence models derive the emergence of anomalous transport and self-similar subdiffusive profiles in systems with heavy-tailed waiting times (Berry et al., 2015, Berry et al., 13 Jan 2026).
• Neuroscience: Renewal equations for “age since last spike” in integrate-and-fire neuronal networks and their mean-field limits (Quiñinao, 2015).
• Population Genetics and Epigenetic Aging: Piecewise-deterministic models for biological (epigenetic) age under random rejuvenation and premature aging events, retrieving long-tailed and broad population distributions (Demongeot et al., 2023).
• Numerical Methods: Upwind finite-difference schemes, explicit and semi-implicit time-stepping, and renewal boundary updates enable accurate simulation of both age and trait dynamics, with concentration observed through Dirac-mass formation in trait space near optima (Nordmann et al., 2020).

7. Methodological Variants and Theoretical Frameworks

Variations and generalizations include:

• Nonlocal and Singular Jump Kernels: Incorporation of singular transition kernels (e.g., for telomere-length dynamics) and control of ergodic properties under non-compact and irregular kernels (Olayé et al., 2024).
• Multi-structured and Spatial Models: Coupling of multiple physiological structures (age, size, organelle age) and incorporation of spatial movement, leading to multidimensional renewal equations (Altus et al., 2022, Chou et al., 2015).
• Non-autonomous and Time-dependent Problems: Analysis of non-autonomous equations under rescaling, leading to pseudo-equilibria which themselves relax to stationary distributions (Berry et al., 2015).
• Closure Schemes: Systematic use of mean-field, pair-approximation, and truncation methods for deriving tractable equations from otherwise infinite hierarchies (Chou et al., 2015).

Age-structured jump models thus constitute a broad and technically rigorous domain. They provide the primary mathematical framework for linking microscopic, stochastic population dynamics with macroscopic, deterministic or stochastic continuum limits, supporting both qualitative and quantitative predictions across population dynamics, evolutionary biology, and anomalous transport (Nordmann et al., 2020, Altus et al., 2022, Méléard et al., 2010, Berry et al., 13 Jan 2026, Berry et al., 2015, Chou et al., 2015, Greenman et al., 2015, Quiñinao, 2015, Demongeot et al., 2023, Olayé et al., 2024).