Range expansion by growth and congestion

1. Problem Statement and Motivation

The paper addresses the mathematical modeling of range expansion in populations where individuals are immotile (e.g., cell colonies, tumors) but expand spatially due to growth and congestion.

• Limitations of Classical Models: Traditional reaction-diffusion equations (like Fisher-KPP) assume random motion (Brownian motion) leading to linear diffusion. These models often fail to capture the behavior of immotile cells, which do not move autonomously but are displaced by mechanical pressure from neighboring cells dividing. Furthermore, classical models do not inherently guarantee time-monotonicity (solutions can decrease locally initially), whereas experimental observations of cell colonies show that cells only move away when pushed by new growth.
• The Phenomenon: The authors model a system where local proliferation (mitosis) creates congestion. When a location reaches maximum packing density ($u=1$), new cells generated there are mechanically pushed to nearby, less congested locations. This creates a nonlocal, nonlinear dispersal mechanism driven by space competition rather than resource depletion.
• Secondary Application: The authors briefly discuss the applicability of this framework to urban sprawl, where population growth in saturated city centers forces expansion into surrounding undeveloped land.

2. Methodology and Model Derivation

A. The Nonlocal Nonlinear Model

The authors introduce a system describing the evolution of population density $u(t, x)$:
$$\partial_t u = F[u] + L[u]$$
Where:

• $F[u]$ (Growth): Represents local production limited by nonlocal congestion: $g(u)(1 - K * p)$.
• $L[u]$ (Dispersal): A conservative integral operator representing the mechanical shuffling of cells. It accounts for the loss of cells from a site $x$ due to pressure and the gain from cells pushed from saturated sites $y$.
• $p = P(u)$: A pressure variable, typically $P(u) = u^\gamma$, where $P(1)=1$ represents saturation.
• $K$: A non-negative, normalized interaction kernel with compact support, representing the range of mechanical interaction.

Key properties of this model include time-monotonicity ($\partial_t u \geq 0$) and the preservation of the bound $0 \leq u \leq 1$, reflecting the incompressibility of the cells.

B. The Singular Limit

The core mathematical contribution is the rigorous derivation of a singular limit as the pressure law becomes infinitely stiff ($\gamma \to \infty$, i.e., $P(u) = u^\gamma$).

• Assumption: In the limit, movement only occurs from saturated areas ($u=1$). Cells created in unsaturated areas ($u<1$) stay put.
• Resulting Equation: The limit yields a free boundary problem (obstacle type):
  $$\begin{cases} \partial_t u = g(u)(1 - K * \mathbb{1}_{\{u=1\}}) + g(1)K * \mathbb{1}_{\{u=1\}} & \text{if } u < 1 \\ \partial_t u = 0 & \text{if } u = 1 \end{cases}$$
  Here, the pressure $p$ becomes the characteristic function $\mathbb{1}_{\{u=1\}}$. This differs from the classical Hele-Shaw limit (associated with porous medium equations) as it describes a "pushing" mechanism rather than a diffusive one.

3. Key Contributions and Mathematical Results

The paper establishes the well-posedness and asymptotic behavior of the singular limit model.

A. Well-Posedness and Regularity

• Existence: The authors prove the existence of solutions to the singular limit by passing to the limit in the regularized problem ($\gamma < \infty$). They establish that the solution $u$ is Lipschitz continuous in both time and space.
• Regularity of the Free Boundary: They show that the characteristic function of the saturated set, $\mathbb{1}_{\{u=1\}}$, belongs to $BV_{loc}$ (Bounded Variation), indicating a well-defined free boundary.
• Comparison Principle: A crucial result is the proof of a comparison principle (Proposition 2.3), which guarantees the uniqueness of the solution. This requires a specific condition on the growth function $g$ (specifically $g(u) \leq g(1)$), ensuring that growth does not accelerate in under-saturated regions relative to saturated ones in a way that breaks monotonicity.

B. Invasion and Stationary States

• Global Invasion: If the initial population is non-zero and the kernel $K$ has positive support, the population eventually invades the entire space ($u(t, x) \to 1$ as $t \to \infty$).
• Stationary States: The only stationary solutions are the trivial states $u \equiv 0$ and $u \equiv 1$.

C. Traveling Waves

The authors prove the existence of planar traveling wave solutions of the form $u(t, x) = \phi(x \cdot e - ct)$.

• Minimal Speed: There exists a minimal speed $c^* > 0$ such that traveling waves exist for all $c \geq c^*$.
• Profile Characteristics:
  • For $c > c^*$, the profile is strictly positive everywhere.
  • For $c = c^*$, the profile is "semi-compactly supported," meaning it is zero for $s > \ell$ (where $\ell$ is the support radius of $K$).
• Bounds: The minimal speed $c^*$ is bounded by $g(1) \int_0^\ell h(s) ds \leq c^* \leq \ell \sup g$.

D. Asymptotic Spreading Speed

The paper rigorously characterizes the spreading of compactly supported initial data:

• Upper Bound: Solutions cannot spread faster than the minimal speed $c^*$.
• Lower Bound: Using a radially symmetric subsolution constructed from the minimal speed traveling wave, the authors prove that the saturated set expands with speed exactly $c^*$.
• Support vs. Congested Set: A significant finding (Proposition 2.9) is that the support of the solution ($\{u > 0\}$) and the saturated set ($\{u = 1\}$) remain within a finite distance (specifically the kernel support radius $\ell$) of each other. The "active" growth zone is confined to an annulus of width $\ell$ around the saturated core.

4. Significance and Implications

1. New Mathematical Framework: The paper introduces a novel class of nonlocal reaction-dispersion equations that differ fundamentally from standard Fisher-KPP or porous medium models. The mechanism of dispersal is driven by congestion-induced displacement rather than random motion or resource gradients.
2. Biological Relevance: The model aligns better with experimental observations of immotile cell colonies (e.g., epithelial tissues, tumors) where:
   • Expansion is monotone in time.
   • Cells do not move unless pushed.
   • The front is sharp, and the "active" region is thin.
3. Numerical Tractability: Unlike the Hele-Shaw limit which can be numerically stiff, the derived singular equation (1.8) is described as "very simple to simulate numerically," making it a practical tool for biological modeling.
4. Generalizability: The framework is adaptable to other fields, such as urban planning (modeling city growth where saturation forces expansion into new zones), demonstrating the versatility of the "growth + congestion" mechanism.

In summary, Berestycki and Mellet provide a rigorous mathematical foundation for range expansion driven by mechanical pressure, bridging the gap between biological observations of immotile colonies and partial differential equation theory through the analysis of a singular free boundary problem.