Triangular SKT Cross-Diffusion Model

• Triangular SKT model is a cross-diffusion system with a strict hierarchy where each species’ dispersal depends only on higher-index species.
• Analytical approaches employ duality, entropy functionals, and compactness techniques to establish global existence and regularity in complex reaction-diffusion settings.
• The model’s framework is pivotal for studying competitive interactions in ecological systems, providing insights into controlled dispersion and food chain dynamics.

The triangular Shigesada-Kawasaki-Teramoto (SKT) model is a class of cross-diffusion systems in which the dispersal of each species or population is influenced by a strict hierarchy—each species’ diffusion rate depends only on higher-indexed species. Initially introduced to model competitive interactions among multiple species, these systems are now foundational in the rigorous study of multi-species population dynamics with structured cross-diffusion and reaction terms. A distinct feature is the “triangular” diffusion matrix, leading to analytical simplifications and enabling new compactness and regularity arguments for global well-posedness.

1. Formulation and Triangular Structure

The general triangular SKT system, for $I$ interacting populations with non-negative densities $u_i=u_i(t,x)\ge0$ $(i=1,\dots,I)$ on a periodic torus $T^N$, is given by: $$\begin{cases} \partial_t u_1 - \Delta\bigl[a_1(x,u_2,\dots,u_I)u_1\bigr] = r_1(u_1,\dots,u_I)u_1,\ \partial_t u_2 - \Delta\bigl[a_2(x,u_3,\dots,u_I)u_2\bigr] = r_2(u_1,\dots,u_I)u_2,\ \quad\vdots \ \partial_t u_{I-1}-\Delta\bigl[a_{I-1}(x,u_I)u_{I-1}\bigr] = r_{I-1}(u_1,\dots,u_I)u_{I-1},\ \partial_t u_I - \Delta\bigl[a_I(x)u_I\bigr] = r_I(u_1,\dots,u_I)u_I, \end{cases}$$ with the diffusion matrix

$$A(u) = \begin{pmatrix} a_1(x,u_2,\dots,u_I) & 0 & \cdots & 0 \ 0 & a_2(x,u_3,\dots,u_I) & \cdots & 0 \ \vdots &&\ddots & \vdots \ 0 & \cdots & 0 & a_I(x) \end{pmatrix}.$$

Each cross-diffusion coefficient $a_i$ is continuous, satisfies positive lower and upper bounds, and depends only on higher-indexed populations, so $a_i$ does not depend on $u_i$ (no self-diffusion). Each reaction function $r_i:\mathbb R^I\to\mathbb R$ is continuous, bounded above, and sub-affine. The triangular structure models a strict hierarchy (e.g., food chains), ensuring only “downward” coupling in $i$ and simplifies the analysis by decoupling the last equation and structuring the PDE system inductively (Moussa, 2017).

2. Analytical Framework, Existence, and Compactness

A central analytical challenge is showing that solutions to relaxed nonlocal triangular SKT systems converge to solutions of the classical system. The nonlocal (relaxed) version is

$$\begin{cases} \partial_t u_1^n - \Delta\bigl[a_1\left(x, u_2^n\ast\rho_2^n, \ldots, u_I^n\ast\rho_I^n\right) u_1^n\bigr] = r_1(\cdots)u_1^n,\ \vdots\ \partial_t u_I^n - \Delta[a_I(x) u_I^n] = r_I(\cdots)u_I^n, \end{cases}$$

where $\rho_i^n$ are mollifier kernels converging to the Dirac delta. As $n\to\infty$, strong $L^2$-compactness of the sequence $(u_i^n)$ is established, and the limit solves the original triangular SKT system. The compactness result relies on a new duality-based estimate for the Kolmogorov equation with variable coefficients and bounded right-hand side: $\partial_t z_n - \Delta(\mu_n z_n) = R_n z_n,$ with $0<\inf\mu_n$, $R_n$ bounded, and $\mu_n$ precompact in $L^1$. This yields strong $L^2$-precompactness and allows an inductive identification of the limit via the triangular structure—starting from the last (scalar) equation and propagating regularity up the system. No classical parabolic regularity or entropy structure is required, with the duality lemma being the principal tool to control concentration effects (Moussa, 2017).

3. Regularity, Boundary Conditions, and Improved Results

Recent advances have sharpened regularity results for the triangular SKT system using parabolic regularisation methods both with Neumann and Dirichlet boundary conditions. With homogeneous Dirichlet data on a smooth bounded domain $\Omega\subset \mathbb R^3$,

$$\begin{cases} \partial_t u - \Delta[(d_1 + \sigma v)u] = u(r_u - d_{11}u - d_{12}v), \ \partial_t v - d_2\Delta v = v(r_v - d_{21}u - d_{22}v), \ u|_{\partial\Omega}=v|_{\partial\Omega}=0, \end{cases}$$

it is shown that maximal global regularity (all derivatives in $L^p$ spaces) and Hölder-continuity hold provided initial data are $W^{2,\infty}$ and compatibility conditions are met. The crux is a dedicated Hölder-regularity theorem for scalar parabolic equations with rough (bounded, elliptic) coefficients and monotonicity $\partial_t w\ge 0$, which facilitates a bootstrap argument to obtain $L^{3+\delta}$ control for $u$ and thus closes the regularity estimates (Bouton et al., 24 Nov 2025, Desvillettes et al., 11 Mar 2025).

4. Entropy, Duality, and Energy Methods

The triangular SKT systems admit a unified treatment using entropy-Lyapunov functionals and duality arguments. Invariant energy-type functionals yield $L^p$ and gradient control for the $u$ equation, while maximum principles apply to the logistic-type $v$ equation, including upper bounds: $$0\le v(t,x)\le\max\{ \|v_{in}\|_{L^\infty}, (r_v/r_c)^{1/c} \}.$$ Duality lemmas for variable-coefficient parabolic equations allow for higher integrability estimates in critical or degenerate regimes, underpinning global existence even when power-law nonlinearities exhibit slow reaction or superdiffusivity. Singular-perturbation approximations splitting species into sub-populations with fast exchange have been proved to recover the triangular SKT as the fast-reaction limit, with uniform-in-$\varepsilon$ estimates and quantitative convergence rates (Desvillettes et al., 2014, Bouton, 8 Jan 2026).

5. Parameter Constraints, Boundedness, and Extensions

Global existence and boundedness for general triangular SKT systems rest on minimal structural conditions. For the two-species case, the system

$$\begin{cases} \partial_t u - \nabla\!\left[(d_1+\alpha_{11}u)\nabla u + b_{11}u\nabla v \right] = u(a_1-b_1u+c_1v), \ \partial_t v - \nabla\!\left[ b_{22}v\nabla u + (d_2+\alpha_{21}u+\alpha_{22}v)\nabla v \right] = v(a_2+b_2u-c_2v), \end{cases}$$

with $\alpha_{12}=0$, has global classical solutions if

$$(\alpha_{11}-\alpha_{21})\alpha_{22} > b_{11}b_{22},$$

and further mild one-sided bounds on the parameters. This condition is less restrictive than full SKT parameter regimes and allows global regularity without space-dimension limitation. The Lyapunov function $H(u,v)=\frac12\lambda u^2 + uv + \frac12\mu v^2$ is central in establishing these bounds. Numerical (spectral Galerkin) simulations confirm rapid convergence to spatially homogeneous steady-states even outside previously known parameter regimes (Kouachi et al., 2014).

6. Biological and Mathematical Significance

Biologically, the triangular case represents a strict hierarchy: species $i$’s dispersal is affected only by higher-index species. This mirrors food-webs, dominance orders, or directed influence, justifying the absence of mutual or “upward” cross-diffusion terms. The cross-diffusion and reaction coefficients are bounded, ensuring limited, saturating interspecific effects—important for mathematical parabolicity and ecological plausibility. The passage from nonlocal to local triangular SKT systems rigorously connects models of long-range sensing or interaction to classic diffusion-driven motifs in population ecology. Mathematically, the triangular structure facilitates induction arguments, sharp $L^2$ compactness via duality, and global regularity in higher dimensions and with weak initial data—a setting where full nondiagonal cross-diffusion would be intractable (Moussa, 2017, Desvillettes et al., 2014).

7. Extensions and Current Directions

The triangular SKT model has seen continued development with the derivation of explicit convergence rates in fast-reaction limits, refined Hölder regularity estimates, and broadened global existence results to higher dimensions and more general reaction-diffusion frameworks. Modern approaches combine duality, energy, interpolation, and regularization tools to extend these results to classes of reaction–cross-diffusion systems with arbitrary polynomial nonlinearities and rough coefficients, confirming robustness and flexibility of the triangular formalism for both theoretical analysis and modeling in population dynamics (Bouton, 8 Jan 2026, Desvillettes et al., 11 Mar 2025, Bouton et al., 24 Nov 2025).