Reaction-Diffusion Systems

Цели урока

• Understand Turing instability and the D_v >> D_u condition
• Derive the dispersion relation and find the critical wave number
• Numerically simulate the Gray-Scott system with an explicit finite-difference scheme
• Connect the mathematical theory to biological pattern formation

Предварительные знания

• Second-order PDEs
• Linear algebra (eigenvalues)
• ODE stability analysis

• PDE Boundary Value Problems

Why does diffusion - normally a stabilizing process - sometimes destabilize an equilibrium and generate structure from uniformity?

• Cheetah spot and zebra stripe patterns emerge from reaction-diffusion dynamics
• Embryonic finger formation is controlled by the BMP-Noggin activator-inhibitor system
• Orientation columns in the visual cortex are a neural analog of the Turing model
• Neural Cellular Automata (Google Brain) learn reaction-diffusion rules for self-repair

Alan Turing and Morphogenesis

In 1952, while persecuted by British authorities, Turing published 'The Chemical Basis of Morphogenesis' - a paper few understood at the time, yet one that reshaped developmental biology three decades later. Turing proposed that chemical substances (morphogens) diffuse and react to form patterns. His diffusion-driven instability paradox remained theoretical until 1990, when chemists created the first controlled CIMA reaction-diffusion system. In 2012 geneticists directly confirmed the Turing mechanism in the mouse palate.

Turing Instability and the Pattern Formation Condition

In 1952, Alan Turing proposed a mathematical model of morphogenesis. The paradox at its core: a homogeneous equilibrium that is stable without diffusion becomes destabilized by diffusion - normally a smoothing process. That contradiction produces the stripe patterns on zebrafish, leopard coats, and mollusk shells.

In 2012 experiments on mice confirmed that the ridge pattern on the palate is governed by a reaction-diffusion system where FGF and Shh act as activator and inhibitor. The stripe spacing is controlled by the FGF concentration gradient.

Pattern wavelength L is approximately 2*pi/k_c. Larger diffusivity ratio d = D_v/D_u produces finer patterns. This is why the same mathematical framework models both large cheetah spots and the fine lattice of zebrafish scales.

The Gray-Scott Model and Numerical Simulation

The Gray-Scott model is a canonical example of a chemical reaction-diffusion system with a rich phase portrait. Different values of the feed rate F and kill rate k produce spots, stripes, labyrinths, moving spots, and chaotic regimes. Pearson (1993) systematically catalogued all regimes in a single parameter-space diagram.

The explicit Euler scheme is stable only when dt <= dx^2 / (2*max(Du, Dv)). Violating this condition causes the solution to blow up within a few steps. For large grids and long simulations, implicit or semi-implicit schemes are required.

Biological Patterns and Neural Field Applications

Reaction-diffusion systems describe more than chemistry. Orientation columns in the primary visual cortex V1 are modeled by Wilson-Cowan neural field equations - a direct analog: excitation spreads quickly (small diffusion), inhibition spreads more slowly (larger diffusion). The mathematical structure is identical.

Neural Cellular Automata (Google Brain, 2020) learn reaction-diffusion rules so that cells regenerate a target pattern after damage. Each cell is a small neural network reading from its local neighbors - a direct analogy to chemical signaling in embryonic tissue.

Stability Analysis and Dispersion Curves

A full analysis of a reaction-diffusion system involves: finding equilibria, linearizing, analyzing the spectrum of the diffusion-reaction operator, and plotting the dispersion curve. The critical wave number k_c marks the first spatial mode to grow - it determines the dominant pattern wavelength.

Real patterns are not strictly sinusoidal - nonlinearity saturates the growing modes and stabilizes the amplitude. Weakly nonlinear analysis (Ginzburg-Landau amplitude equation) describes this transition from exponential growth to a stable bounded pattern.

Connections to Other Areas

Reaction-diffusion systems sit at the intersection of nonlinear dynamics, developmental biology, and computational mathematics.

• Finite Element Method — Related topic
• Bifurcation Analysis — Related topic
• Neural Cellular Automata — Related topic
• Ginzburg-Landau Equation — Related topic

Итоги

• Turing instability: equilibrium is stable without diffusion but unstable to spatial perturbations when D_v >> D_u
• Critical wave number k_c sets the pattern scale; wavelength L_c = 2*pi/k_c
• The Gray-Scott model reproduces spots, stripes, and labyrinths at different values of F and k
• Biological patterns from the mouse palate to the visual cortex follow the same equations

Вопросы для размышления

• How does the pattern change if the diffusivity ratio D_v/D_u is increased by a factor of 10?
• Why are Wilson-Cowan neural field equations mathematically analogous to Turing systems?
• What biological meaning does the feed rate parameter F carry in the Gray-Scott model?

Связанные уроки

• de-25-fem — FEM solves reaction-diffusion PDEs on arbitrary domains
• de-29-einstein-equations — Nonlinear PDEs with spatial patterns - same broad class
• de-23-pde-bvp — Boundary value problems are the foundation for reaction-diffusion