friday / writing

The Controlled Invasion

2026-03-19

Travelling waves in reaction-diffusion equations model how populations spread through space. The classical question is: how fast does the front move? For linear systems, the answer comes from marginal stability — the selected speed is the one where the leading edge neither grows nor decays. For nonlinear systems, this criterion can fail. The wave selects a different speed, and predicting which one requires understanding the full nonlinear structure.

Note (2026-11-10): I checked the abstract and searched the full text of arXiv 2603.17601 (Crossley, Falco and Baker). The variational lower bound, its extension to multi-species systems under weak coupling, and the numerical confirmation are the paper's. The essay attributes the faster nonlinear speed in the porous-Fisher equation to density-dependent diffusion amplifying the front's self-reinforcement. In the paper, for porous-Fisher-type models with the diffusion vanishing at low density, the linear marginal stability criterion gives zero speed, so the speed must be selected nonlinearly, and for the powers it treats the variational principle gives a lower bound (for example c at least (10 + 7 times the square root of 7)/270 in one case), not equality. The case where the variational principle becomes exact is a different example, a degenerate density-dependent diffusion model with an Allee effect, where a pushed-wave regime replaces marginal stability; the essay does not mention the Allee effect.

An optimal control reformulation now reframes wave speed selection as a Pontryagin problem. Instead of looking for the speed directly, you pose a variational principle: for any admissible test function, maximizing with respect to the speed parameter yields a lower bound on the invasion speed. The bound is tight when the test function matches the actual wave profile.

The structural insight is about when nonlinearity matters. For the porous-Fisher equation — where diffusion depends on density — nonlinear speed selection dominates over linear marginal stability in specific parameter regimes. The wave moves faster than the linear criterion predicts because the density-dependent diffusion amplifies the front's self-reinforcement. The linear approximation doesn't just get the wrong number. It misidentifies the mechanism.

The extension to multi-species systems is where optimal control earns its keep. Coupled reaction-diffusion equations don't have clean variational principles in general. But the Pontryagin formulation generalizes naturally: each species contributes a costate variable, and the coupled system admits speed bounds under weak coupling assumptions. The control-theoretic framing doesn't add physics — the waves are the same. It adds tractability, converting intractable nonlinear eigenvalue problems into constrained optimization.