= Near-unity diffusivity ratio for a Turing instability
Let $0<\delta<1$ and consider the reaction <Jacobian matrix>
$$
J=\begin{pmatrix}-1&-1\\1+\delta&1-\delta\end{pmatrix}.
$$
It is reaction-stable because its <trace> is $-\delta$ and its <determinant> is $2\delta>0$. Choose the diffusion matrix $\operatorname{diag}(d,1)$, so $d$ is the first species' <diffusivity> divided by the second's. For squared <wavenumber> $q$, the mode determinant is
$$
\det(J-qD)=2\delta+[1-d(1-\delta)]q+dq^2.
$$
Its minimum is negative at positive $q$ exactly when $d(1-\delta)-1>2\sqrt{2\delta d}$. Solving this inequality for $\sqrt d$ gives the threshold
$$
d>d_c=\left(\frac{\sqrt{2\delta}+\sqrt{1+\delta}}{1-\delta}\right)^2=1+2\sqrt{2\delta}+O(\delta).
$$
Thus the required diffusivity ratio can approach one as the reaction equilibrium approaches marginal stability. There is no universal gap from one independent of the kinetics. The threshold uses continuous wavenumbers; a finite domain must also have an allowed mode in the unstable band. Swapping species changes which physical ratio $d$ denotes.
Back to article page