Skip to main content

Turing Space

Module 4: Space and Pattern Formation

NoteReading
  • Compulsory: Britton (2003), Chapter 7, Section 7.4.1 (pp. 214 to 217): the curve of potential marginal Turing instability in the plane of the diffusion coefficients.
  • Recommended: Murray (2003), Chapter 2, Section 2.5.

The Turing space is the set of parameters for which the four Turing conditions hold (Murray 2003, sec. 2.5). For Gierer-Meinhardt you can draw it in the \((a, d)\) plane at fixed \(b\).

Britton draws the same set in the plane of the two diffusion coefficients. Its boundary is the curve \((D_2 f_u + D_1 g_v)^2 = 4 D_1 D_2 \det J\), taken where \(D_2 f_u + D_1 g_v > 0\). It is the envelope of the curves \(a_2(\lambda) = 0\) over all \(\lambda\) and separates the region of potential Turing instability from the region without it (Britton 2003, Equations 7.4.28 and 7.4.29). With \(D_1 = 1\) and \(D_2 = d\) this is the critical ratio \(d_c\) of the Turing Conditions page. The diagram below fixes the kinetics through \(a\) and \(b\) and varies \(d\) instead.

The Region in Closed Form

From Linearization, the Gierer-Meinhardt Jacobian at the steady state has \(f_u = b(1-a)/(1+a)\), \(g_v = -1\), and \(\det J = b\). The four conditions reduce to:

  • \(\det J = b > 0\), always true.
  • \(\operatorname{tr} J < 0\), that is \(f_u < 1\).
  • \(f_u > 0\), that is \(a < 1\) (needed for \(d f_u + g_v > 0\)).
  • \(d f_u - 1 > 2\sqrt{b d}\), which gives

\[ \sqrt{d} > \frac{\sqrt{b} + \sqrt{b + f_u}}{f_u}, \qquad f_u = \frac{b(1-a)}{1+a}. \tag{1}\]

For \(b = 1\) the trace condition holds for every \(a > 0\), and the Turing space is the region above the curve Equation 1 with \(0 < a < 1\).

Figure 1: Turing space of Gierer-Meinhardt for \(b = 1\) (shaded), computed with amlab.pdes_1d.is_turing_instability. The dashed line is the boundary from Equation 1. The markers are the two test cases of the laboratory.

Reading the Diagram

  • The curve diverges as \(a \to 1\), because \(f_u \to 0\) and activation disappears.
  • For \(a = 0.4\) the boundary sits at \(d_c \approx 26.24\). Case A lies inside, Case B outside.
  • The diagram depends only on \(a\), \(b\) and \(d\). It does not depend on \(\gamma\) or on the domain. Whether a pattern appears on a given domain also needs an admissible mode in the unstable band (Dispersion Relation).

The same diagram appears in the interactive simulator at bam-applied-math-lab.streamlit.app, where you can click on a point of the \((a, d)\) plane and watch the resulting 1D simulation.

NoteExercise

Increase \(b\) from \(1\) to \(2\) at \(a = 0.4\). Does the critical ratio \(d_c\) increase or decrease? Compute it with Equation 1, and check whether the trace condition still holds.

With \(b = 2\), \(f_u = 2 \cdot 0.6 / 1.4 = 6/7 \approx 0.857 < 1\), so the trace condition holds. Then \(\sqrt{d_c} = (\sqrt 2 + \sqrt{2.857})/0.857 \approx (1.414 + 1.690)/0.857 \approx 3.62\), so \(d_c \approx 13.1\). The critical ratio decreases.

import numpy as np
from amlab.turing import critical_diffusion_ratio, gierer_meinhardt_jacobian

print(critical_diffusion_ratio(*gierer_meinhardt_jacobian(a=0.4, b=2.0)))

Next: Gierer-Meinhardt Model in 1D.

References

Britton, Nicholas F. 2003. Essential Mathematical Biology. Springer Undergraduate Mathematics Series. Springer. https://doi.org/10.1007/978-1-4471-0049-2.
Murray, James D. 2003. Mathematical Biology II: Spatial Models and Biomedical Applications. 3rd ed. Vol. 18. Interdisciplinary Applied Mathematics. Springer. https://doi.org/10.1007/b98869.