How does anisotropy make a moving active fluid membrane rough or crumpled?

Two membrane fates hinge on one ratio

The paper asks what happens to a nearly flat patch of an anisotropic, inversion-asymmetric, permeable active fluid membrane that drifts along its own local normal. Its answer is that the same hydrodynamic equation produces two qualitatively different conformations, and the split is controlled by the anisotropy parameters rather than by the noise strength.

The relevant parameters are β = ν_y/ν_x and α = ν/ν_x, built from the effective anisotropic tension matrix ν_ij in the equation of motion. With ν_x and ν_y both positive, linear stability requires ν_x ν_y ≥ ν², that is β ≥ α². The threshold α_c = ±√β therefore separates a stable regime, |α| < √β, from an unstable one, |α| > √β. The instability can occur even at zero noise, so it is a property of the anisotropic drift term, not of fluctuations being amplified by chance.

Anisotropy forces a different renormalization scheme

The model is an anisotropic generalization of the Kardar-Parisi-Zhang (KPZ) equation, the standard description of a randomly growing interface whose height field roughens through a nonlinear term. In the isotropic KPZ equation, the nonlinear coupling is protected from renormalization by Galilean invariance. The paper shows the same protection holds here: the pseudo-Galilean invariance of the anisotropic equation keeps λ unrenormalized exactly, and the one-loop diagrams that would correct λ cancel.

What does get corrected are the damping parameters ν_x, ν_y, ν, and the noise strength D. Because the model lacks spherical symmetry, the authors cannot integrate wave vectors inside a spherical shell as in standard momentum-shell renormalization. Instead they integrate q_x over the whole real line and q_y over two thin strips near the cutoff, a scheme they note is equivalent to the reverse choice because anisotropic terms of the same order scale alike. This is a technical but consequential detail: the whole phase diagram rests on one-loop flow equations derived this way.

Stable anisotropy flows back to isotropy

The one-loop recursion relation for the dimensionless coupling g shows that g grows without bound as the renormalization scale increases, diverging at a nonuniversal length l_c of order one. Perturbation theory therefore breaks down once g becomes of order one, which the authors read as a signal of a strong-coupling algebraic rough phase, analogous to the two-dimensional isotropic KPZ equation.

Assuming a strong-coupling fixed point, the flow equations for α and β give β = 1 and α = 0 as the stable fixed point. In the long-wavelength limit this means ν_x → ν_y and ν → 0: the membrane becomes effectively isotropic even though it started anisotropic. The authors connect this to earlier mappings between this equation and the standard anisotropic KPZ form, and note that the flow toward an isotropic strong-coupling fixed point has also been found with nonperturbative renormalization methods. The white region of the β-α phase diagram, bounded by |α| < √β, is linearly stable but perturbatively inaccessible; flow lines starting there escape to that strong-coupling fixed point.

Simulations recover KPZ-like exponents

To test the renormalization predictions, the authors solve the equation numerically on a square grid with lattice spacing a₀ = 1, using forward-backward finite differences and an Euler time step Δt = 0.01. They measure the interface width W, the root-mean-square deviation of height from its spatial mean, averaged over realizations. W grows initially and then saturates at W_sat, which scales with system size L as L^χ.

In the regime argued to be algebraically rough, the growth exponent is insensitive to the damping parameters: varying ν_y at fixed ν_x and ν, or varying ν at fixed ν_y, leaves the slope of the growth regime nearly unchanged. What changes is the saturation level. Lowering ν_y at fixed ν_x raises W_sat because it moves the system closer to the instability threshold, and increasing |ν| at fixed ν_x does the same. Fitting the growth regime for L = 300 gives β̃ = 0.22, and measuring W_sat for L = 100, 200, 300, 400, and 500 gives χ ≈ 0.38, hence z = χ/β̃ = 1.72 with χ + z ≈ 2. The authors note that χ = 0.38 compares reasonably with existing results and is not far from mode-coupling and functional renormalization predictions, supporting two-dimensional KPZ-like scaling.

These numbers describe the rough phase only. They are measured with fixed ν_x = 0.50, D = 0.005, and λ = 24 in the exponent fits, and the authors report that different parameter sets give nearly the same exponents, which is their evidence that the rough phase is model-parameter independent.

Order survives differently in each direction

The rough phase is not simply disordered. The paper characterizes it as having translational short-range order but orientational long-range order, the same combination it attributes to a two-dimensional KPZ surface. The distinction matters because a reader might otherwise read algebraic roughness as a loss of all structure. Here the height correlations decay algebraically rather than saturating, yet the local orientation retains long-range order.

The crumpled phase is different in kind. The authors describe it as having both translational and orientational short-range order, meaning order is lost in both senses. This is a stronger statement than large roughness: it is a change in the ordering character of the membrane, not just a larger amplitude of the same fluctuations.

The unstable regime outruns the model itself

In the unstable region, |α| > √β, the authors add a stabilizing hyperdiffusion term −ν₄∇⁴h with ν₄ = 0.10 to keep high wave vectors under control. Even so, the interface width diverges rapidly after a short time. The paper is explicit that this likely means additional higher-order nonlinear terms, dropped from the equation by scaling arguments, are needed to describe this region reliably.

The snapshots taken just before divergence show ripples at an arbitrary angle for one parameter set and at π/4 for another, with characteristic lengths L_c = 0.46 and L_c = 0.32, both far smaller than the membrane size L = 200. The authors also give a crossover scale: membranes of linear size below or above L_c ~ √(ν₄/|min(ν₁, ν₂)|) are dominated by stabilizing hyperdiffusion or destabilizing diffusion respectively. So the crumpled label is supported by a linear instability and by the observed divergence, but the paper does not claim a settled steady-state description of the crumpled phase.

Reading the threshold as a practical split

The most useful takeaway is the comparison between the two regimes, not a single number. The rough phase is reached when the anisotropy stays inside |α| < √β; there the long-wavelength behavior is effectively isotropic and KPZ-like, with χ ≈ 0.38 and z ≈ 1.72 from the simulations. The crumpled outcome is reached when |α| > √β, where the linear instability dominates and the width grows without settling.

For a reader deciding which regime a membrane is in, the paper's own variables give the checklist: the ratios β = ν_y/ν_x and α = ν/ν_x, the sign and size of the off-diagonal tension ν, and the system size relative to L_c. The paper's microscopic picture, a nearly flat patch with actin filaments grafted locally normal to it, supplies one concrete setting where these anisotropic tensions arise, but the theory is hydrodynamic and the authors do not claim a specific measured membrane. The priority the paper establishes is that anisotropy is not a small correction to an isotropic membrane: it selects between two phases, and the off-diagonal term ν is the parameter that pushes a membrane across the threshold.

Sources used in this answer