Pattern formation by turbulent cascades

The study asks whether seemingly structureless turbulence can be harnessed to generate patterns with a selected wavelength. Classical pattern formation typically stems from a linear instability of a homogeneous state that selects a wavelength via the most unstable mode. By contrast, the proposed mechanism is fully nonlinear and relies on combining a direct energy cascade (from large to small scales) with an inverse cascade (from small to large scales) so that energy piles up at an intermediate wavenumber k_c. This requires a non-dissipative arrest mechanism. The authors propose odd viscosity—a non-dissipative, antisymmetric component of the viscosity tensor present in parity- and time-reversal-breaking fluids—as a scale-dependent Coriolis-like effect that induces two-dimensionalization at small scales, thereby reversing the order of cascades compared to rotating fluids. They explore theoretically and numerically how this mixed cascade produces spectral condensation and wavelength selection at an intermediate scale.

The work builds on turbulence phenomenology and cascade theory, including forward (direct) cascades in 3D turbulence and inverse cascades in 2D and rotating flows. Prior studies have shown inverse cascades can arise via mirror symmetry breaking or imposed shear in 3D. Cascades also occur in diverse media such as optics and elastic plates. Rotating turbulence exhibits inertial waves and a Zeman scale that separates quasi-2D from 3D behavior, leading to split cascades. In geophysical and plasma contexts, Rossby/drift waves arrest inverse cascades at the Rhines scale, generating zonal patterns. In space plasmas like the solar wind, a helicity barrier has been proposed to arrest cascades due to changes in inviscid invariants. Beyond energy cascades, mass cascades (e.g., droplet coalescence and breakup, aerosols, rain formation) can exhibit scale selection from the balance of aggregation and fragmentation rates. Odd viscosity, Hall viscosity, and gyroviscosity have been discussed in magnetized fluids, graphene electron fluids, and active chiral colloids, providing experimental motivation for non-dissipative transport in chiral media.

• Theory: The Navier–Stokes equations are extended to include odd viscosity: D_t v = -∇P + ν ∇^2 v + ν_odd e_z × ∇^2 v + f(t,x), with incompressibility ∇·v = 0. Here ν is shear viscosity, ν_odd is the odd (antisymmetric) viscosity coefficient, and e_z sets the chiral axis. The odd term acts like a scale-dependent Coriolis force and is non-dissipative (drops out of the energy balance).
• Wave-induced two-dimensionalization: Linear analysis shows odd waves with dispersion ω(k) = ± ν_odd k^2 |k| decorrelate triadic interactions except for 2D modes with k_z = 0 (ω = 0), forming a resonant manifold that dominates nonlinear energy transfer and induces a quasi-2D inverse cascade at sufficiently large k.
• Energy balance and transfer: The energy spectrum E(k,t) evolves as ∂_t E = -T - ν k^2 E + F, with T the nonlinear transfer and F the injection. Triadic interactions among k, p, q with k+p+q=0 govern T. The phase factor e^{i(ω_k+ω_p+ω_q)t} suppresses transfers except on the resonant manifold.
• Scaling theory and characteristic scales: Compare the odd-wave frequency to the eddy turnover frequency to define an odd-viscosity crossover wavenumber k_odd ≈ (ν_odd/ν)^{1/4}. For k << k_odd, odd viscosity is important and flow is quasi-2D; for k >> k_odd, effects are negligible and flow is 3D. Dimensional analysis with triad correlation time τ_3(k) yields: (i) Kolmogorov scaling E(k) ∝ ε^{2/3} k^{-5/3} for k << k_odd; (ii) odd-wave-dominated scaling E(k) ∝ ε^{1/2} ν_odd^{-1/2} k^{-1} for k >> k_odd. Balancing injection and dissipation gives the spectral condensation peak at k_c ∝ (ε/(ν_odd ν))^{1/4}. Peak height scales as [E(k_c)/E_0(k_c)] ≈ (ν_odd/ν)^{1/3} relative to the Kolmogorov spectrum E_0.
• Comparison to rotation: Rotating turbulence has inertial waves with ω = ±2Ω k_z/k and Zeman scale k_o ∝ Ω^{3/2} ν^{-1/2}. Crucially, the ordering of inverse/direct cascades relative to the crossover differs: odd fluids yield convergent fluxes (double arrest at intermediate scale) whereas rotating fluids yield divergent fluxes (condensation only at system size).
• Numerical simulations: Direct numerical simulations of the odd Navier–Stokes model are performed using a parallel pseudo-spectral solver. For direct cascades (injection at k_in > k_odd), they compute spectra E(k), flux Π(k) = Σ T(k′), and compensated spectra E(k)/E_0(k) to quantify spectral condensation and identify k_c. For inverse cascades (k_in ≈ or < k_odd), they analyze flux loops by decomposing the energy flux into homochiral (inverse) and heterochiral (direct) channels to show mutual cancellation below k_in and energy recirculation to small scales for dissipation. Anisotropy analysis links k_c predominantly to horizontal structure while vertical scales are set by k_odd.

• Non-dissipative arrest mechanism: Odd viscosity, a non-dissipative antisymmetric transport coefficient, induces odd waves that decorrelate 3D triadic interactions, channeling nonlinear energy transfer onto a resonant (2D) manifold and two-dimensionalizing small scales.
• Double cascade arrest and wavelength selection: Combining a direct cascade (large to small scales) with an odd-wave-induced inverse cascade (at higher k) leads to spectral condensation at an intermediate scale, selecting a characteristic wavelength independent of system size.
• Two characteristic scales: k_odd ≈ (ν_odd/ν)^{1/4} marks the onset of odd-wave dominance and quasi-2D behavior; the spectral peak occurs at k_c ∝ (ε/(ν_odd ν))^{1/4}. The peak amplification relative to a Kolmogorov reference scales as (ν_odd/ν)^{1/3}.
• Spectral regimes: For k << k_odd, E(k) ∝ ε^{2/3} k^{-5/3} (Kolmogorov). For k >> k_odd, E(k) ∝ ε^{1/2} ν_odd^{-1/2} k^{-1}, indicating accumulation and enhanced dissipation leading to a peak at k_c.
• Simulation confirmation: DNS shows quasi-2D columnar structures with selected horizontal (≈ k_c^{-1}) and vertical (≈ k_odd^{-1}) scales. Energy flux Π(k) decays as k crosses k_odd, evidencing cascade arrest and spectral condensation. Compensated spectra collapse when scaled by k/k_odd and peak near k_c as predicted.
• Inverse-cascade regime and flux loop: When energy is injected above k_odd, the inverse cascade is arrested at larger scales by a flux-loop mechanism: homochiral (inverse) and heterochiral (direct) flux channels cancel below k_in, routing energy back to small scales for dissipation and preventing blow-up at the largest scales.
• Distinction from rotation: Unlike rotating turbulence (with condensation at system size), odd fluids permute the order of direct/inverse cascades leading to convergent fluxes and pattern formation at an intermediate, tunable scale.
• Generality: Similar cascade-induced scale selection can arise via other non-dissipative arrest mechanisms (e.g., Rossby-wave Rhines scale in geophysical flows, helicity barriers in solar wind MHD) and in mass cascades (balance of coalescence and breakup rates).

The findings demonstrate that turbulent flows can self-organize into patterns via a nonlinear mechanism that arrests cascades at an intermediate scale, answering the central question of whether turbulence can generate patterns with selected wavelengths. Odd viscosity provides a tunable, non-dissipative control parameter: increasing ν_odd raises the condensation amplitude and shifts the selected scale according to k_odd ∼ (ν_odd/ν)^{1/4} and k_c ∼ (ε/(ν_odd ν))^{1/4}. This mechanism differs fundamentally from classical, linear-instability-driven pattern formation by relying on triadic nonlinear interactions and wave-induced decorrelation that create a resonant manifold. It also differs from rotating turbulence, where the cascade ordering leads to condensation only at system size. The work unifies and extends ideas of wave–eddy interactions and cascade directionality by showing how non-dissipative waves can reshape energy fluxes to converge at an intermediate scale, thereby selecting a wavelength. The implications span chiral active fluids, electron hydrodynamics, and potentially engineered media (optical/elastic metamaterials). The flux-loop arrest of inverse cascades highlights the role of helicity-channel competition in setting large-scale behavior without runaway growth. Overall, the results position odd viscosity as a practical knob to induce and tune pattern scales in turbulent systems and suggest analogous mechanisms in other domains where non-dissipative processes reshape cascades.

The study develops a theory and numerical evidence for cascade-induced pattern formation: odd viscosity introduces non-dissipative waves that two-dimensionalize the flow at small scales, reverse cascade ordering relative to rotation, and arrest both direct and inverse cascades at an intermediate, tunable scale. Two key scales emerge—k_odd and k_c—governing the onset of quasi-2D dynamics and the location of spectral condensation, respectively. Simulations confirm wavelength selection, flux convergence, and flux loops. Beyond odd fluids, similar scale selection can arise in geophysical flows (Rhines scale), space plasmas (helicity barriers), wave turbulence in parity-violating media, elastodynamics with odd elasticity, and mass cascades or even time-domain cascades. Future research directions include experimental realization in chiral active fluids or electron fluids, quantitative measurement of ν_odd/ν dependencies, controlled metamaterial platforms to design dispersion relations, and comprehensive weak-turbulence theories for odd waves.

• Experimental realization: While mechanisms are supported by simulations and theory, experimental observation requires sufficiently large ν_odd/ν and Reynolds numbers; orders of magnitude are system dependent.
• Anisotropy and approximations: Scaling arguments use assumptions such as locality, k ≈ k_⊥, and partial isotropization. Anisotropy is treated approximately and refined in Methods, but full anisotropic effects may shift exponents or prefactors.
• Finite-size and numerical choices: DNS uses pseudo-spectral methods and, in some cases, hyperdissipation to highlight flux loops; such choices and finite resolution/box size can affect spectra near peaks and flux decomposition details.
• Regime dependence: Pattern visibility is stronger for arrested direct cascades than for inverse cascades (where energy spreads over a broader k-range), potentially limiting detectability in some conditions.
• Parameter sensitivity: Selected scales depend on ε, ν, and ν_odd; uncertainties in measuring or controlling these in real systems may limit predictive accuracy. The mapping from microscopic chirality or magnetic fields to ν_odd is system specific.