Effective hyperuniformity in time-integrated stochastic Turing patterns

Anirban Mukherjee 1, Hong-Yan Shih 1,2 · 2026

Read Effective hyperuniformity in time-integrated stochastic Turing patterns by Anirban Mukherjee 1, Hong-Yan Shih 1,2 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-23677.txt.

Introduction

Mukherjee and Shih show that averaging noisy biological patterns over time reveals a hidden large scale order, in a 2026 paper on pattern formation. Turing patterns, spots and stripes arising when chemicals react and diffuse at different rates, appear across living systems. Random noise can excite such patterns even where the deterministic system would stay uniform. Cells, though, often respond to accumulated signals over time, a fact most theory ignores.

The authors integrate the fluctuating configurations over time and study the variance of particle counts inside large windows. In the Levin Segel model they prove, and confirm in simulations, that this variance falls toward a fixed floor set by reaction kinetics as windows grow. The fall signals effective hyperuniformity, an unusually even large scale arrangement, appearing without any fine tuning. The ordered range widens hugely near the Turing instability, suggesting time integration is a general route to large scale organization.

Catalog record

CREATORAnirban Mukherjee 1, Hong-Yan Shih 1,2
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE56 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.23677

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Effective hyperuniformity in time-integrated stochastic Turing patterns Anirban Mukherjee 1, Hong-Yan Shih 1,2 Abstract Demographic noise generates stochastic Turing patterns even when reaction-diffusion systems are deterministically stable. We show analytically and verify numerically in the Levin-Segel model that temporal integration of configurations reveals emergent large-scale organization. The intensive number variance in a window of size R ≫ 1 R\gg 1 approaches a finite reaction-kinetic floor as 1 / R 1/R , over a spatial range growing by orders of magnitude near the Turing instability. This yields an effectively hyperuniform, fine-tuning-free regime previously unidentified in non-conserved multispecies stochastic systems. Introduction. — Living systems often display striking spatial patterns—from stripes and spots in animal pigmentation to vegetation bands in drylands—raising a fundamental question: how does robust macroscopic organization emerge at scales far larger than those of individual constituents Murray 2003 ; Cross and Hohenberg 1993 ; Camazine et al. 2003 ; Rietkerk et al. 2021 ? Turing’s landmark theory showed that such patterns can emerge spontaneously from a homogeneous state through the coupling of a slowly diffusing activator to a faster-diffusing inhibitor, a central paradigm of reaction–diffusion pattern formation Turing 1952 ; Kondo and Miura 2010 ; Rietkerk and van de Koppel 2008 . Central to this picture, particularly in developmental biology Weber 2022 , are morphogens Smith 1996 ; Gurdon and Bourillot 2001 ; Wolpert 1996 , signaling molecules whose local concentration acts as a positional cue for how cells differentiate and organize. In most such systems noise is intrinsic and unavoidable Tsimring 2014 ; Lande et al. 2003 : stochastic reaction–diffusion models capture this discreteness through individual reaction and diffusion events, generating finite-wavelength patterns even below the deterministic Turing threshold by selectively amplifying weakly damped modes at an intermediate wave number McKane and Newman 2004 ; Lugo and McKane 2008 ; McKane and Newman 2005 ; Butler and Goldenfeld 2009 ; Butler and Goldenfeld 2011 ; Biancalani et al. 2010 ; Asslani et al. 2012 , a mechanism invoked for patterning in synthetic bacterial populations Karig et al. 2018 , cyanobacterial filaments Di Patti et al. 2018 , and Arabidopsis trichomes Di Patti et al. 2023 (for a review, see Ref. Di Patti et al. 2026 ). The prevailing treatment assumes that biological entities read the local morphogen concentration at a given instant Turing 1952 ; Wolpert 1969 ; Pagès and Kerridge 2000 ; Ashe and Briscoe 2006 , as do activator-inhibitor models of ecological pattern formation Murray 2003 ; Rietkerk and van de Koppel 2008 ; Borgogno et al. 2009 . Yet cells and organisms often do not act on instantaneous snapshots alone; in several patterning systems they accumulate signals and make fate decisions reflecting a history of morphogen Pagès and Kerridge 2000 ; Harfe et al. 2004 ; Dessaud et al. 2007 ; Dessaud et al. 2008 ; Sagner and Briscoe 2017 and stressor Ogle et al. 2015 ; Johnstone et al. 2016 activity rather than—or in addition to—the instantaneous concentration. This raises a question: can large-scale spatial organization emerge from accumulated fluctuations, beyond the instantaneous patterns at intermediate Turing scales? A particularly stringent form of large-scale spatial organization is hyperuniformity, in which long-wavelength density fluctuations are anomalously suppressed: the number variance grows more slowly than the observation volume, or equivalently the structure factor vanishes in the long-wavelength limit, S(\mathbf{q})\rightarrow 0 as |\mathbf{q}|\to 0 Torquato and Stillinger 2003 ; Torquato 2018 . Beyond crystals and quasicrystals, hyperuniformity also occurs in disordered systems that appear irregular locally but highly uniform at large scales, and has been identified in biological and ecological systems where such uniformity is often linked to function, including the avian retinal photoreceptor mosaic Jiao et al. 2014 , looped leaf-vein networks Liu et al. 2024 , dryland vegetation Ge 2023 , swimming algae Huang et al. 2021 , immune receptors Mayer et al. 2015 , and epithelial tissue layers Tan 2024 ; Siegert et al. 2026 . Deterministic pattern-forming equations, including the Swift–Hohenberg and Cahn–Hilliard equations Ma and Torquato 2017 ; Zheng et al. 2024 and generic Gray–Scott models Ballestero et al. 2025 , can also generate hyperuniform states. In equilibrium systems, hyperuniformity arises naturally in crystals, quasicrystals, and certain disordered ground states with long-range interactions or constrained correlations, whereas in stochastic nonequilibrium systems known routes to hyperuniformity are more restrictive and often rely on conservation laws: particle-number conservation can suppress long-wavelength density fluctuations Hexner and Levine 2015 ; Mukherjee and Pradhan 2023 , while conservation of the center of mass Hexner and Levine 2017 ; Mukherjee et al. 2024 or higher moments of the mass distribution Maire and Chaix 2025 enforces stronger forms of hyperuniformity. Although hyperuniform density fluctuations were recently reported in a nonconserved stochastic population model through self-organized criticality Agranov et al. 2025 , whether noisy reaction–diffusion systems can generically produce hyperuniform-like order without conservation laws or imposed constraints remains unresolved. Here we show that time integration can reveal effective hyperuniform organization in stochastic reaction–diffusion systems exhibiting finite-wavelength Turing patterns. Time integration acts as a dynamical filter for persistent fluctuations, so that the zero-frequency dynamic structure factor at small wavenumber and the long-time intensive subsystem fluctuations at large scales are controlled by a noise floor pinned by reaction kinetics. In the stochastic Turing regime, this reaction-induced floor is subdominant over an intermediate range of length scales above the Turing scale, taking over only beyond a crossover length that diverges as the Turing instability is approached. The resulting suppression of long-wavelength fluctuations reflects an incomplete but increasingly extended spatiotemporal compensation: local clustering promotes growth but also enhances inhibition, generating anticorrelated fluctuations whose spatial sum leaves only a finite reaction-controlled residual. Near the instability, the weak damping of the Turing modes sharpens the pattern and extends the oscillatory correlations, making this compensating response long-ranged and long-lived enough to let the hyperuniform-like regime dominate the floor across a broad range of scales. Thus, although particle number is not conserved, reactions impose an emergent constraint on accumulated-density fluctuations, providing a nonconserved route to effective hyperuniformity. Our stochastic simulations confirm the theory and show that temporal accumulation sharpens spatial structures nearly invisible in instantaneous snapshots, suggesting effective hyperuniformity as a diagnostic for hidden order in stochastic pattern-forming systems. Time-integrated configurations.— We consider a generic two-species stochastic reaction-diffusion process on a discrete d -dimensional hypercubic lattice with periodic boundary conditions. Each site \mathbf{r} has internal volume V and hosts integer populations N^{(X)}_{\mathbf{r}}(t) , with X=P (activator) or Q (inhibitor). Using the van Kampen system-size expansion Van Kampen 2007 ; Lugo and McKane 2008 , we decompose the local populations as N_{\mathbf{r}}^{(X)}\simeq V\rho_{\mathbf{r}}^{(X)}+\sqrt{V}\xi_{\mathbf{r}}^{(X)} . The deterministic part \rho_{\mathbf{r}}^{(X)} obeys the mean-field reaction-diffusion equation \partial_{t}\rho_{\mathbf{r}}^{(X)}=f_{X}\!\left(\boldsymbol{\rho}_{\mathbf{r}}\right)+\sum_{Y}\mathcal{D}_{XY}\nabla^{2}\rho_{\mathbf{r}}^{(Y)}, (1) where f_{X} is the local reaction kinetics and \mathcal{D} is the diffusion matrix coupling species X and Y . The fluctuating part \xi_{\mathbf{r}}^{(X)} obeys the linear Langevin equation \partial_{t}\xi_{\mathbf{r}}^{(X)}=\sum_{Y}\mathcal{R}_{XY}\,\xi_{\mathbf{r}}^{(Y)}+\sum_{Y}\mathcal{D}_{XY}\nabla^{2}\xi_{\mathbf{r}}^{(Y)}+\eta_{\mathbf{r}}^{(X)}(t), (2) where \mathcal{R} is the reaction matrix, obtained by linearizing f_{X} about the homogeneous steady state, and is the same matrix that enters the standard deterministic Turing stability analysis. The noise \eta_{\mathbf{r}}^{(X)}(t) is Gaussian and white, \langle\eta_{\mathbf{r}}^{(X)}(t)\,\eta_{\mathbf{r}^{\prime}}^{(Y)}(t^{\prime})\rangle=B^{(XY)}(\boldsymbol{\rho})\,\delta_{\mathbf{r}\mathbf{r}^{\prime}}\,\delta(t-t^{\prime}) , and encodes the demographic noise from the underlying reactions Sup . As V\to\infty at fixed time, \xi_{\mathbf{r}}^{(X)} converges to the Gaussian process governed by Eq. ( 2 )

Read the complete text