Hyperuniform Disordered Patterns

• Hyperuniform disordered patterns (HUDS) are disordered configurations characterized by a vanishing structure factor at small wave numbers, ensuring suppressed infinite-wavelength density fluctuations.
• HUDS extend beyond point patterns to scalar fields and heterogeneous media, with classification based on scaling laws and exponents that reveal distinct hyperuniformity classes.
• Applications of HUDS range from photonic band gap materials and high-Q optical cavities to optimized transport properties and dynamic self-assembly in nonequilibrium systems.

Searching arXiv for recent and foundational papers on hyperuniform disordered patterns, including point patterns, scalar fields, curved manifolds, dynamics, and applications. Hyperuniform disordered patterns (HUDS) are disordered configurations that suppress long-wavelength density fluctuations while lacking crystalline long-range order and Bragg peaks. In the standard point-pattern formulation, hyperuniformity is defined by the vanishing of the structure factor at small wave number, $\lim_{|\mathbf{k}|\to 0} S(\mathbf{k}) = 0$, which is equivalent to sub-volume growth of local number fluctuations in large sampling windows (Torquato, 2018). The term is used for statistically isotropic, amorphous states in Euclidean space, but the same organizing principle has been extended to spherical surfaces, scalar fields, two-phase media, driven nonequilibrium systems, and engineered heterogeneous materials (Meyra et al., 2018).

1. Definition and classification

For a statistically homogeneous point pattern of number density $\rho$, with pair correlation function $g_2(\mathbf{r})$ and total correlation function $h(\mathbf{r}) = g_2(\mathbf{r}) - 1$, the structure factor is

$$S(\mathbf{k}) = 1 + \rho\, \widetilde{h}(\mathbf{k}),$$

and hyperuniformity requires

$\lim_{|\mathbf{k}|\to 0} S(\mathbf{k}) = 0.$

This criterion expresses the suppression of infinite-wavelength density fluctuations and distinguishes HUDS from ordinary disordered systems such as Poisson processes, for which $S(0) > 0$ (Torquato, 2018).

The reciprocal-space criterion is equivalent to a real-space statement about number fluctuations in large sampling windows. For a spherical window of radius $R$ in $d$ dimensions, hyperuniform systems have number variance $\sigma_N^2(R)$ that grows more slowly than the window volume. When the small-$\rho$0 behavior obeys

$\rho$1

the asymptotic scaling falls into three standard classes: $\rho$2 These correspond to Class I, II, and III hyperuniformity, respectively (Torquato, 2018).

The same framework has been generalized to continuous scalar fields and two-phase media. For a scalar field with autocovariance $\rho$3 and spectral density $\rho$4, hyperuniformity requires $\rho$5 as $\rho$6; for two-phase media, the analogous quantity is the spectral density $\rho$7 (Ma et al., 2017). In recent materials work, this continuous-field formulation has been used to describe disordered hyperuniform materials whose local property field $\rho$8 has spectral density $\rho$9, thereby extending HUDS from point processes to heterogeneous media with continuously varying local properties (Zhong et al., 10 Apr 2025).

A stricter subclass is stealthy hyperuniformity, defined by

$g_2(\mathbf{r})$0

which eliminates scattering over a finite range of wavevectors (Morse et al., 2024). Stealthiness is not required for HUDS, but it is central in several photonic and inverse-design constructions.

2. Geometry, variance, and diagnostics

The principal diagnostics of HUDS are the structure factor or spectral density at small wave number and the corresponding real-space fluctuation statistics. In Euclidean space, the local number variance in a spherical window has an exact representation in terms of $g_2(\mathbf{r})$1 and the scaled intersection volume of two windows, while for scalar fields the analogous local field variance is expressed through the field autocovariance (Torquato, 2018).

On finite curved manifolds, the reciprocal-space criterion must be reformulated. For points on a spherical surface $g_2(\mathbf{r})$2 of radius $g_2(\mathbf{r})$3, finite size precludes an infinite-wavelength limit of the structure factor, so hyperuniformity is defined through the scaling of the local particle-number variance in spherical-cap windows. If $g_2(\mathbf{r})$4 is the cap area and $g_2(\mathbf{r})$5 the ensemble-averaged number variance, the spherical criterion is

$g_2(\mathbf{r})$6

For spherical caps,

$g_2(\mathbf{r})$7

and the cap-perimeter scaling is particularly informative because area-based variance is non-monotonic on a closed surface (Meyra et al., 2018).

The spherical study identifies three broad regimes. Strongly hyperuniform ordered point sets obey

$g_2(\mathbf{r})$8

with $g_2(\mathbf{r})$9 the cap perimeter. Spherical HUDS satisfy

$h(\mathbf{r}) = g_2(\mathbf{r}) - 1$0

and non-hyperuniform configurations exhibit either $h(\mathbf{r}) = g_2(\mathbf{r}) - 1$1 for fixed-$h(\mathbf{r}) = g_2(\mathbf{r}) - 1$2 uniform point sets or $h(\mathbf{r}) = g_2(\mathbf{r}) - 1$3 for Poisson processes (Meyra et al., 2018). This perimeter-based criterion is the spherical analog of the Euclidean distinction between surface-area and volume-dominated fluctuations.

An alternative real-space diagnostic is the hyperuniformity disorder length $h(\mathbf{r}) = g_2(\mathbf{r}) - 1$4, defined through the variance of volume-fraction fluctuations in windows of size $h(\mathbf{r}) = g_2(\mathbf{r}) - 1$5. In this picture, fluctuations are controlled by a boundary layer of thickness $h(\mathbf{r}) = g_2(\mathbf{r}) - 1$6. For strongly hyperuniform patterns, $h(\mathbf{r}) = g_2(\mathbf{r}) - 1$7 approaches a constant at large $h(\mathbf{r}) = g_2(\mathbf{r}) - 1$8, whereas non-hyperuniform patterns exhibit $h(\mathbf{r}) = g_2(\mathbf{r}) - 1$9 (Chieco et al., 2017). This approach is useful for pixel data, grayscale images, and point patterns represented on finite grids.

For finite systems, direct small-$$S(\mathbf{k}) = 1 + \rho\, \widetilde{h}(\mathbf{k}),$$0 extrapolation can be difficult. A practical scalar-field metric is

$$S(\mathbf{k}) = 1 + \rho\, \widetilde{h}(\mathbf{k}),$$1

with effectively hyperuniform fields corresponding to small $$S(\mathbf{k}) = 1 + \rho\, \widetilde{h}(\mathbf{k}),$$2 values. Persistent-homology-based analyses of scalar fields have used this finite-size indicator together with Wasserstein distances between persistence diagrams to relate local topology to global hyperuniformity class (Milor et al., 1 Sep 2025).

3. Ordered analogs, disordered realizations, and generative mechanisms

HUDS occupy the intermediate regime between crystals and uncorrelated disorder. Crystals and quasicrystals are hyperuniform, but they display Bragg peaks and conventional long-range order. HUDS retain the suppression of large-scale fluctuations while remaining statistically isotropic and disordered (Torquato, 2018).

A canonical biological example is the avian cone photoreceptor mosaic. In chicken retina, both the total photoreceptor population and each of the five cone subtypes are hyperuniform. The small-$$S(\mathbf{k}) = 1 + \rho\, \widetilde{h}(\mathbf{k}),$$3 behavior is

$$S(\mathbf{k}) = 1 + \rho\, \widetilde{h}(\mathbf{k}),$$4

which corresponds to Class II hyperuniformity, and the number variance in two dimensions is well fit by

$$S(\mathbf{k}) = 1 + \rho\, \widetilde{h}(\mathbf{k}),$$5

with a negligible $$S(\mathbf{k}) = 1 + \rho\, \widetilde{h}(\mathbf{k}),$$6 term, confirming effective hyperuniformity (Jiao et al., 2014). Because each subtype is itself hyperuniform, the pattern is described as “multi-hyperuniform.” A multiscale packing model with short-range hard-core exclusion between all cells and finite-range soft repulsion between like cells reproduces this behavior and shows how competition between packing constraints and homotypic repulsion yields a disordered hyperuniform state (Jiao et al., 2014).

In synthetic point patterns, one important route is collective-coordinate design and stealthy optimization. Stealthy hyperuniform point patterns satisfy $$S(\mathbf{k}) = 1 + \rho\, \widetilde{h}(\mathbf{k}),$$7 for all constrained wavevectors in a spherical exclusion region, and their structure can be tuned by the stealthiness parameter

$$S(\mathbf{k}) = 1 + \rho\, \widetilde{h}(\mathbf{k}),$$8

where $$S(\mathbf{k}) = 1 + \rho\, \widetilde{h}(\mathbf{k}),$$9 is the number of independent constrained wavevectors (Morse et al., 2024). For $\lim_{|\mathbf{k}|\to 0} S(\mathbf{k}) = 0.$0, fully disordered stealthy ground states exist for $\lim_{|\mathbf{k}|\to 0} S(\mathbf{k}) = 0.$1, whereas $\lim_{|\mathbf{k}|\to 0} S(\mathbf{k}) = 0.$2 yields ordered ground states such as crystals or stacked-slider phases. Numerical work in dimensions $\lim_{|\mathbf{k}|\to 0} S(\mathbf{k}) = 0.$3 through $\lim_{|\mathbf{k}|\to 0} S(\mathbf{k}) = 0.$4 shows that disordered stealthy ground states decorrelate as dimension increases, with the structure factor approaching a step-function form outside the exclusion region (Morse et al., 2024).

Scalar and field-based HUDS can also be constructed directly. Gaussian random fields with prescribed power spectra, Cahn–Hilliard spinodal-decomposition patterns, and Swift–Hohenberg labyrinths all provide explicit routes to hyperuniform scalar fields (Ma et al., 2017). For Gaussian random fields, specifying a target spectral density $\lim_{|\mathbf{k}|\to 0} S(\mathbf{k}) = 0.$5 yields a field with the desired small-$\lim_{|\mathbf{k}|\to 0} S(\mathbf{k}) = 0.$6 class. In late-stage Cahn–Hilliard coarsening, the spectral density shows

$\lim_{|\mathbf{k}|\to 0} S(\mathbf{k}) = 0.$7

at small $\lim_{|\mathbf{k}|\to 0} S(\mathbf{k}) = 0.$8, while in Swift–Hohenberg labyrinths the spectrum develops an effectively stealthy low-$\lim_{|\mathbf{k}|\to 0} S(\mathbf{k}) = 0.$9 hole with very small hyperuniformity index values (Ma et al., 2017). Thresholding a hyperuniform Gaussian field generally destroys hyperuniformity, but thresholding a strongly non-Gaussian, nearly bimodal field can preserve effective hyperuniformity (Ma et al., 2017).

A recent continuous-property generalization uses stationary Gaussian random fields for the local material property $S(0) > 0$0, with analytical spectral density

$S(0) > 0$1

to generate hyperuniform, nonhyperuniform, and antihyperuniform materials by Fourier filtering (Zhong et al., 10 Apr 2025). This suggests that HUDS can be regarded not only as point sets but also as engineered spectra of continuous fields.

4. Nonequilibrium and dynamical hyperuniformity

A large class of HUDS arises far from equilibrium. In periodically driven and absorbing-state systems, hyperuniformity emerges dynamically through fluctuation suppression near nonequilibrium critical points (Lei et al., 2024).

In a vibrated granular layer at the continuous liquid-to-solid transition, the static structure factor satisfies

$S(0) > 0$2

with $S(0) > 0$3 in experiments guided by a field-theoretic model, and the scaled offset $S(0) > 0$4 vanishes at criticality, implying dynamic hyperuniformity (Castillo et al., 2018). The mechanism is coupling of a conserved density field to a critically fluctuating friction or order field: solid-like patches with long lifetimes and diverging correlation lengths increasingly block long-wavelength density modes, so $S(0) > 0$5 as $S(0) > 0$6 (Castillo et al., 2018). In this setting, the structure factor is more sensitive than number variance, because window averaging smears the small-$S(0) > 0$7 behavior and finite-size effects can mask the asymptotic scaling (Castillo et al., 2018).

More generally, dynamic HUDS have been reported in absorbing phase transitions, random organization, center-of-mass-conserving active matter, driven dissipative fluids, chiral active particles, and spinodal decomposition (Lei et al., 2024). The review of nonequilibrium dynamic hyperuniform states distinguishes several mechanisms. Conserved directed-percolation criticality yields exponents near $S(0) > 0$8 in two dimensions; center-of-mass-conserving active states produce

$S(0) > 0$9

and scalar-field coarsening systems such as Model B and Model H show

$R$0

at late times (Lei et al., 2024).

Spinodal decomposition is especially significant because it provides a bottom-up fabrication route. In phase-field and Cahn–Hilliard models, late-time coarsening produces hyperuniform scalar fields with $R$1 spectral suppression (Ma et al., 2017). This mechanism has been connected to experiments on spinodal solid-state dewetting of strained $R$2 films on ultrathin silicon-on-insulator substrates, where AFM height fields exhibit spectral-density scaling

$R$3

with $R$4–$R$5 for connected patterns, together with a practical hyperuniformity figure of merit $R$6 (Salvalaglio et al., 2019). In that system, Asaro–Tiller–Grinfeld instability and near-simultaneous dewetting generate effective hyperuniformity in monocrystalline semiconductor morphologies (Salvalaglio et al., 2019).

5. Materials, optical phenomena, and structure–property relations

HUDS have become central in photonics because isotropic suppression of long-wavelength scattering can produce band gaps without crystalline periodicity. A 2D isotropic hyperuniform disordered solid formed by a connected dielectric-wall network enclosing air cells exhibits a TE photonic band gap at very low dielectric contrast, with measured stop band centered near $R$7 GHz and plane-wave-expansion calculations showing a zero-density-of-states interval

$R$8

mid-gap $R$9 GHz, and normalized width $d$0 (Man et al., 2013). The key ingredients are statistical isotropy, stealthy suppression of small-$d$1 scattering, and uniform local topology. Complementary resonant modes below and above the gap localize energy predominantly in dielectric and air regions, respectively, producing a robust gap mechanism even without Bragg scattering (Man et al., 2013).

The same principle enables optical cavities. Hyperuniform disordered honeycomb networks derived from stealthy point patterns support high-$d$2 TE cavity modes in both ideal 2D and slab geometries. In 2D, simulations yield $d$3, while finite-height slabs maintain $d$4 after local cavity optimization (Amoah et al., 2015). The design strategy is to create an H1-type defect by filling one cell, adding a central air hole to tune resonance frequency, and adjusting neighboring cells to suppress Fourier components inside the light cone (Amoah et al., 2015).

These cavity concepts have been realized experimentally in GaAs slabs. Near-field optical imaging of engineered cavities in hyperuniform disordered architectures shows dipole-, quadrupole-, hexapole-, and octupole-like modes, together with coexisting Anderson-localized and topological-defect modes in a narrow spectral window (Granchi et al., 2023). The highest measured cavity quality factor is of order $d$5, exceeding the quality factors of the observed Anderson states, and the isotropic band-gap background permits cavity placement without the axis-alignment constraints of periodic photonic crystals (Granchi et al., 2023).

A microfluidic self-assembly route demonstrates how hydrodynamically formed HUDS can template isotropic photonic band gaps. In a periodically driven emulsion, droplet centers self-organize into hyperuniform point sets below a critical displacement amplitude $d$6. Using those experimentally measured coordinates as templates for arrays of infinitely tall dielectric cylinders, simulations show that the minimum dielectric contrast for opening a TM photonic band gap is

$d$7

in the hyperuniform regime and

$d$8

above the reversibility transition, with optimal conditions occurring just below $d$9 (Traktman et al., 11 Jul 2025). This suggests that hydrodynamic tuning of hyperuniformity directly controls photonic performance.

Beyond wave physics, HUDS have been connected to effective transport and field fluctuations in random media. For continuous-property materials with $\sigma_N^2(R)$0, first-order perturbation theory for the transport equation

$\sigma_N^2(R)$1

gives

$\sigma_N^2(R)$2

so the induced physical field is hyperuniform only when $\sigma_N^2(R)$3, with effective exponent $\sigma_N^2(R)$4 in the weak-contrast limit (Zhong et al., 10 Apr 2025). Numerical homogenization further shows that the effective conductivity becomes nearly isotropic and exhibits sharply reduced across-realization variance as $\sigma_N^2(R)$5 increases (Zhong et al., 10 Apr 2025).

In solid-state materials, recent work has identified disordered hyperuniform states in amorphous carbon nanotubes, amorphous 2D silica, amorphous graphene, defected transition-metal dichalcogenides, pentagonal 2D materials, and medium/high-entropy alloys (Chen et al., 2022). In these systems, hyperuniformity is associated with distinctive electronic and thermal transport mechanisms, including enhanced electronic transport in amorphous 2D silica and near-Vegard behavior with altered thermal conductivity in multihyperuniform SiGeSn alloys (Chen et al., 2022).

6. Variants, misconceptions, and open questions

A persistent misconception is that hyperuniformity is equivalent to crystallinity. The literature instead defines HUDS precisely by the coexistence of short-range disorder and large-scale uniformity: they are disordered at local scales, lack Bragg peaks, and yet satisfy the same infinite-wavelength suppression criterion as crystals (Jiao et al., 2014). Another misconception is that pair correlation functions alone reveal hyperuniformity. In finite spherical systems, for example, the pair correlation $\sigma_N^2(R)$6 does not exhibit clear hyperuniform signatures except at inaccessible asymptotic limits, so direct variance measurements or small-$\sigma_N^2(R)$7 spectral analysis are required (Meyra et al., 2018).

Window choice and finite size matter. On the sphere, area-based number variance is necessarily non-monotonic because the variance vanishes as the window approaches the full surface area; cap perimeter therefore becomes the preferred scaling variable (Meyra et al., 2018). In finite Euclidean experiments, number-variance measurements can miss small-$\sigma_N^2(R)$8 hyperuniformity because the window transform smears reciprocal-space information, whereas the structure factor can detect the onset sharply (Castillo et al., 2018).

Another unresolved issue is the transfer of hyperuniformity across coupled fields. Continuous-property random-field models show that property-field hyperuniformity with exponent $\sigma_N^2(R)$9 does not automatically imply hyperuniformity of the induced physical fields; in transport problems, the $\rho$00 weighting of the elliptic operator makes $\rho$01 the relevant threshold in the weak-contrast regime, and away from weak contrast second-order terms repopulate low-$\rho$02 modes (Zhong et al., 10 Apr 2025). This suggests that “hyperuniform material” and “hyperuniform response” are distinct notions.

The relation between geometry and topology is an active frontier. Persistent-homology analyses of Cahn–Hilliard fields and Gaussian random fields indicate that local topological features encoded in persistence diagrams correlate systematically with global hyperuniformity exponents and spectral cutoffs (Milor et al., 1 Sep 2025). This suggests that topological descriptors may complement $\rho$03, $\rho$04, and spectral-density fits in finite or noisy datasets.

Open questions remain about hyperuniformity on manifolds other than the sphere, the distinction between local and global hyperuniformity on closed surfaces, the role of topology and curvature, and generalization to cylinders, ellipsoids, and manifolds of non-constant curvature (Meyra et al., 2018). In nonequilibrium systems, the universality of observed exponents remains unresolved because different conservation laws, hydrodynamic couplings, and noise structures yield values ranging from approximately $\rho$05 to $\rho$06 in point-pattern systems and from $\rho$07 to $\rho$08 in scalar-field settings (Lei et al., 2024). In materials design, a major outstanding problem is how to realize large-scale hyperuniform structures by scalable bottom-up assembly while controlling local topology, defect distributions, and the targeted small-$\rho$09 class (Salvalaglio et al., 2019).

Taken together, these developments establish HUDS as a unifying concept across point processes, scalar fields, curved spaces, nonequilibrium dynamics, and functional materials. The central invariant is always the same: anomalous suppression of long-wavelength fluctuations without conventional crystalline order. The diversity of realizations indicates that hyperuniformity is not a single microstructure but a fluctuation principle that can be encoded in positions, interfaces, composition fields, and response fields, with consequences ranging from biological sampling and urban organization to photonic band gaps and transport optimization (Torquato, 2018).