---
title: Stealthy-Hyperuniformity Overview
url: https://www.emergentmind.com/topics/stealthy-hyperuniformity
type: topic
---

# Stealthy-Hyperuniformity Overview

Stealthy-hyperuniformity denotes a stronger-than-hyperuniform suppression of density fluctuations in statistically homogeneous point configurations and two-phase media: the structure factor satisfies \(S(\mathbf{k})=0\) throughout a finite exclusion region \(0<|\mathbf{k}|\le K\), rather than merely vanishing in the limit \( |\mathbf{k}| \to 0\). Because this imposes an exact spectral gap around the origin, stealthy systems are Class I hyperuniform, exhibit anomalously small number-variance growth, and completely suppress single scattering over a finite band of wavevectors [1508.04749]. In current usage, the concept spans point patterns, particle packings, spin chains, networks, and composite media whose microstructure is constrained in Fourier space so that long-range homogeneity coexists with disorder, isotropy, and tunable short-range order.

## 1. Formal definition and statistical descriptors

For a statistically homogeneous point configuration of number density \(\rho\), the pair correlation function \(g_2(r)\) and total correlation function \(h(r)=g_2(r)-1\) determine the structure factor through
\[
S(\mathbf{k})=1+\rho \,\tilde h(\mathbf{k}).
\]
Hyperuniformity is the condition
\[
\lim_{|\mathbf{k}|\to 0} S(\mathbf{k})=0,
\]
which is equivalent to anomalously suppressed long-wavelength density fluctuations. In real space, the number variance \(\sigma_N^2(R)\) in a spherical observation window of radius \(R\) grows more slowly than the window volume; for Class I systems, including stealthy ones, \(\sigma_N^2(R)\sim R^{d-1}\) as \(R\to\infty\) [1508.04749].

Stealthy-hyperuniformity strengthens this requirement by enforcing
\[
S(\mathbf{k})=0 \qquad \text{for} \qquad 0<|\mathbf{k}|\le K,
\]
or, more generally for two-phase media,
\[
\tilde{\chi}_V(\mathbf{k})=0 \qquad \text{for} \qquad 0<|\mathbf{k}|\le K,
\]
where \(\tilde{\chi}_V\) is the spectral density, i.e. the Fourier transform of the autocovariance. In packings of identical spheres or disks, these descriptions are directly linked through
\[
\tilde{\chi}_V(k)=\phi_2\,\tilde{\alpha}_2(k;a)\,S(k),
\]
so stealthiness at the point-pattern level induces stealthiness in the corresponding two-phase medium [2401.08448].

This finite exclusion region in reciprocal space is the defining distinction from generic hyperuniformity. It produces an exact low-\(k\) gap in scattering, not merely a small-\(k\) asymptotic decay, and this sharper constraint is responsible for many of the optical, transport, and rigidity properties associated with the subject.

## 2. Collective-coordinate formulation and model interactions

A standard realization of stealthy-hyperuniformity uses collective coordinates in a periodic fundamental cell of volume \(v_F\). For \(N\) points at positions \(\mathbf{r}_j\), the collective density variable is
\[
\tilde n(\mathbf{k})=\sum_{j=1}^N e^{-i\mathbf{k}\cdot \mathbf{r}_j},
\]
with single-configuration structure factor \(S(\mathbf{k})=|\tilde n(\mathbf{k})|^2/N\) for \(\mathbf{k}\neq \mathbf{0}\). If a pair interaction has Fourier transform \(\tilde v(\mathbf{k})\) that is nonnegative and supported only on \(0<|\mathbf{k}|\le K\), then the total energy can be written as
\[
\Phi(\mathbf{r}^N)=\frac{1}{2v_F}\left[\sum_{\mathbf{k}}\tilde v(\mathbf{k})|\tilde n(\mathbf{k})|^2-N\sum_{\mathbf{k}}\tilde v(\mathbf{k})\right].
\]
Equivalently, over symmetry-independent constrained modes,
\[
\Phi(\mathbf{r}^N)=\frac{1}{v_F}\sum_{\mathbf{k}\in K\text{-indep}}\tilde v(\mathbf{k})|\tilde n(\mathbf{k})|^2+\Phi_0.
\]
Since \(\tilde v(\mathbf{k})\ge 0\) inside the stealthy window, minimizing \(\Phi\) enforces \(\tilde n(\mathbf{k})=0\), hence \(S(\mathbf{k})=0\), for all constrained wavevectors [1508.04749].

The canonical model family takes
\[
\tilde v(\mathbf{k})=
\begin{cases}
V(k), & |\mathbf{k}|\le K,\\
0, & |\mathbf{k}|>K,
\end{cases}
\qquad V(k)\ge 0.
\]
With \(K\) fixed, the inverse Fourier transform yields bounded, long-ranged, oscillatory real-space pair potentials. The force on particle \(j\) follows directly:
\[
\mathbf{F}_j=-\nabla_j \Phi(\mathbf{r}^N)=\frac{2}{v_F}\sum_{\mathbf{k}\in K\text{-indep}} \mathbf{k}\,\tilde v(\mathbf{k})\,\mathrm{Im}\!\left[\tilde n(\mathbf{k})e^{i\mathbf{k}\cdot \mathbf{r}_j}\right].
\]

This formulation also clarifies why numerical generation is subtle. Pure energy minimization from random initial conditions samples the degenerate ground-state manifold in an algorithm-dependent way. Canonical \(T\to 0\) sampling instead uses low-temperature molecular dynamics followed by minimization, which removes that artifact and, when done at sufficiently low equilibration temperature, yields ensembles that are insensitive to the detailed choice of \(V(k)\) within the stealthy window [1508.04749]. High-accuracy collective-coordinate optimization later pushed the residual “distance to stealthiness” to \(S_{\max}\approx 10^{-51}\) for systems up to \(N=2\times 10^6\) in two dimensions, showing that ultra-accurate enforcement of the constraints can be essential for diagnosing the energy landscape and wave-transport behavior [2304.09139].

## 3. Constraint fraction, phase behavior, and entropic selection

The natural control parameter is the fraction of constrained degrees of freedom,
\[
\chi=\frac{M}{d(N-1)},
\]
where \(M\) is the number of independent constrained wavevectors with \(0<|\mathbf{k}|\le K\). For fixed \(K\), decreasing the number density increases \(\chi\), while increasing density decreases \(\chi\). In the thermodynamic limit, \(\chi\) is proportional to the volume of the exclusion ball in \(k\)-space and inversely proportional to density [1508.04749].

For \(\chi<0.5\), the canonical \(T\to 0\) ensemble selects disordered stealthy ground states. These states have no Bragg peaks, satisfy \(S(k)=0\) for \(0<k\le K\), and develop increasing short-range order as \(\chi\) increases. Their structure factor just above \(K\) obeys the “pseudo” hard-sphere ansatz in Fourier space at small \(\chi\): the first peak of \(S(k)\) above \(K\) plays the role of the direct-space first-neighbor peak of an equilibrium hard-sphere fluid. This correspondence is accurate up to \(\chi\) values on the order of \(0.25\)–\(0.33\), depending on dimension, and then progressively breaks down as higher-order structure grows [1508.04749].

At \(\chi\ge 0.5\), the entropically favored states are crystalline in the infinite-system-size limit. In two dimensions, triangular lattices are favored over square lattices in Wang–Landau sampling; in three dimensions, evidence indicates that face-centered cubic order is favored near \(\chi\approx 0.58\), with face-centered cubic and body-centered cubic entropy scalings becoming nearly indistinguishable at higher \(\chi\) [1508.04749]. Stacked-slider phases also belong to the exact ground-state manifold above \(\chi=0.5\), but they are not entropically favored in the canonical zero-temperature ensemble. They are nonperiodic, statistically anisotropic, possess long-range orientational order, and have zero shear modulus because exact sliding motions preserve all stealthy constraints [1508.07669].

A complementary geometric result sharpens this phase picture. Under the Fourier-space hard-sphere analogy, the densest Fourier-space hard-sphere configuration is a Bravais lattice, specifically the reciprocal lattice of the densest real-space Bravais lattice. Thus, while disordered stealthy ground states exist for \(\chi\le 1/2\) in \(d\ge 2\), the densest possible stealthy configurations are ordered in every dimension [2404.16819].

## 4. Structural realizations: glasses, packings, polycrystals, and stackings

As \(\chi\to 0\), the degeneracy of the stealthy ground-state manifold becomes maximal. A modified collective-coordinate optimization that adds a short-ranged soft-core repulsion to the usual stealthy energy exploits this degeneracy to construct ultradense packings. In three dimensions, the resulting \(\chi\to 0^+\) packings are configurationally extremely close to maximally random jammed sphere packings: they have packing fraction \(\varphi\approx 0.638\), mean contact number \(Z(D^+)\approx 6\), gap exponent \(\gamma=0.44(1)\), and small-\(k\) scaling \(S(k)\sim Ck\), i.e. \(\alpha=1\), matching the maximally random jammed state [2504.10310].

The same soft-core construction yields ultradense disordered stealthy-hyperuniform packings throughout the disordered regime \(0<\chi<1/2\). In two dimensions the attainable packing fractions span \(\phi\in[0.67,0.86]\), and in three dimensions \(\phi\in[0.47,0.63]\), values far exceeding those obtained without the extra repulsion. Without soft-core repulsion, the maximal achievable packing fraction decreases to zero on average as \(N\) increases for \(\chi<0.5\); with it, the maximal packing fraction remains large and essentially \(N\)-independent [2503.24297]. In this regime, increasing \(\chi\) thins the contact network, eventually producing linear polymer-like chains of contacting particles with progressively shorter mean chain lengths.

Polycrystalline stealthy-hyperuniform media constitute a distinct realization. In two dimensions, collective-coordinate optimization generates ultradense polycrystalline disk packings with \(\phi\) between \(0.80\) and \(0.86\), local triangular order, and grain boundaries, while still maintaining \(S(k)=0\) for \(0<k\le K\). These polycrystals differ sharply from Lubachevsky–Stillinger reference polycrystals: their grain-size distribution develops a pronounced peak, and the grain–boundary two-phase autocovariance exhibits long-range oscillations, indicating inter-grain correlations induced by the stealthy constraint [2606.23628].

Stealthiness also occurs in ordered stacking families. All Barlow packings, including stacking-disordered ones, are stealthy hyperuniform by the stealthy-stacking theorem, with the common lower bound
\[
K\ge \frac{4\pi}{\sqrt{3}}.
\]
Within that family, the hyperuniformity order metric \(\bar{\Lambda}\) is, to a very good approximation, linear in the fraction of fcc-like clusters, showing that large-scale fluctuation suppression is governed primarily by local three-layer stacking geometry [1810.09883]. This illustrates a broader point: stealthiness does not imply either full periodicity or full amorphousness, but rather a reciprocal-space constraint that can coexist with crystals, polycrystals, and highly degenerate disordered states.

## 5. Wave, transport, and network consequences

The most direct physical consequence of stealthy-hyperuniformity is the exact suppression of scattering for \(0<|\mathbf{k}|\le K\). In two-phase media, nonlocal strong-contrast theory shows that layered and transversely isotropic stealthy-hyperuniform media possess finite transparency windows in wavevector space, and that for layered transverse polarization and transversely isotropic TM polarization this perfect transparency persists through third order in the expansion. Within the predicted interval there is, in practice, no Anderson localization because the localization length is much larger than any practically large sample size [2401.08448]. The same low-\(k\) gap also accelerates interphase diffusion: the excess spreadability decays exponentially in stealthy media, whereas nonstealthy media exhibit algebraic long-time tails.

These principles have been realized experimentally in photonics. In three dimensions, an optimized disordered stealthy-hyperuniform dielectric network fabricated for microwaves exhibits the first experimental isotropic photonic band gap of this type. For refractive index \(n=2.1\), the complete isotropic gap is centered at \(f\approx 0.34\,(c/a)\) with width \(\approx 0.6\%\); at \(n\approx 3.4\), not fabricated in that work, the calculated gap widens to \(\approx 14\%\) and shifts to \(f\approx 0.24\,(c/a)\) [2403.08404]. A periodic diamond network of similar node density shows anisotropic gaps, while an amorphous non-stealthy reference shows no complete gap, demonstrating that isotropy alone is insufficient.

A related semiconductor realization uses a hole-based stealthy-hyperuniform pattern on a quantum-cascade platform for the mid-infrared. The measured TE reflection band is centered near \(12.0\,\mu\mathrm{m}\) with a gap-midgap ratio of \(9.8\%\), and the reflection spectrum remains unchanged under in-plane rotation, directly confirming spatial isotropy in the optical response [2406.17154].

Transport and effective-medium advantages extend beyond photonic gaps. Two-phase media derived from ultradense stealthy-hyperuniform packings exhibit faster diffusion spreadability than media generated without soft-core repulsion, and the imaginary part of the effective dynamic dielectric constant vanishes at a small wavevector, implying perfect transparency for the corresponding modes. Cross-property relations link the transparency-window size to the long-time spreadability rate, since both are governed by the same spectral density \(\tilde{\chi}_V(k)\) [2503.24297].

Network models inherit related benefits. Delaunay triangulation networks built from disordered stealthy-hyperuniform point sets have lower percolation thresholds than Poisson networks under distance-dependent bond occupation, and the threshold decreases as \(\chi\) increases. In two dimensions, systems with large \(\chi\) fall into the same universality class as lattice percolation, whereas Poisson and low-\(\chi\) systems exhibit shifted critical exponents [2603.16167]. This suggests that suppressing long-wavelength density fluctuations systematically improves global connectivity in statistically homogeneous disordered networks.

## 6. Generalizations, rigorous results, and current points of interpretation

Stealthy-hyperuniformity is not restricted to isotropic Hermitian point sets. In two-phase composites, inverse design can impose arbitrary realizable exclusion regions \(\Omega\) in Fourier space, producing anisotropic stealthy-hyperuniform microstructures with circular, elliptical, square, rectangular, butterfly-shaped, or lemniscate-shaped spectral gaps. These media may be directionally hyperuniform—stealthy along selected directions but not others—while still possessing nearly isotropic local morphology [2307.13833]. In one-dimensional deep-subwavelength multilayers, a stealthy gap in the spectrum of \(n^2(z)\) classifies intermediate phases between crystals and uncorrelated disorder and enables angle-selective control of localization through the breakdown of effective-medium theory by Goos–Hänchen-dominated interface physics [2410.07875].

Open-system wave physics has introduced a further extension. For complex material potentials \(V(\mathbf{r})=V_R(\mathbf{r})+iV_I(\mathbf{r})\), the scattering intensity splits into an even part
\[
S_A(\mathbf{k})=\langle |V_R(\mathbf{k})|^2\rangle+\langle |V_I(\mathbf{k})|^2\rangle
\]
and an odd real–imaginary cross term
\[
S_C(\mathbf{k})=2\,\mathrm{Im}\langle V_R(\mathbf{k})V_I^*(\mathbf{k})\rangle.
\]
In this setting, hyperuniformity requires only that the real and imaginary parts are each hyperuniform, whereas stealthiness depends essentially on the cross-correlations because cancellation between \(S_A\) and \(S_C\) can create unidirectional stealthy phases unavailable in Hermitian media and in non-Hermitian crystals [2602.09458]. Experimentally, residual single scattering inside the nominal stealthy region has also been observed in two-dimensional photonic-crystal slabs with stealthy-hyperuniform disorder, where a complex effective mass induced by radiative loss makes the dynamics intrinsically non-Hermitian and leaves a finite linewidth even below the nominal \(|\mathbf{k}|=K/2\) threshold for Hermitian single scattering [2507.05253].

The concept also has rigorous probabilistic consequences. For translation-invariant stochastic processes whose structure function vanishes on an open set \(U\), the exterior configuration outside any bounded domain determines the exact interior configuration with probability \(1\); such systems are maximally rigid. When \(0\in U\), stealthy-hyperuniform point processes additionally have bounded holes, with the maximal hole size bounded above by a constant proportional to the inverse radius of the spectral gap [1707.04328]. These results supply mathematically precise versions of two empirically central features: suppressed voids and unusually strong long-range constraints.

Several apparent counterexamples have therefore been re-evaluated. Gyromorphs, whose structure factors contain rings of Bragg-like peaks with discrete rotational symmetry, were proposed as disordered media reproducing stealthy-hyperuniform advantages without stealthiness or hyperuniformity. Subsequent analysis showed instead that, in the large-\(G\) nearly isotropic limit, gyromorphs are hyperuniform of Class III, not stealthy, and display size-dependent pseudogaps densely populated by localized states rather than the smooth gaps of highly stealthy hyperuniform materials [2606.22300]. The broader implication is that isotropy, disorder, or even hyperuniformity alone do not reproduce the full phenomenology of stealthy-hyperuniformity; the finite exclusion region \(0<|\mathbf{k}|\le K\) remains the distinctive structural ingredient.

Source: https://www.emergentmind.com/topics/stealthy-hyperuniformity