betaFeedback

WorksPhysics and Astronomy

Contact-number hyperuniformity at glass close packing in infinite dimensions

Abstract

We investigate contact-number fluctuations of harmonic spheres near glass close packing (GCP), the amorphous jamming endpoint reached by compressing an ideal glass. Using replica theory, we show that the Gardner-unstable region shrinks toward jamming as the spatial dimension increases, allowing GCP to be approached along a stable one-step replica symmetry breaking branch in the high-dimensional limit. By coupling an external field to the local contact number, we derive the variance of the total contact number and the wave-number-dependent structure factor directly from the microscopic interactions. At GCP, the structure factor vanishes quadratically at small scaled wave numbers, demonstrating contact-number hyperuniformity. Upon weak compression above GCP, the structure factor develops a low-wave-number plateau whose height is proportional to the excess coordination. These results provide an analytical description of contact-number hyperuniformity and its breakdown upon compression.

Introduction

The jamming transition is characterized by the emergence of mechanical rigidity upon compression of a particle assembly [1–3]. Systems of frictionless repulsive spheres have been extensively studied as a simple model of this transition [4]. Numerical and theoretical studies have established isostaticity at the jamming transition and revealed scaling laws for elastic moduli, vibrational properties, and the distributions of contact forces and interparticle gaps [5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16] [4].

In addition to these scaling laws, numerical studies have revealed an anomalous suppression of contact-number fluctuations on large length scales near jamming [17, 18]. This behavior is known as contact-number hyperuniformity. In general, hyperuniformity is characterized by a variance of integrated fluctuations that grows more slowly than the volume of an observation window [19, 20]. For contact-number fluctuations at jamming, the variance was found to scale with the surface area of the window rather than its volume [17]. This suppression is also reflected in the contact-number structure factor Sz(k)S_z(k), which vanishes with power law Sz(k)∝kαS_z(k) \propto k^\alpha as k→0k \to0. The original numerical study reported an approximate scaling α≈1.5\alpha\approx1.5 [17, 21, 22]. More recently, Wang et al. found α≃2\alpha\simeq2 in jammed harmonic-sphere packings with a tuned particle-size distribution [23].

To explain the origin of contact-number hyperuniformity, Shang et al. proposed a minimal network model based on global isostaticity and local mechanical stability constraints [24]. Numerical simulations of this model yielded contact-number hyperuniformity with an exponent α≈1.5\alpha\approx1.5. However, to our knowledge, the contact-number structure factor near jamming has not yet been derived analytically from a microscopic model of interacting particles.

Replica theory provides a framework for describing glassy states starting from microscopic particle interactions [25, 26]. Early calculations for harmonic spheres used a one-step replica symmetry breaking (1RSB) approximation to study glassy and jamming transitions [27]. In the infinite-dimensional limit, the replicated free-energy functional can be derived exactly [28]. As approaches jamming, the 1RSB solution can become unstable at a Gardner transition [29, 30, 31, 32, 33, 34, 35]. Beyond the Gardner transition, a hierarchy of glassy subbasins emerges, requiring a full replica symmetry breaking (fullRSB) description. The fullRSB solution predicts the nontrivial critical scaling of the force and gap distributions near jamming [9, 10, 13, 14]. However, calculating spatial contact-number correlations within the fullRSB framework remains technically challenging.

To make analytical progress, we focus on glass close packing (GCP), the amorphous jamming endpoint of the ideal-glass branch [26]. The reduced packing fraction at GCP, φ^=2dφ/d\hat{\varphi} = 2^{d}\varphi/d, grows as log⁡d\log d, and the cage size becomes small [26, 30]. The Gardner boundary moves toward jamming as the reduced density increases [9]. For harmonic spheres, we show that the width of the Gardner-unstable region, expressed in terms of the relative excess coordination, decreases as 1/log⁡d1/\log d. In the limit d→∞d \to\infty, GCP can thus be approached along a stable 1RSB branch.

Within this 1RSB description, we introduce an external field conjugate to the local contact number. The response to a uniform field yields the mean contact number and the variance of the total contact number. At GCP, the mean reaches the isostatic value 2d2d and the variance per particle vanishes. Using a spatially varying field, we derive an explicit expression for the contact-number structure factor Sz(q)S_z(q). At GCP, it vanishes quadratically in the scaled wave number k=σq/dk = \sigma q/\sqrt{d}, yielding contact-number hyperuniformity. Upon weak compression above GCP, a finite plateau develops at small kk, with a height proportional to the excess coordination.

The remainder of this paper is organized as follows. In Sec. II, we introduce the model and explain how to compute the mean contact number and its fluctuations. In Sec. III, we derive the replicated free energy in the high-dimensional and zero-temperature limits. In Sec. IV, we analyze the response to a uniform field and locate the Gardner stability boundary. In Sec. V, we extend the calculation to a spatially varying field and derive the contact-number structure factor. We conclude in Sec. VI.

Setup

Model

We consider NN harmonic spheres of diameter σ\sigma in dd dimensions [4]. The total potential energy is

H=∑i<jv(∣xi−xj∣),(1)H = \sum_{i<j} v\left(\lvert\mathbf{x}_i-\mathbf{x}_j\rvert\right), \tag*{(1)}

where xi\mathbf{x}_i is the position of particle ii and the pair potential is

v(r)=κ2(rσ−1)2θ(σ−r).(2)v(r) = \frac{\kappa}{2}\left(\frac{r}{\sigma}-1\right)^2 \theta(\sigma-r). \tag*{(2)}

Here, κ\kappa sets the energy scale and θ\theta is the Heaviside step function.

Contact-number observables

The contact number of particle ii is

zi=∑j≠iθ(σ−∣xi−xj∣).(3)z_i = \sum_{j\ne i} \theta(\sigma-\lvert x_i-x_j\rvert). \tag*{(3)}

We define the contact field as

z(x)=∑i=1Nziδ(x−xi).(4)z(x) = \sum_{i=1}^{N} z_i\delta(x-x_i). \tag*{(4)}

Its Fourier transform is

zq=∫dx eiq⋅xz(x)=∑i=1Nzieiq⋅xi.(5)z_q = \int dx\,e^{\mathrm{i}q\cdot x}z(x) = \sum_{i=1}^{N} z_i e^{\mathrm{i}q\cdot x_i}. \tag*{(5)}

The contact-number structure factor is

Sz(q)=1N⟨δzqδz−q⟩=⟨zqz−q⟩−⟨zq⟩⟨z−q⟩N,(6)S_z(q) = \frac{1}{N}\langle\delta z_q\delta z_{-q}\rangle= \frac{\langle z_qz_{-q}\rangle-\langle z_q\rangle\langle z_{-q}\rangle}{N}, \tag*{(6)}

where δzq=zq−⟨zq⟩\delta z_q=z_q-\langle z_q\rangle. An alternative definition uses the Fourier transformation of zi−zˉz_i-\bar{z}, where zˉ\bar{z} is the mean contact number [17], leading to

Sz′(q)=1N⟨∣∑ieiq⋅xi(zi−zˉ)∣2⟩=1N[⟨zqz−q⟩−zˉ⟨zqρ−q⟩−zˉ⟨z−qρq⟩+zˉ2⟨ρqρ−q⟩].(7)S'_z(q) = \frac{1}{N}\left\langle\left\lvert\sum_i e^{\mathrm{i}q\cdot x_i}(z_i-\bar{z})\right\rvert^2\right\rangle= \frac{1}{N}\left[\langle z_qz_{-q}\rangle-\bar{z}\langle z_q\rho_{-q}\rangle-\bar{z}\langle z_{-q}\rho_q\rangle+\bar{z}^2\langle\rho_q\rho_{-q}\rangle\right]. \tag*{(7)}

The two definitions agree to leading order in the high-dimensional limit considered here because density fluctuations contribute only at subleading order (see Appendix A).

Partition function

To compute the mean contact number and its correlations, we introduce a partition function with an external field λ(x)\lambda(x) conjugate to z(x)z(x),

Z=∫∏i=1Ndxi exp⁡[−βH+∫dx z(x)λ(x)],(8)Z = \int\prod_{i=1}^{N} dx_i\,\exp\left[-\beta H+\int dx\,z(x)\lambda(x)\right], \tag*{(8)}

where β\beta is the inverse temperature. A functional derivative with respect to the field gives

⟨z(x)⟩=δlog⁡Zδλ(x).(9)\langle z(x)\rangle= \frac{\delta\log Z}{\delta\lambda(x)}. \tag*{(9)}

For a uniform field λ(x)=λ\lambda(x)=\lambda, the mean contact number per particle is

ddλ(log⁡ZN)=1N⟨∫dx z(x)⟩=1N∑i=1N⟨zi⟩.(10)\frac{d}{d\lambda}\left(\frac{\log Z}{N}\right)=\frac{1}{N}\left\langle\int dx\,z(x)\right\rangle=\frac{1}{N}\sum_{i=1}^{N}\langle z_i\rangle. \tag*{(10)}

Using the same Fourier convention for λ(x)\lambda(x) as for z(x)z(x), Parseval’s identity gives

∫dx z(x)λ(x)=1(2π)d∫dq zqλ−q.(11)\int dx\,z(x)\lambda(x)=\frac{1}{(2\pi)^d}\int dq\,z_q\lambda_{-q}. \tag*{(11)}

Differentiating with respect to the Fourier components of the field then gives

⟨zq⟩=(2π)dδlog⁡Zδλ−q,⟨zqzq′⟩−⟨zq⟩⟨zq′⟩=(2π)dδ⟨zq⟩δλ−q′.(12)\langle z_q\rangle=(2\pi)^d\frac{\delta\log Z}{\delta\lambda_{-q}},\qquad\langle z_qz_{q'}\rangle-\langle z_q\rangle\langle z_{q'}\rangle=(2\pi)^d\frac{\delta\langle z_q\rangle}{\delta\lambda_{-q'}}. \tag*{(12)}

Replica method

Replicated free energy

Following Monasson [25], we consider mm replicas constrained to occupy the same metastable state. The replicated free energy FmF_m is defined by

−βFm=log⁡Zm,(13)-\beta F_m=\log Z_m, \tag*{(13)}

where ZmZ_m is the replicated partition function. After analytic continuation in mm, the equilibrium liquid corresponds to m=1m=1, whereas the ideal-glass branch has m<1m<1. On the glass branch considered here, m→0m\to0 as the temperature goes to zero.

When the replicas occupy the same metastable state, the mm copies of particle ii remain localized near one another and can be treated as a single molecule [26]. We denote its coordinates by x‾i=(xi1,…,xim)\overline{x}_i=(x_i^1,\ldots,x_i^m) and write dx‾i=∏a=1mdxiad\overline{x}_i=\prod_{a=1}^m dx_i^a for integration over all replica coordinates of molecule ii. The replicated partition function on this glass branch is then written as

Zm=1N!∫∏i=1Ndx‾iexp⁡{∑a=1m[−βHa+∫dx za(x)λa(x)]}.(14)Z_m=\frac{1}{N!}\int\prod_{i=1}^{N}d\overline{x}_i\exp\left\{\sum_{a=1}^{m}\left[-\beta H_a+\int dx\,z_a(x)\lambda_a(x)\right]\right\}. \tag*{(14)}

where

Ha=∑i<jv(∣xia−xja∣),za(x)=∑i=1Nziaδ(x−xia).(15)H_a=\sum_{i<j}v\left(\left|x_i^a-x_j^a\right|\right),\qquad z_a(x)=\sum_{i=1}^{N}z_i^a\delta\left(x-x_i^a\right). \tag*{(15)}

Here, ziaz_i^a is the contact number of particle ii in replica aa. The source term can be written as a sum over pairs,

∫dx za(x)λa(x)=∑i=1Nziaλa(xia)=∑i<j[λa(xia)+λa(xja)]θ(σ−∣xia−xja∣).(16)\int dx\,z_a(x)\lambda_a(x)=\sum_{i=1}^{N}z_i^a\lambda_a(x_i^a) =\sum_{i<j}\left[\lambda_a(x_i^a)+\lambda_a(x_j^a)\right]\theta\left(\sigma-\left|x_i^a-x_j^a\right|\right). \tag*{(16)}

The Boltzmann weight thus factorizes over molecular pairs, giving

Zm=1N!∫∏i=1Ndx‾i∏i<j[1+f(x‾i,x‾j)].(17)Z_m=\frac{1}{N!}\int\prod_{i=1}^{N}d\overline{x}_i\prod_{i<j}\left[1+f(\overline{x}_i,\overline{x}_j)\right]. \tag*{(17)}

where the molecular Mayer function is defined by

f(xˉ,yˉ)=∏a=1me−βv(∣xa−ya∣)+[λa(xa)+λa(ya)]θ(σ−∣xa−ya∣)−1.(18)f(\bar{\mathbf{x}},\bar{\mathbf{y}})=\prod_{a=1}^{m}e^{-\beta v(\lvert\mathbf{x}^{a}-\mathbf{y}^{a}\rvert)+[\lambda_{a}(\mathbf{x}^{a})+\lambda_{a}(\mathbf{y}^{a})]\theta(\sigma-\lvert\mathbf{x}^{a}-\mathbf{y}^{a}\rvert)}-1. \tag*{(18)}

Following the standard procedure of Morita and Hiroike [36, 37], we obtain the replicated free energy as a functional of the molecular density ρ(xˉ)\rho(\bar{\mathbf{x}}),

−βFm=∫dxˉ ρ(xˉ)[1−log⁡ρ(xˉ)]+12∫dxˉ dyˉ ρ(xˉ)ρ(yˉ)f(xˉ,yˉ)+⋯ .(19)-\beta F_m=\int d\bar{\mathbf{x}}\,\rho(\bar{\mathbf{x}})[1-\log\rho(\bar{\mathbf{x}})]+\frac{1}{2}\int d\bar{\mathbf{x}}\,d\bar{\mathbf{y}}\,\rho(\bar{\mathbf{x}})\rho(\bar{\mathbf{y}})f(\bar{\mathbf{x}},\bar{\mathbf{y}})+\cdots. \tag*{(19)}

The higher-order terms are subleading in the high-dimensional limit [14, 26]. The first term is the ideal-gas contribution SidS_{\mathrm{id}}, and the second is the leading interaction contribution SintS_{\mathrm{int}}.

Gaussian ansatz

Tomita and Yoshino developed a replicated liquid theory in which the density and the glass order parameter vary along a single spatial direction [38]. Here we allow the density and cage variances to depend on the full molecular-center coordinate X\mathbf{X}. We use a Gaussian ansatz for the molecular density,

ρ(xˉ)=∫dX n(X)∏a=1mγAa(X)(xa−X),(20)\rho(\bar{\mathbf{x}})=\int d\mathbf{X}\,n(\mathbf{X})\prod_{a=1}^{m}\gamma_{A_a(\mathbf{X})}(\mathbf{x}^{a}-\mathbf{X}), \tag*{(20)}

where γA\gamma_A is a normalized Gaussian with variance AA per Cartesian component. The local density of molecular centers, n(X)n(\mathbf{X}), is normalized to the total number of molecules,

∫dX n(X)=N.(21)\int d\mathbf{X}\,n(\mathbf{X})=N. \tag*{(21)}

Since the source λa(x)\lambda_a(\mathbf{x}) can differ between replicas, we initially allow the cage variance Aa(X)A_a(\mathbf{X}) to depend on the replica index aa. In the zero-temperature limit, however, the saddle-point equations admit a solution with the same rescaled cage variance for all replicas at each X\mathbf{X}, as we will see below.

High-dimensional limit

Substituting the Gaussian ansatz for the density outside the logarithm gives

Sid≡∫dxˉ ρ(xˉ)[1−log⁡ρ(xˉ)]=∫dX n(X)∫dxˉ∏a=1mγAa(X)(xa−X)[1−log⁡ρ(xˉ)].(22)\begin{aligned} S_{\mathrm{id}}\equiv\int d\bar{\mathbf{x}}\,\rho(\bar{\mathbf{x}})[1-\log\rho(\bar{\mathbf{x}})] \\ &=\int d\mathbf{X}\,n(\mathbf{X})\int d\bar{\mathbf{x}}\prod_{a=1}^{m}\gamma_{A_a(\mathbf{X})}(\mathbf{x}^{a}-\mathbf{X})[1-\log\rho(\bar{\mathbf{x}})]. \tag*{(22)} \end{aligned}

The logarithm contains a separate integral over the molecular center,

log⁡ρ(xˉ)=log⁡[∫dY n(Y)∏a=1mγAa(Y)(xa−Y)].(23)\log\rho(\bar{x}) = \log\left[\int dY\, n(Y) \prod_{a=1}^{m} \gamma_{A_a(Y)}(x^a-Y)\right]. \tag*{(23)}

At large dd, the cage variances are of order d−2d^{-2} [26]. At fixed X\mathbf{X}, the outer Gaussian factors therefore localize the replica coordinates xax^a around the molecular center X\mathbf{X}. The Gaussian product inside the logarithm is appreciable when Y\mathbf{Y} is close to these replica coordinates, so the dominant contribution to the inner integral comes from Y\mathbf{Y} near X\mathbf{X}. We thus approximate n(Y)n(\mathbf{Y}) and Aa(Y)A_a(\mathbf{Y}) by n(X)n(\mathbf{X}) and Aa(X)A_a(\mathbf{X}), respectively, in the inner integral,

log⁡ρ(xˉ)≃log⁡[n(X)∫dY∏a=1mγAa(X)(xa−Y)].(24)\log\rho(\bar{x}) \simeq\log\left[n(\mathbf{X}) \int dY \prod_{a=1}^{m} \gamma_{A_a(\mathbf{X})}(x^a-Y)\right]. \tag*{(24)}

We perform the Gaussian integral over YY and average the resulting logarithm over the replica coordinates at fixed X\mathbf{X}. To express the result, we introduce the dimensionless variances A^a(X)\widehat{A}_a(\mathbf{X}) through

Aa(X)=σ2d2A^a(X).(25)A_a(\mathbf{X}) = \frac{\sigma^2}{d^2}\widehat{A}_a(\mathbf{X}). \tag*{(25)}

The ideal contribution then reads [26]

Sid≃∫dX n(X){1−log⁡n(X)+d2[(m−1)log⁡(2πeσ2d2)+∑a=1mlog⁡A^a(X)+log⁡(∑a=1m1A^a(X))]}.(26)S_{\mathrm{id}} \simeq\int d\mathbf{X}\, n(\mathbf{X}) \left\{1-\log n(\mathbf{X})+\frac{d}{2}\left[(m-1)\log\left(\frac{2\pi e\sigma^2}{d^2}\right)+\sum_{a=1}^{m}\log\widehat{A}_a(\mathbf{X})+\log\left(\sum_{a=1}^{m}\frac{1}{\widehat{A}_a(\mathbf{X})}\right)\right]\right\}. \tag*{(26)}

Substituting the Gaussian ansatz into the interaction contribution gives

Sint≡12∫dxˉ dyˉ ρ(xˉ)ρ(yˉ)f(xˉ,yˉ)=12∫dX dY n(X)n(Y)[∏a=1mQa(X,Y)−1],(27)\begin{aligned} S_{\mathrm{int}} &\equiv\frac{1}{2}\int d\bar{x}\,d\bar{y}\,\rho(\bar{x})\rho(\bar{y})f(\bar{x},\bar{y}) \\ &= \frac{1}{2}\int d\mathbf{X}\,d\mathbf{Y}\,n(\mathbf{X})n(\mathbf{Y})\left[\prod_{a=1}^{m}Q_a(\mathbf{X},\mathbf{Y})-1\right], \tag*{(27)} \end{aligned}

where

Qa(X,Y)=∫dx dy γAa(X)(x−X)γAa(Y)(y−Y)exp⁡[−βv(∣x−y∣)+[λa(x)+λa(y)]θ(σ−∣x−y∣)].(28)Q_a(\mathbf{X},\mathbf{Y})=\int d\mathbf{x}\,d\mathbf{y}\,\gamma_{A_a(\mathbf{X})}(\mathbf{x}-\mathbf{X})\gamma_{A_a(\mathbf{Y})}(\mathbf{y}-\mathbf{Y})\exp\left[-\beta v(|\mathbf{x}-\mathbf{y}|)+[\lambda_a(\mathbf{x})+\lambda_a(\mathbf{y})]\theta(\sigma-|\mathbf{x}-\mathbf{y}|)\right]. \tag*{(28)}

When the sources vary slowly on the cage scale, we evaluate λa(x)\lambda_a(\mathbf{x}) and λa(y)\lambda_a(\mathbf{y}) at the corresponding molecular centers X\mathbf{X} and Y\mathbf{Y}. We define the scaled gap and inverse temperature as

h=d(∣X−Y∣σ−1),β^=βκd2.(29)h=d\left(\frac{|\mathbf{X}-\mathbf{Y}|}{\sigma}-1\right), \qquad\widehat{\beta}=\frac{\beta\kappa}{d^2}. \tag*{(29)}

At fixed hh in the high-dimensional limit, we obtain [26, 34]

Qa(X,Y)≃∫ds γA^a(X)+A^a(Y)(s)exp⁡{[−β^2ga2+λa(X)+λa(Y)]θ(−ga)},(30)Q_a(\mathbf{X},\mathbf{Y}) \simeq\int ds\,\gamma_{\widehat{A}_a(\mathbf{X})+\widehat{A}_a(\mathbf{Y})}(s)\exp\left\{\left[-\frac{\widehat{\beta}}{2}g_a^2+\lambda_a(\mathbf{X})+\lambda_a(\mathbf{Y})\right]\theta(-g_a)\right\}, \tag*{(30)}
ga=h−s+A^a(X)+A^a(Y)2.g_a=h-s+\frac{\widehat{A}_a(\mathbf{X})+\widehat{A}_a(\mathbf{Y})}{2}.

Zero-temperature limit

We take the joint limit β^→∞\hat{\beta} \to\infty and m→0m \to0 at fixed ratio, writing

A^a(X)=mαa(X),β^=1mτ.(31)\hat{A}_a(\mathbf{X}) = m\alpha_a(\mathbf{X}), \qquad\hat{\beta} = \frac{1}{m\tau}. \tag*{(31)}

For fixed hh, the saddle point method leads to

∏a=1mQa(X,Y)≃θ(h)+θ(−h)exp⁡[Λ(X)+Λ(Y)−h22m∑a=1m1τ+αa(X)+αa(Y)].(32)\prod_{a=1}^{m} Q_a(\mathbf{X},\mathbf{Y}) \simeq\theta(h) + \theta(-h) \exp\left[\Lambda(\mathbf{X})+\Lambda(\mathbf{Y})-\frac{h^2}{2m}\sum_{a=1}^{m}\frac{1}{\tau+\alpha_a(\mathbf{X})+\alpha_a(\mathbf{Y})}\right]. \tag*{(32)}

where

Λ(X)≡∑a=1mλa(X).(33)\Lambda(\mathbf{X}) \equiv\sum_{a=1}^{m}\lambda_a(\mathbf{X}). \tag*{(33)}

The sources enter the free-energy only through their sum Λ(X)=∑a=1mλa(X)\Lambda(\mathbf{X}) = \sum_{a=1}^{m}\lambda_a(\mathbf{X}). The free-energy is thus invariant under permutations of the replica labels on the cage fields αa(X)\alpha_a(\mathbf{X}). We seek a saddle that preserves this symmetry,

αa(X)=α(X),a=1,…,m.(34)\alpha_a(\mathbf{X}) = \alpha(\mathbf{X}), \qquad a=1,\ldots,m. \tag*{(34)}

With this choice, the stationarity conditions for the individual replicas reduce to a single equation for α(X)\alpha(\mathbf{X}). At this saddle, taking m→0m \to0 in the ideal term gives

Sid=∫dX n(X)[1−log⁡n(X)−d2log⁡(2πeσ2α(X)d2)],(35)S_{\mathrm{id}} = \int dX\, n(\mathbf{X})\left[1-\log n(\mathbf{X})-\frac{d}{2}\log\left(\frac{2\pi e\sigma^2\alpha(\mathbf{X})}{d^2}\right)\right], \tag*{(35)}

and the interaction term becomes

Sint=12∫dX dY n(X)n(Y)θ(−h){exp⁡[Λ(X)+Λ(Y)−h22[τ+α(X)+α(Y)]]−1}.(36)S_{\mathrm{int}} = \frac{1}{2}\int dX\,dY\,n(\mathbf{X})n(\mathbf{Y})\theta(-h)\left\{\exp\left[\Lambda(\mathbf{X})+\Lambda(\mathbf{Y})-\frac{h^2}{2[\tau+\alpha(\mathbf{X})+\alpha(\mathbf{Y})]}\right]-1\right\}. \tag*{(36)}

Homogeneous case

Free energy

We first consider a spatially uniform source, λa(x)=λa\lambda_a(\mathbf{x})=\lambda_a. The homogeneous saddle has n(X)=nn(\mathbf{X})=n and α(X)=α\alpha(\mathbf{X})=\alpha. From Eq. (35), the ideal contribution per particle is

SidN=1−log⁡n+d2log⁡(d22πeσ2)−d2log⁡α.(37)\frac{S_{\mathrm{id}}}{N}=1-\log n+\frac{d}{2}\log\left(\frac{d^2}{2\pi e\sigma^2}\right)-\frac{d}{2}\log\alpha. \tag*{(37)}

Let vd=πd/2σd/Γ(d/2+1)v_d=\pi^{d/2}\sigma^d/\Gamma(d/2+1) be the volume of a dd-dimensional ball of radius σ\sigma. Then nvd=2dφ=dφ^nv_d=2^d\varphi=d\hat{\varphi}, where φ^=2dφ/d\hat{\varphi}=2^d\varphi/d. At the GCP, φ^=O(log⁡d)\hat{\varphi}=O(\log d) [26]. Using Stirling’s formula for vdv_d, we obtain

SidN=−d2log⁡α+d2log⁡d+O(log⁡d).(38)\frac{S_{\mathrm{id}}}{N}=-\frac{d}{2}\log\alpha+\frac{d}{2}\log d+O(\log d). \tag*{(38)}

After angular integration, the radial measure in Eq. (36) becomes dϕ^ eh dhd\hat{\phi}\,e^h\,dh in the high-dimensional limit. With Λ=∑aλa\Lambda= \sum_a \lambda_a, the interaction contribution per particle is

SintN=dϕ^2[e2ΛG(α+τ2)−1],(39)\frac{S_{\mathrm{int}}}{N} = \frac{d\hat{\phi}}{2}\left[e^{2\Lambda}G\left(\alpha+\frac{\tau}{2}\right)-1\right], \tag*{(39)}

where

G(x)=∫−∞0dh eh−h2/(4x)=πxexerfc⁡(x).(40)G(x) = \int_{-\infty}^{0} dh\,e^{h-h^2/(4x)} = \sqrt{\pi}x e^x \operatorname{erfc}(\sqrt{x}). \tag*{(40)}

For small xx, this function has the expansion

G(x)=πx−2x+O(x3/2).(41)G(x) = \sqrt{\pi x} - 2x + O(x^{3/2}). \tag*{(41)}

Cage size

The cage size is calculated by minimizing the free-energy

s≡Sid+SintN.(42)s \equiv\frac{S_{\mathrm{id}}+S_{\mathrm{int}}}{N}. \tag*{(42)}

The cage size is determined by the saddle-point condition:

0=∂s∂α=−d2α+dϕ^2e2ΛG′(α+τ2),(43)0 = \frac{\partial s}{\partial\alpha} = -\frac{d}{2\alpha}+\frac{d\hat{\phi}}{2}e^{2\Lambda}G'\left(\alpha+\frac{\tau}{2}\right), \tag*{(43)}

which gives

1ϕ^=e2ΛαG′(α+τ2).(44)\frac{1}{\hat{\phi}} = e^{2\Lambda}\alpha G'\left(\alpha+\frac{\tau}{2}\right). \tag*{(44)}

We introduce the dimensionless ratio

ε≡τ2α.(45)\varepsilon\equiv\frac{\tau}{2\alpha}. \tag*{(45)}

For weak compression along the ideal-glass branch in the high-dimensional limit, ε\varepsilon increases linearly with compression above GCP and measures the relative excess coordination, as we will show in the following subsections. Using G′(x)≃π/(2x)G'(x)\simeq\sqrt{\pi}/(2\sqrt{x}) in Eq. (44), we find

α≃4(1+ε)e−4Λπϕ^2.(46)\alpha\simeq\frac{4(1+\varepsilon)e^{-4\Lambda}}{\pi\hat{\phi}^2}. \tag*{(46)}

Thus, at fixed ε\varepsilon and Λ\Lambda, the cage size decreases as α∼ϕ^−2∼(log⁡d)−2\alpha\sim\hat{\phi}^{-2}\sim(\log d)^{-2}, which validates the small-cage expansion at large $d.

Complexity

The zero-complexity condition selects the ideal-glass branch. Together with the saddle-point equation, it determines τ\tau and hence ε\varepsilon as functions of ϕ^\hat{\phi}, as we will see below. At zero source, the complexity per particle is

Σ=sm−m∂sm∂m,(47)\Sigma= s_m - m\frac{\partial s_m}{\partial m}, \tag*{(47)}

where sms_m is the replicated entropy per particle at finite mm [25, 26]. Writing T^=mτ\hat{T}=m\tau and taking the zero-temperature limit gives [27]

Σ(τ)=s(τ)+τ∂s∂τ.(48)\Sigma(\tau)=s(\tau)+\tau\frac{\partial s}{\partial\tau}. \tag*{(48)}

To reach the jamming endpoint, we first take T^→0\hat{T}\to0 at fixed m>0m>0. This suppresses particle overlaps and reduces the model to hard spheres. We then take m→0m\to0 along the hard-sphere glass branch, reaching its infinite-pressure jamming endpoint [26]. This order of limits corresponds to τ=T^/m=0\tau=\hat{T}/m=0, for which Σ(0)=s(0)\Sigma(0)=s(0). In the density regime ϕ^=O(log⁡d)\hat{\phi}=O(\log d), the leading terms are

Σ(0)=d2log⁡d−dϕ^2+O(dlog⁡log⁡d).(49)\Sigma(0)=\frac{d}{2}\log d-\frac{d\hat{\phi}}{2}+O(d\log\log d). \tag*{(49)}

Setting Σ(0)=0\Sigma(0)=0 gives the GCP density,

ϕ^JGCP=log⁡d+O(log⁡log⁡d),(50)\hat{\phi}^{\mathrm{GCP}}_J=\log d+O(\log\log d), \tag*{(50)}

consistent with Ref. [26].

To determine how ε\varepsilon changes upon compression above GCP, we expand the complexity at small τ\tau. The stationarity of ss with respect to α\alpha gives

∂s∂τ∣ϕ^=dϕ^4G′(α+τ2)=d4α,(51)\left.\frac{\partial s}{\partial\tau}\right|_{\hat{\phi}}=\frac{d\hat{\phi}}{4}G'\left(\alpha+\frac{\tau}{2}\right)=\frac{d}{4\alpha}, \tag*{(51)}

where the last equality follows from Eq. (44) at zero source. With α0=α(τ=0)\alpha_0=\alpha(\tau=0), we obtain

Σ(τ)−Σ(0)=τd2α0+O(d(τα0)2)=dε+O(dε2).(52)\begin{aligned} \Sigma(\tau)-\Sigma(0)&=\frac{\tau d}{2\alpha_0}+O\left(d\left(\frac{\tau}{\alpha_0}\right)^2\right)\\ &=d\varepsilon+O(d\varepsilon^2). \tag*{(52)} \end{aligned}

At τ=0\tau=0, differentiation with respect to density gives

dΣ(0)dϕ^=−1ϕ^+d2[G(α0)−1]=−d2+O(dϕ^).(53)\frac{d\Sigma(0)}{d\hat{\phi}}=-\frac{1}{\hat{\phi}}+\frac{d}{2}[G(\alpha_0)-1]=-\frac{d}{2}+O\left(\frac{d}{\hat{\phi}}\right). \tag*{(53)}

Expanding about ϕ^JGCP\hat{\phi}^{\mathrm{GCP}}_J and imposing Σ(τ)=0\Sigma(\tau)=0, we find

ε≃ϕ^−ϕ^JGCP2(54)\varepsilon\simeq\frac{\hat{\phi}-\hat{\phi}^{\mathrm{GCP}}_J}{2} \tag*{(54)}

for weak compression above GCP.

Gardner transition

At zero source, we locate the Gardner transition of the 1RSB solution using the condition derived by Biroli and Urbani [34],

1=ϕ^2e−A^∫−∞∞dh ehQ(h)m[2A^d2dh2log⁡Q(h)]2,(55)1 = \frac{\hat{\phi}}{2}e^{-\hat{A}}\int_{-\infty}^{\infty} dh\,e^h Q(h)^m \left[2\hat{A}\frac{d^2}{dh^2}\log Q(h)\right]^2, \tag*{(55)}

where

Q(h)=∫dz γ2A^(z)exp⁡[−β^2(h−z)2θ(z−h)].(56)Q(h) = \int dz\,\gamma_{2\hat{A}}(z)\exp\left[-\frac{\hat{\beta}}{2}(h-z)^2\theta(z-h)\right]. \tag*{(56)}

We set A^=mα\hat{A}=m\alpha and β^=1/(mτ)\hat{\beta}=1/(m\tau). In the limit m→0m\to0 at fixed τ>0\tau>0, we obtain, for h≠0h\ne0,

Q(h)m⟶exp⁡[−h24(α+τ/2)]θ(−h)+θ(h),2A^d2dh2log⁡Q(h)⟶−αα+τ/2θ(−h).(57)Q(h)^m \longrightarrow\exp\left[-\frac{h^2}{4(\alpha+\tau/2)}\right]\theta(-h)+\theta(h),\qquad2\hat{A}\frac{d^2}{dh^2}\log Q(h)\longrightarrow-\frac{\alpha}{\alpha+\tau/2}\theta(-h). \tag*{(57)}

Substituting these limits into Eq. (55) gives

1=ϕ^2α2(α+τ/2)2G(α+τ2).(58)1=\frac{\hat{\phi}}{2}\frac{\alpha^2}{(\alpha+\tau/2)^2}G\left(\alpha+\frac{\tau}{2}\right). \tag*{(58)}

Using the zero-source saddle-point condition (44), we eliminate ϕ^\hat{\phi} to obtain

1=α2(α+τ/2)2G(α+τ/2)G′(α+τ/2).(59)1=\frac{\alpha}{2(\alpha+\tau/2)^2}\frac{G(\alpha+\tau/2)}{G'(\alpha+\tau/2)}. \tag*{(59)}

For small xx, the ratio in this equation has the expansion

G(x)G′(x)=2x[1+2xπ+O(x)].(60)\frac{G(x)}{G'(x)}=2x\left[1+2\sqrt{\frac{x}{\pi}}+O(x)\right]. \tag*{(60)}

Using α+τ/2=α(1+ε)\alpha+\tau/2=\alpha(1+\varepsilon), we find that the Gardner boundary occurs at

εG=2αGπ+O(αG),(61)\varepsilon_{\mathrm{G}}=2\sqrt{\frac{\alpha_{\mathrm{G}}}{\pi}}+O(\alpha_{\mathrm{G}}), \tag*{(61)}

where αG\alpha_{\mathrm{G}} is the cage size at the transition. Near GCP, αG=O((log⁡d)−2)\alpha_{\mathrm{G}}=O((\log d)^{-2}), so the width of the unstable region is

εG=O(1log⁡d).(62)\varepsilon_{\mathrm{G}}=O\left(\frac{1}{\log d}\right). \tag*{(62)}

The Gardner boundary thus approaches jamming as dd increases. At fixed ε>0\varepsilon>0 near GCP, the 1RSB solution is stable against the Gardner instability in the limit d→∞d\to\infty. The jamming endpoint ε=0\varepsilon=0 is marginal in this limit [30].

Schematic stability diagram of the 1RSB solution for harmonic spheres near GCP

Figure 1. Schematic stability diagram of the 1RSB solution for harmonic spheres near GCP. The parameter ε=τ/(2α)\varepsilon=\tau/(2\alpha) measures the relative excess coordination (zˉ−2d)/(2d)(\bar{z}-2d)/(2d) to leading order at large dd. (a) At large but finite dd, the 1RSB solution is unstable for 0<ε<εG0<\varepsilon<\varepsilon_G, with εG=O(1/log⁡d)\varepsilon_G=O(1/\log d). (b) As d→∞d\to\infty, the Gardner boundary approaches the marginal jamming endpoint ε=0\varepsilon=0, and the 1RSB solution is stable at fixed ε>0\varepsilon>0 near GCP.

Mean contact number and its fluctuations

The uniform source λa\lambda_a couples to the sum of the particle contact numbers in replica aa:

∫dx za(x)λa(x)=λaZa,(63)\int d\mathbf{x}\,z_a(\mathbf{x})\lambda_a(\mathbf{x})=\lambda_a Z_a, \tag*{(63)}

where

Za=∑i=1Nzia.(64)Z^a=\sum_{i=1}^{N}z_i^a. \tag*{(64)}

We obtain the mean contact number per particle by differentiating the free-energy ss with respect to λa\lambda_a. Since ss is stationary with respect to α\alpha, only the explicit source dependence contributes to the first derivative,

zˉa=1N⟨Za⟩=∂s∂λa=dϕ^e2ΛG(α+τ2).(65)\bar{z}_a=\frac{1}{N}\langle Z^a\rangle=\frac{\partial s}{\partial\lambda_a}=d\hat{\phi}e^{2\Lambda}G\left(\alpha+\frac{\tau}{2}\right). \tag*{(65)}

Using Eq. (44), we can also write

zˉa=dαG(α+τ/2)G′(α+τ/2).(66)\bar{z}_a=\frac{d}{\alpha}\frac{G\left(\alpha+\tau/2\right)}{G'\left(\alpha+\tau/2\right)}. \tag*{(66)}

The leading small-cage result is

zˉ≃2d(1+τ2α)=2d(1+ε).(67)\bar{z}\simeq2d\left(1+\frac{\tau}{2\alpha}\right)=2d(1+\varepsilon). \tag*{(67)}

Thus, near GCP in the high-dimensional limit, ε\varepsilon measures the relative excess coordination

ε≃zˉ−2d2d.(68)\varepsilon\simeq\frac{\bar{z}-2d}{2d}. \tag*{(68)}

At ε=0\varepsilon=0, the limiting mean contact number reaches the isostatic value 2d2d. Combining Eq. (67) with the zero-complexity condition gives a linear increase upon weak compression in the high-dimensional 1RSB regime,

zˉ−2d≃d(ϕ^−ϕ^JGCP).(69)\bar{z}-2d\simeq d\left(\hat{\phi}-\hat{\phi}^{\mathrm{GCP}}_J\right). \tag*{(69)}

In finite-dimensional simulations of harmonic spheres, the excess coordination grows as zˉ−2d∝(φ−φJ)1/2\bar{z}-2d \propto(\varphi-\varphi_J)^{1/2} near jamming [4]. The difference between these scaling laws reflects the replica structure governing the approach to jamming. At GCP, where φ^JGCP∼log⁡d\hat{\varphi}_J^{\mathrm{GCP}} \sim\log d, the Gardner-unstable region shrinks as d→∞d \to\infty, allowing jamming to be approached along a stable 1RSB branch. At finite dd, a glass can undergo a Gardner transition before reaching jamming. This can also occur in the limit d→∞d \to\infty when the jamming density φ^J\hat{\varphi}_J remains finite [34, 35]. In either case, full replica symmetry breaking governs the asymptotic scaling near jamming [9, 10].

Further derivative with respect to λa\lambda_a gives the contact-number susceptibility,

χz=1NVar⁡(Za)=∂zˉa∂λa.(70)\chi_z = \frac{1}{N}\operatorname{Var}(Z^a) = \frac{\partial\bar{z}_a}{\partial\lambda_a}. \tag*{(70)}

Using the leading small-cage expression zˉa≃2d(1+ε)\bar{z}_a \simeq2d(1+\varepsilon) and differentiating at fixed φ^\hat{\varphi} and τ\tau, we obtain

χz≃2d∂ε∂λa=−2dε∂log⁡α∂λa.(71)\chi_z \simeq2d\frac{\partial\varepsilon}{\partial\lambda_a} = -2d\varepsilon\frac{\partial\log\alpha}{\partial\lambda_a}. \tag*{(71)}

Taking the logarithm of the small-cage form of Eq. (44) and differentiating with respect to λa\lambda_a gives

0≃2+[1−12(1+ε)]∂log⁡α∂λa.(72)0 \simeq2+\left[1-\frac{1}{2(1+\varepsilon)}\right]\frac{\partial\log\alpha}{\partial\lambda_a}. \tag*{(72)}

Substituting this result into Eq. (71) yields

χz≃8dε(1+ε)1+2ε.(73)\chi_z \simeq\frac{8d\varepsilon(1+\varepsilon)}{1+2\varepsilon}. \tag*{(73)}

Figure 2 shows this result. As jamming is approached, χz≃8dε≃4δz\chi_z \simeq8d\varepsilon\simeq4\delta z, where δz≡zˉ−2d\delta z \equiv\bar{z}-2d denotes the excess coordination. The 1RSB calculation thus gives

Plot of contact-number susceptibility $\chi_z/d$ as a function of $\varepsilon$

Figure 2. Contact-number susceptibility χz=N−1Var⁡(Za)\chi_z = N^{-1}\operatorname{Var}(Z^a) as a function of ε=τ/(2α)\varepsilon= \tau/(2\alpha). The parameter ε=τ/(2α)\varepsilon= \tau/(2\alpha) measures the relative excess coordination (zˉ−2d)/(2d)(\bar{z}-2d)/(2d) to leading order at large dd. The solid line shows the leading 1RSB result and the dashed line shows its small-ε\varepsilon asymptote, χz≃8dε\chi_z \simeq8d\varepsilon.

χz∝(δz)β,β=1.(74)\chi_z \propto(\delta z)^\beta,\qquad\beta=1. \tag*{(74)}

Numerical simulations in three dimensions suggest β≃1.2\beta\simeq1.2–1.251.25 [18, 39, 40]. Whether extending our calculation to the fullRSB regime yields an exponent closer to these estimates remains an open question.

Inhomogeneous case

In this section, we extend the calculation to a spatially varying source to determine how the contact-number fluctuations depend on wave number.

Free energy

We apply a spatially varying source with zero spatial average and expand Eqs. (35) and (36) about the homogeneous saddle at zero source. We write the density and cage fields as

n(X)=n[1+r(X)],α(X)=α[1+a(X)].(75)n(\mathbf{X}) = n[1+r(\mathbf{X})],\qquad\alpha(\mathbf{X}) = \alpha[1+a(\mathbf{X})]. \tag*{(75)}

For a source at nonzero wave number, the linear responses rr and aa have zero spatial average. Near GCP, α=O((log⁡d)−2)\alpha= O((\log d)^{-2}) and, at fixed ε=τ/(2α)\varepsilon= \tau/(2\alpha), τ\tau is of the same order. The Gaussian in the interaction term is therefore confined to a narrow shell on the inside of the contact surface ∣X−Y∣=σ|\mathbf{X}-\mathbf{Y}|=\sigma. To leading order in the shell width [41],

exp⁡[−h22[τ+α(X)+α(Y)]]θ(−h)≃π2[τ+α(X)+α(Y)]σdδ(∣X−Y∣−σ).(76)\exp\left[-\frac{h^2}{2[\tau+\alpha(\mathbf{X})+\alpha(\mathbf{Y})]}\right]\theta(-h) \simeq\sqrt{\frac{\pi}{2}[\tau+\alpha(\mathbf{X})+\alpha(\mathbf{Y})]}\frac{\sigma}{d}\delta(|\mathbf{X}-\mathbf{Y}|-\sigma). \tag*{(76)}

The square-root factor is the integrated weight of the Gaussian over h<0h<0, and σ/d\sigma/d converts the delta function in hh to one in the radial distance.

We use ⟨f⟩V=V−1∫dX f(X)\langle f\rangle_V=V^{-1}\int d\mathbf{X}\,f(\mathbf{X}) for spatial averages. For functions of two positions, we define averages over the surface and interior of the contact sphere,

⟨f⟩sd=1Vsd∫dX dY δ(∣X−Y∣−σ)f(X,Y),⟨f⟩vd=1Vvd∫dX dY θ(σ−∣X−Y∣)f(X,Y),(77)\begin{aligned} \langle f\rangle_{s_d} &= \frac{1}{V s_d}\int d\mathbf{X}\,d\mathbf{Y}\,\delta(|\mathbf{X}-\mathbf{Y}|-\sigma)f(\mathbf{X},\mathbf{Y}),\\ \langle f\rangle_{v_d} &= \frac{1}{V v_d}\int d\mathbf{X}\,d\mathbf{Y}\,\theta(\sigma-|\mathbf{X}-\mathbf{Y}|)f(\mathbf{X},\mathbf{Y}), \tag*{(77)} \end{aligned}

where sd=dvd/σs_d=dv_d/\sigma is the surface area of a sphere of radius σ\sigma. With rX=r(X)r_X=r(\mathbf{X}), aX=a(X)a_X=a(\mathbf{X}), and ΛX=Λ(X)\Lambda_X=\Lambda(\mathbf{X}), the ideal and interaction contributions are, up to additive constants,

SidN=−⟨(1+rX)log⁡(1+rX)⟩V−d2⟨(1+rX)log⁡(1+aX)⟩V,(78)\frac{S_{\mathrm{id}}}{N}=-\langle(1+r_X)\log(1+r_X)\rangle_V-\frac{d}{2}\langle(1+r_X)\log(1+a_X)\rangle_V, \tag*{(78)}
SintN≃zˉ2⟨(1+rX)(1+rY)eΛX+ΛY1+aX+aY2(1+ε)⟩sd−dφ^2⟨rXrY⟩vd.(79)\frac{S_{\mathrm{int}}}{N}\simeq\frac{\bar{z}}{2}\left\langle(1+r_X)(1+r_Y)e^{\Lambda_X+\Lambda_Y}\sqrt{1+\frac{a_X+a_Y}{2(1+\varepsilon)}}\right\rangle_{s_d}-\frac{d\hat{\varphi}}{2}\langle r_Xr_Y\rangle_{v_d}. \tag*{(79)}

Here zˉ≃dφ^πα(1+ε)≃2d(1+ε)\bar{z}\simeq d\hat{\varphi}\sqrt{\pi\alpha(1+\varepsilon)}\simeq2d(1+\varepsilon) is the mean contact number of the homogeneous saddle to leading order.

For a modulation at fixed nonzero scaled wave number k=σ∣q∣/dk=\sigma|\mathbf{q}|/\sqrt{d}, the excluded-volume term −dϕ^⟨rXrY⟩vd/2-d\hat{\phi}\langle r_Xr_Y\rangle_{v_d}/2 gives a density stiffness of order dϕ^=O(dlog⁡d)d\hat{\phi}=O(d\log d) near GCP. The couplings of the density mode to the cage modulation and the contact-number source are both of order dd. The density saddle point therefore gives

rX=O(aXlog⁡d,ΛXlog⁡d)(80)r_X=O\left(\frac{a_X}{\log d},\frac{\Lambda_X}{\log d}\right) \tag*{(80)}

so that we can set rX=0r_X=0 when computing the leading-order contact-number response. Setting rX=0r_X=0 and expanding to second order in the cage modulation aXa_X and the source ΛX\Lambda_X, we obtain

Sid(2)N=d4⟨aX2⟩vd,\frac{S_{\mathrm{id}}^{(2)}}{N}=\frac{d}{4}\langle a_X^2\rangle_{v_d},
Sint(2)N=d(1+ε)2⟨(ΛX+ΛY)2⟩sd−d32(1+ε)⟨(aX+aY)2⟩sd+d4⟨(aX+aY)(ΛX+ΛY)⟩sd.(81)\frac{S_{\mathrm{int}}^{(2)}}{N}=\frac{d(1+\varepsilon)}{2}\langle(\Lambda_X+\Lambda_Y)^2\rangle_{s_d}-\frac{d}{32(1+\varepsilon)}\langle(a_X+a_Y)^2\rangle_{s_d}+\frac{d}{4}\langle(a_X+a_Y)(\Lambda_X+\Lambda_Y)\rangle_{s_d}. \tag*{(81)}

With the Fourier convention introduced above, the surface average of two fields ff and gg takes the form

⟨(fX+fY)(gX+gY)⟩sd=2(2π)dV∫dq (1+Kq)fqg−q,(82)\langle(f_X+f_Y)(g_X+g_Y)\rangle_{s_d}=\frac{2}{(2\pi)^dV}\int d\mathbf{q}\,(1+K_q)f_qg_{-\mathbf{q}}, \tag*{(82)}

where

Kq=1sd∫dx eiq⋅xδ(∣x∣−σ)(83)K_q=\frac{1}{s_d}\int d\mathbf{x}\,e^{i\mathbf{q}\cdot\mathbf{x}}\delta(|\mathbf{x}|-\sigma) \tag*{(83)}

is the Fourier transform of the normalized surface measure on the contact sphere. In the limit d→∞d\to\infty, it converges to

Kq→d→∞exp⁡[−k22],k=σqd.(84)K_q\xrightarrow{d\to\infty}\exp\left[-\frac{k^2}{2}\right],\qquad k=\frac{\sigma q}{\sqrt{d}}. \tag*{(84)}

The quadratic contribution to −βFm-\beta F_m is then

(−βFm)(2)=N(2π)dV∫dq[Aqaqa−q+BqaqΛ−q+CqΛqΛ−q],(85)(-\beta F_m)^{(2)}=\frac{N}{(2\pi)^dV}\int d\mathbf{q}\left[A_qa_qa_{-\mathbf{q}}+B_qa_q\Lambda_{-\mathbf{q}}+C_q\Lambda_q\Lambda_{-\mathbf{q}}\right], \tag*{(85)}

with

Aq=d4−d(1+Kq)16(1+ε),A_q=\frac{d}{4}-\frac{d(1+K_q)}{16(1+\varepsilon)},
Bq=d2(1+Kq),B_q=\frac{d}{2}(1+K_q),
Cq=d(1+ε)(1+Kq).(86)C_q=d(1+\varepsilon)(1+K_q). \tag*{(86)}

Figure 3. Contact-number structure factor Sz(q)/dS_z(q)/d as a function of the scaled wave number k=σq/dk = \sigma q/\sqrt{d} for several values of ε=τ/(2α)\varepsilon= \tau/(2\alpha). The parameter ε=τ/(2α)\varepsilon= \tau/(2\alpha) measures the relative excess coordination (zˉ−2d)/(2d)(\bar{z}-2d)/(2d) to leading order at large dd. The small-kk plateau decreases as ε\varepsilon approaches zero. At GCP (ε=0\varepsilon= 0), the structure factor vanishes quadratically at small kk.

Contact-number structure factor

Stationarity with respect to a−qa_{-q} gives the cage response,

aq=−Bq2AqΛq.(87)a_q=-\frac{B_q}{2A_q}\Lambda_q. \tag*{(87)}

Substituting this response into the quadratic free energy gives

(−βFm,∗)(2)=n(2π)d∫dq(Cq−Bq24Aq)ΛqΛ−q,(88)(-\beta F_{m,*})^{(2)}=\frac{n}{(2\pi)^d}\int dq\left(C_q-\frac{B_q^2}{4A_q}\right)\Lambda_q\Lambda_{-q}, \tag*{(88)}

where Fm,∗F_{m,*} denotes the value at the saddle. Since each replica source enters through Λ\Lambda, differentiating twice with respect to the source in one replica yields the connected correlation function

⟨δzqδzq′⟩≃(2π)dn(2Cq−Bq22Aq)δ(q+q′).(89)\langle\delta z_q\delta z_{q'}\rangle\simeq(2\pi)^d n\left(2C_q-\frac{B_q^2}{2A_q}\right)\delta(q+q'). \tag*{(89)}

In a periodic volume VV, the Dirac delta is replaced by V/(2π)dV/(2\pi)^d times the corresponding Kronecker delta. Dividing the covariance at q′=−qq'=-q by NN therefore gives

Sz(q)≃2Cq−Bq22Aq=4d(1+ε)(1+Kq)(1+2ε−Kq)3+4ε−Kq.(90)\begin{aligned} S_z(q)\simeq2C_q-\frac{B_q^2}{2A_q} \\ &=4d(1+\varepsilon)\frac{(1+K_q)(1+2\varepsilon-K_q)}{3+4\varepsilon-K_q}. \tag*{(90)} \end{aligned}

For fixed k=σq/dk=\sigma q/\sqrt{d} in the high-dimensional limit, using Eq. (84), we get

Sz(k)=4d(1+ε)(1+e−k2/2)(1+2ε−e−k2/2)3+4ε−e−k2/2.(91)S_z(k)=4d(1+\varepsilon)\frac{(1+e^{-k^2/2})(1+2\varepsilon-e^{-k^2/2})}{3+4\varepsilon-e^{-k^2/2}}. \tag*{(91)}

Figure 3 shows the resulting structure factor as a function of kk. The plateau at small kk decreases as jamming is approached. Taking k→0k \to0 in the limiting spectrum gives

lim⁡k→0Sz(k)=χz.(92)\lim_{k \to0} S_z(k) = \chi_z. \tag*{(92)}

At GCP, where χz=0\chi_z = 0, the plateau vanishes and the leading small-kk behavior is

Sz(k)/d=2k2+O(k4).(93)S_z(k)/d = 2k^2 + O(k^4). \tag*{(93)}

The quadratic decay establishes contact-number hyperuniformity at GCP in the high-dimensional limit. The exponent 2 is larger than the value of about 1.5 reported by Hexner et al. [17]. Whether full replica symmetry breaking changes the wave-number exponent in the marginal regime remains an open question.

Summary and discussion

We derived the replicated free-energy functional of harmonic spheres near glass close packing (GCP) in the high-dimensional limit within the 1RSB ansatz. By coupling a source to the contact field, we obtained both the uniform susceptibility and the wave-number-dependent structure factor. For a uniform source, the mean contact number zˉ\bar{z} reaches the isostatic value 2d2d at GCP and increases linearly with the excess packing fraction ϕ^−ϕ^JGCP\hat{\phi} - \hat{\phi}_{J}^{\mathrm{GCP}} upon compression. The variance of the total contact number per particle, χz\chi_z, vanishes at the transition point as χz∝δzβ\chi_z \propto\delta z^\beta with β=1\beta= 1. For a spatially varying source, we obtained the contact-number structure factor. Taking the high-dimensional limit at fixed scaled wave number k=σq/dk = \sigma q/\sqrt{d} gives

Sz(k)d≃4δzd+2k2,δzd≪1,k≪1.(94)\frac{S_z(k)}{d} \simeq\frac{4\delta z}{d} + 2k^2, \qquad\frac{\delta z}{d} \ll1,\quad k \ll1. \tag*{(94)}

At GCP, δz=0\delta z = 0 and the contact number exhibits hyperuniformity Sz(k)∝kαS_z(k) \propto k^\alpha with α=2\alpha= 2. Above GCP, δz>0\delta z > 0, and thus, hyperuniformity is not observed.

Our results β=1\beta= 1 and α=2\alpha= 2 deviate from those of three-dimensional simulations, which reported α≃1.5\alpha\simeq1.5 [17] and β≃1.2\beta\simeq1.2–1.251.25 [18, 39, 40]. The difference may be related to the stability of the glass on the approach to jamming. At high-dimensional GCP, the width of the Gardner-unstable region scales as 1/log⁡d1/\log d. In the limit d→∞d \to\infty, this region shrinks to the jamming point, which can then be approached along a stable 1RSB branch and is marginal at its endpoint. A direction for future work is to extend the current calculation to the fullRSB regime and determine whether the hierarchy of glass subbasins yields values of α\alpha and β\beta closer to the numerical results in finite dimension.

Wang et al. found α≃2\alpha\simeq2 in jammed harmonic-sphere packings with a tuned particle-size distribution [23], in agreement with our 1RSB prediction. They also found enhanced vibrational stability in glasses prepared by particle resizing well above jamming, where low-frequency quasilocalized modes were strongly suppressed. This enhanced stability raises the possibility that these glasses approach jamming along a stable 1RSB branch, which could explain the agreement in α\alpha. Further work is needed to elucidate this point.

Acknowledgments

This project has received JSPS KAKENHI Grant Numbers 23K13031, and 25H01401.

AI disclosure. I used ChatGPT–6.1 (OpenAI) to assist with drafting and editing the manuscript, and checking intermediate steps in the derivations. I checked the derivations against my calculations, and reviewed and revised the AI-assisted text. I take full responsibility for the content of the manuscript.

Data availability

This study is based on analytical calculations. All quantitative plots can be reproduced from the expressions given in the article.

Comparison of the two contact-number structure factors

We compare the two contact-number structure factors Sz(q)S_z(\mathbf{q}) and Sz′(q)S'_z(\mathbf{q}) introduced in Sec. II. Throughout this appendix, we take d→∞d \to\infty at fixed nonzero scaled wave number k=σ∣q∣/dk = \sigma|\mathbf{q}|/\sqrt{d}.

The density stiffness derived in § 2.4 is of order dϕ^d\hat{\phi}. Near GCP, where ϕ^∼log⁡d\hat{\phi} \sim\log d, its inverse gives

Sρ(q)≡1N⟨ρqρ−q⟩=O(1dlog⁡d).S_\rho(\mathbf{q}) \equiv\frac{1}{N}\langle\rho_{\mathbf{q}}\rho_{-\mathbf{q}}\rangle= O\left(\frac{1}{d\log d}\right).

The same order is obtained by evaluating the high-dimensional hard-sphere liquid result, Sliq(k)≃[1+dϕ^e−k2/2]−1S_{\mathrm{liq}}(k) \simeq[1+d\hat{\phi}e^{-k^2/2}]^{-1}, at the GCP density scale [42].

For q≠0\mathbf{q} \ne0, translational invariance gives ⟨zq⟩=0\langle z_{\mathbf{q}}\rangle= 0. The first term in Eq. (7) is therefore

Sz(q)=1N⟨zqz−q⟩=O(d),S_z(\mathbf{q}) = \frac{1}{N}\langle z_{\mathbf{q}}z_{-\mathbf{q}}\rangle= O(d),

where the scaling follows from the calculation in the main text. Since zˉ=O(d)\bar{z} = O(d), the density term is

zˉ2Sρ(q)=O(dlog⁡d).\bar{z}^{2}S_\rho(\mathbf{q}) = O\left(\frac{d}{\log d}\right).

The Cauchy–Schwarz inequality bounds the mixed term as

zˉN∣⟨ρqz−q⟩∣≤zˉSρ(q)Sz(q)=O(dlog⁡d).\frac{\bar{z}}{N}|\langle\rho_{\mathbf{q}}z_{-\mathbf{q}}\rangle| \le\bar{z}\sqrt{S_\rho(\mathbf{q})S_z(\mathbf{q})} = O\left(\frac{d}{\sqrt{\log d}}\right).

Thus, at fixed kk, the density-dependent terms are subleading, and Sz(q)S_z(\mathbf{q}) and Sz′(q)S'_z(\mathbf{q}) agree to leading order as d→∞d \to\infty.

References

  1. [5]M. Wyart, L. E. Silbert, S. R. Nagel, and T. A. Witten, Phys. Rev. E 72, 051306 (2005).
  2. [6]C. P. Goodrich, A. J. Liu, and S. R. Nagel, Phys. Rev. Lett. 109, 095704 (2012).
  3. [7]E. DeGiuli, A. Laversanne-Finot, G. Düring, E. Lerner, and M. Wyart, Soft Matter 10, 5628 (2014).
  4. [8]E. DeGiuli, E. Lerner, C. Brito, and M. Wyart, Proceedings of the National Academy of Sciences 111, 17054–17059 (2014).
  5. [9]P. Charbonneau, J. Kurchan, G. Parisi, P. Urbani, and F. Zamponi, Nature Communications 5 (2014), 10.1038/ncomms4725.
  6. [10]P. Charbonneau, J. Kurchan, G. Parisi, P. Urbani, and F. Zamponi, Journal of Statistical Mechanics: Theory and Experiment 2014, P10009 (2014).
  7. [11]P. Charbonneau, E. I. Corwin, G. Parisi, and F. Zamponi, Phys. Rev. Lett. 114, 125504 (2015).
  8. [12]L. Yan, E. DeGiuli, and M. Wyart, EPL (Europhysics Letters) 114, 26003 (2016).
  9. [13]P. Charbonneau, J. Kurchan, G. Parisi, P. Urbani, and F. Zamponi, Annual Review of Condensed Matter Physics 8, 265–288 (2017).
  10. [14]G. Parisi, P. Urbani, and F. Zamponi, Theory of Simple Glasses: Exact Solutions in Infinite Dimensions (Cambridge University Press, 2019).
  11. [15]P. Charbonneau, E. I. Corwin, R. C. Dennis, R. Díaz Hernández Rojas, H. Ikeda, G. Parisi, and F. Ricci-Tersenghi, Phys. Rev. E 104, 014102 (2021).
  12. [16]H. Ikeda and M. Shimada, Phys. Rev. E 106, 024904 (2022).
  13. [17]D. Hexner, A. J. Liu, and S. R. Nagel, Phys. Rev. Lett. 121, 115501 (2018).
  14. [18]D. Hexner, P. Urbani, and F. Zamponi, Phys. Rev. Lett. 123, 068003 (2019).
  15. [19]S. Torquato and F. H. Stillinger, Phys. Rev. E 68, 041113 (2003).
  16. [20]S. Torquato, Physics Reports 745, 1–95 (2018).
  17. [21]S. Franz, A. Sclocchi, and P. Urbani, SciPost Physics 9 (2020), 10.21468/scipostphys.9.1.012.
  18. [22]J. Zhang, J. Si, N. Xu, and H. Tong, Phys. Rev. Lett. 137, 078201 (2026).
  19. [23]Y. Wang, Z. Qian, H. Tong, and H. Tanaka, Nature Communications 16 (2025), 10.1038/s41467-025-56283-1.
  20. [24]J. Shang, Y. Wang, D. Pan, Y. Jin, and J. Zhang, Proceedings of the National Academy of Sciences 123 (2026), 10.1073/pnas.2517241123.
  21. [25]R. Monasson, Phys. Rev. Lett. 75, 2847 (1995).
  22. [26]G. Parisi and F. Zamponi, Rev. Mod. Phys. 82, 789 (2010).
  23. [27]L. Berthier, H. Jacquin, and F. Zamponi, Phys. Rev. E 84, 051103 (2011).
  24. [28]J. Kurchan, G. Parisi, and F. Zamponi, Journal of Statistical Mechanics: Theory and Experiment 2012, P10012 (2012).
  25. [29]E. Gardner, Nuclear Physics B 257, 747–765 (1985).
  26. [30]J. Kurchan, G. Parisi, P. Urbani, and F. Zamponi, The Journal of Physical Chemistry B 117, 12979–12994 (2013).
  27. [31]P. Charbonneau, Y. Jin, G. Parisi, C. Rainone, B. Seoane, and F. Zamponi, Phys. Rev. E 92, 012316 (2015).
  28. [32]L. Berthier, P. Charbonneau, Y. Jin, G. Parisi, B. Seoane, and F. Zamponi, Proceedings of the National Academy of Sciences 113, 8397–8401 (2016).
  29. [33]A. Seguin and O. Dauchot, Phys. Rev. Lett. 117, 228001 (2016).
  30. [34]G. Biroli and P. Urbani, SciPost Physics 4 (2018), 10.21468/scipostphys.4.4.020.
  31. [35]C. Scalliet, L. Berthier, and F. Zamponi, Phys. Rev. E 99, 012107 (2019).
  32. [36]T. Morita and K. Hiroike, Progress of Theoretical Physics 23, 1003–1027 (1960).
  33. [37]J.-P. Hansen and I. R. McDonald, Theory of simple liquids: with applications to soft matter (Academic press, 2013).
  34. [38]Y. Tomita and H. Yoshino, The Journal of Chemical Physics 163 (2025), 10.1063/5.0300245.
  35. [39]J. A. Giannini, E. Lerner, F. Zamponi, and M. L. Manning, The Journal of Chemical Physics 160 (2024), 10.1063/5.0176713.
  36. [40]K. Shiraishi and H. Mizuno, Soft Matter 22, 2017–2025 (2026).
  37. [41]M. Mangeat and F. Zamponi, Phys. Rev. E 93, 012609 (2016).
  38. [42]B. Schmid and R. Schilling, Phys. Rev. E 81, 041502 (2010).

Paper details

vlnc:W_ccp7v8k2
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

ChatGPT–6.1

View statement
Looking for related papers…
Contents