Emergent cohesion via self-caging in maximally entangled rod packings Preprint: APS/123-QED

Yeonsu Jung, L. Mahadevan · 2026

Read Emergent cohesion via self-caging in maximally entangled rod packings Preprint: APS/123-QED by Yeonsu Jung, L. Mahadevan (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-2606-03952.txt.

Introduction

Jung and Mahadevan ask why a loose pile of rods can hold together. A twig nest stands on its own with no glue or ties. They study a rod packing that boosts the mean crossing count. Rods may not pass through each other in their setup. That rule pushes the packing toward tight entanglement.

They show the packing cages itself in that state. Group constraints block rods from escape, even at free edges. Pure repulsion and friction then keep the pile sound. Self-caging turns on free space in shift and turn directions. It is least when a set ratio hits one third. The result gives a plain geometric root for cohesion in grain-like matter.

Catalog record

CREATORYeonsu Jung, L. Mahadevan
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE28 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2606.03952

Browse the full library · About the Tower of Babel

Full text

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

Emergent cohesion via self-caging in maximally entangled rod packings Preprint: APS/123-QED Yeonsu Jung, L. Mahadevan Abstract Random packings of disordered rigid rods exhibit emergent cohesion, as exemplified in a nest of twigs that is self-equilibrated, free-standing structures. We analyze the geometric motif underlying this cohesion using a rod packing that maximizes the average crossing number subject to non-penetration constraints. We show that this protocol leads to self-caging: collective geometric constraints that prevent rod escape even in finite systems with free boundaries, leading to packings that remain mechanically cohesive due to a combination of purely repulsive and frictional interactions. We show that self-caging is controlled by the available free-volume in translational and rotational configuration spaces, which is minimal when N / ( Z ​ α ) = 1 / 3 N/(Z\alpha)=1/3 where N N is the number of rods, α \alpha is the aspect ratio, and Z Z is the average coordination number. Our results establish a minimal geometric motif for entanglement-induced cohesion in athermal rod packings, with implications for cohesive granular matter without attractive forces. Disordered packings with steric repulsion and dry friction exhibit collective rigidity by creating jammed structures 2 ; 13 ; 17 ; 10 ; 3 ; 18 . For spheres, jamming is commonly defined as the inability to rearrange without energetic cost, a notion that typically assumes fixed boundaries or the (infinite particle) thermodynamic limit; with free boundaries, grains can always be removed from the surface. In addition, there are robust algorithms for the generation and stability of jamming 11 ; 13 ; 16 . Rod packings display a distinct form of rigidity: even finite systems with free boundaries can remain cohesive because rods obstruct each other’s motion through geometry, contact, and friction—a phenomenon often termed entanglement 7 ; 9 ; 14 ; 1 ; 15 ; 8 . Despite its ubiquity, entanglement lacks both systematic protocols for its generation, and quantitative links to mechanical stability. Here we identify a minimal geometric motif that underlies cohesion in the random packings of rods: self-caging . We generate maximally entangled packings by optimizing a geometric functional subject to non-penetration constraints, and show that these packings can remain cohesive under mechanical perturbations even when the interactions are purely repulsive and frictional. Unlike jammed sphere packings –where grains near a free boundary can always escape – entangled rods packings can mutually block all lateral motions, trapping each another even with free boundaries, due to their extended nature. Figure 1: Maximizing entanglement for a packing with N=200 . (A) Schematic of a pair of rods. (B) Maximization of entanglement without a non-penetration constraint. Rods overlap near the global centroid. (C,D) Relaxation of the packing in (B) with an one-sided harmonic potential enforcing non-overlap, with the aspect ratios \alpha=200 (C) and \alpha=50 (D) The dotted circles of radius R in (C) and (D) denote spheres that contain all contact points in the packing. Quantification of entanglement. A packing of N rigid spherocylindrical rods of diameter d can be characterized by their generalized coordinates corresponding to the position of their centers and the orientation relative to some fixed origin \mathbf{q}=\{\mathbf{q}_{i}\}=\{(x_{i},y_{i},z_{i},\theta_{i},\phi_{i})\} for i=1,\dots,N . From these generalized coordinates, we can reconstruct their centerline by \mathbf{r}_{i}(s)=\mathbf{c}_{i}+s\mathbf{t}_{i} with \mathbf{c}_{i}=(x_{i},y_{i},z_{i}) , \mathbf{t}_{i}=(\sin\theta_{i}\cos\phi_{i},\sin\theta_{i}\sin\phi_{i},\cos\theta_{i}) , and s\in[-l/2,l/2] . We fix length l=1 and vary d to tune their aspect ratio \alpha=l/d . Then the equations describing the geometry (kinematics) and dynamics of the rods can be shown to depend only on \alpha=l/d (Suppl. Note 1 and 2). We quantify entanglement using the average crossing number (ACN) which provides an upper bound for true topological measures such as various polynomial metrics 5 . For rods i and j with centerlines \mathbf{r}_{i}(s) and \mathbf{r}_{j}(t) , with unit tangent vectors \mathbf{t}_{i} and \mathbf{t}_{j} , we define \mathrm{ACN}_{ij}=\frac{1}{4\pi}\int_{-l/2}^{l/2}\int_{-l/2}^{l/2}\frac{|(\mathbf{t}_{i}\times\mathbf{t}_{j})\cdot(\mathbf{r}_{i}-\mathbf{r}_{j})|}{|\mathbf{r}_{i}-\mathbf{r}_{j}|^{3}}\,ds\,dt, (1) which measures the average solid angle of directions for which two rods intersect in projection divided by the total solid angle ( 4\pi ), since the integrand is the local solid angle of direction along \mathbf{r}_{i}(s)-\mathbf{r}_{j}(t) . The maximum value of pairwise ACN is 0.5 and is attained when the two rods are perpendicular and in contact with each other (Fig. S1) consistent with intuition. In terms of an (extensive) entanglement quantity obtained for the whole network e=\sum_{i<j}\mathrm{ACN}_{ij} , we can define the normalized entanglement of a packing of N rods as a functional of \mathbf{q} 1 ; 8 \tilde{e}(\mathbf{q})=\frac{e(\mathbf{q})}{n_{p}}=\frac{1}{n_{p}}\sum_{i<j}\mathrm{ACN}_{ij} (2) with n_{p}=N(N-1)/2 . Maximizing entanglement. But how do we generate entangled rod packings? We do so by maximizing the entanglement measure \tilde{e}(\mathbf{q}) over configurations \mathbf{q} . Starting from initial states prepared via random sequential adsorption 6 , we could evolve the system under the gradient flow \frac{d\mathbf{q}}{dt}=\nabla_{\mathbf{q}}\tilde{e}(\mathbf{q}) , which drives the system toward a local maximizer \mathbf{q}^{*}=\arg\!\max_{\mathbf{q}}\tilde{e}(\mathbf{q}) . In practice, we employ the Fast Inertial Relaxation Engine (FIRE) 4 to efficiently reach such maximally entangled states. Since maximizing \tilde{e} alone leads to unphysical overlaps between rods, we subsequently enforce non-penetration constraints using a one-sided harmonic repulsion potential activated only when d_{ij}<d . See Supplementary Note 4 for detailed discussion on numerical implementation. For sufficiently large \alpha , the resulting relaxed configurations, \mathbf{q}^{**} , exhibit a dense contact-rich core surrounding a “thorny” shell (Fig. 1 C). For small \alpha , the thorny shell structure disappear and the packing starts to look spherical (Fig. 1 D). It is worth pointing out that in sharp contrast with an isolated pair, when \mathrm{ACN}_{ij} is maximized by perpendicular orientation at minimal separation, for N>3 rods, the frustration due to non-penetration leads to highly non-trivial hedgehog-like structures (as in Fig. 1 B–D; see Supplementary Video 1). Figure 2: The normalized contact spread R/l in Eq. 5 as a function of N/(Z\alpha) with Z=4 . The theoretical predictions (red lines) are given by Eqs. 3 and 4 . To quantify these structures, we use the radius R of the smallest sphere containing all contact points in a packing of N rods with aspect ratio \alpha . A packing with number density \rho=N/V has an average coordination number Z=\rho V_{\mathrm{ex}} , where the excluded volume for rods of length \zeta is V_{\mathrm{ex}}=\pi d\zeta^{2}/2 12 . When the core radius is large compared to the rod length ( R/l>1/2 ), the length scale \zeta is l , and subsequently Z=NV_{\mathrm{ex}}/V=(3N/4\pi R^{3})(\pi dl^{2}/2) . This yields \frac{R}{l}=\left(\frac{3N}{2\alpha Z}\right)^{1/3}. (3) In contrast, when R<l/2 , the core region truncates the rods, and the effective length scale becomes \zeta=2R . In this regime, the same argument gives, Z=NV_{\mathrm{ex}}/V=(3N/4\pi R^{3})(2\pi dR^{2}) , \frac{R}{l}=\frac{3N}{2\alpha Z},. (4) These two regimes define a crossover at N/(Z\alpha)=1/3 1 1 1 The solutions to (3x/8)^{1/3}=3x/2 are x=0,-\tfrac{1}{3},\tfrac{1}{3} . , separating an isotropic compact packing regime (Fig. 1 D) from the radially oriented “hedgehog” regime with a dense core and an outward-pointing shell (Fig. 1 C). To test these predictions, we compute the radius of gyration of contact points, R^{2}=\frac{1}{N_{c}}\sum_{i=1}^{N_{c}}\lvert\mathbf{p}_{i}-\mathbf{r}_{0}\rvert^{2}, (5) where \mathbf{p}_{i} are contact positions 2 2 2 In fully relaxed packings, \mathbf{q}^{**} , obtained using the one-sided repulsive potential for d_{ij}<d , every inter-rod distance d_{ij} in \mathbf{q}^{**} is strictly greater than d up to a numerical tolerance of \sim 10^{-9} . We therefore define a contact as a pair of rods whose separation is smaller than 1.01~d . and

Read the complete text