Geometric Fragmentation: Models & Mechanisms

• Geometric fragmentation is a field that studies how explicit geometric parameters, such as wall thickness and cross-sectional area, actively control fragment outcomes.
• Dynamic fracture models demonstrate that fragment size follows power-law scaling with variables like cross-sectional dimension and strain rate, integrating geometry into energy balances.
• Applications range from deterministic cutting and polyhedral reconstruction to stochastic partitioning and quantum state fragmentation, positioning geometry as a central organizing principle.

Searching arXiv for recent and foundational papers on “geometric fragmentation” and closely related uses of the term. Geometric fragmentation denotes a family of models and theories in which fragmentation is governed, constrained, or diagnosed primarily by geometry rather than by a fully detailed constitutive or stochastic failure law. In the literature represented here, the term spans several distinct but related uses: energy-based fracture models in which fragment size depends explicitly on geometric scales such as wall thickness; deterministic cutting-and-reassembly constructions in which occupied volume is controlled by fragment geometry; geometric reconstruction methods that infer fragmentation history from polyhedral shape; and abstract connectivity-based fragmentation in constrained many-body systems, where geometry organizes dynamically disconnected sectors. Across these settings, a common theme is that fragment statistics or fragment accessibility are determined by geometric variables—cross-sectional dimensions, aspect ratios, convex-mosaic combinatorics, path geometry, or plane-wise winding data—rather than by geometry-free bulk scaling alone (Goloveshkin et al., 2013).

1. Geometric fragmentation as a general concept

The strongest common denominator across the cited works is that geometry is not a passive descriptor of fragments but an active control parameter. In rapidly expanding ductile cylinders and rings, the fracture law depends explicitly on wall thickness or on a geometric functional of the cross-section, so fragment size is set by a balance between inertia and geometry-dependent plastic work (Goloveshkin et al., 2013). In granular matter, geometric fragmentation is defined as a deterministic sequence of cuts of a long square prism, followed by reassembly into volume-maximizing towers, so the central observable is the occupied volume generated by fragment geometry alone (Meladze, 13 Feb 2026). In polyhedral reconstruction, the premise is that fragment geometry encodes the stress field that created it, so coarse combinatorics such as the number of faces $F$ and vertices $V$ are treated as signatures of fragmentation mode (Torok et al., 11 Apr 2025).

A broader synthesis suggests several recurring meanings. One meaning is geometry-controlled fragmentation, where cross-sectional dimensions, thickness, or aspect ratio enter explicitly into fragment-size laws. A second is geometry-generated fragmentation, where fragmentation is defined by an idealized cutting rule and studied through resulting geometric configurations. A third is geometry-as-evidence, where fragment morphology is used to infer the formative process. A fourth, present in constrained quantum systems, is geometric fragmentation of state space, where geometry organizes disconnected dynamical sectors. This broader umbrella is an interpretation, but it is consistent with the range of usages documented in the cited papers (Fu et al., 26 Apr 2026).

2. Geometry as a control variable in dynamic fracture

In ductile dynamic fragmentation, the canonical example is the rapidly expanding thin-walled cylinder under plane strain. The model of Goloveshkin and Myagkov treats a uniformly expanding cylinder with circumferential strain rate

$\dot\varepsilon_\varphi = \frac{V_R}{R},$

thickness $2h$, and $2h \ll R$, with incompressible ideally rigid-plastic material of yield stress $Y$. Fragmentation is represented as circumferential cracking induced by localized necking in the wall, and the core balance is between fragment kinetic energy and fracture energy computed from a two-dimensional necking solution (Goloveshkin et al., 2013). The fracture energy per unit axial length is

$A_p = \frac{4}{3}Y h^2,$

and, after neglecting elastic energy because $P/T$ is small for the experiments considered, the mean fragment length and fragment number are

$$2a = \left(16\sqrt{3}\,\frac{Y h}{\rho\,\dot\varepsilon_\varphi^2}\right)^{1/3}, \qquad N = \pi R\left(\frac{\rho\,\dot\varepsilon_\varphi^2}{16\sqrt{3}\,Y h}\right)^{1/3}.$$

These relations exhibit the classical $2/3$-power strain-rate dependence, but they also introduce an explicit $V$0 dependence that is absent from one-dimensional Grady–Kipp theory (Goloveshkin et al., 2013).

The corresponding ring generalization extends the same geometric logic to arbitrary convex, centrally symmetric cross-sections. There the key quantity is not wall thickness alone but the geometric functional

$V$1

minimized over orientations and represented through

$V$2

The fragment number is then expressed in terms of $V$3, density $V$4, yield stress $V$5, strain rate $V$6, cross-sectional area $V$7, and radius $V$8, preserving the same $V$9 structure while making cross-sectional geometry explicit (Goloveshkin et al., 2018). For rectangles, $\dot\varepsilon_\varphi = \frac{V_R}{R},$0 reduces to $\dot\varepsilon_\varphi = \frac{V_R}{R},$1, $\dot\varepsilon_\varphi = \frac{V_R}{R},$2, or $\dot\varepsilon_\varphi = \frac{V_R}{R},$3 depending on aspect ratio, and for a circular cross-section $\dot\varepsilon_\varphi = \frac{V_R}{R},$4, showing that geometry enters mainly through a characteristic linear size with a modest shape factor (Goloveshkin et al., 2018).

These models define one major technical meaning of geometric fragmentation: fragmentation laws in which geometry enters directly through local necking work. Their significance is that fragment size is not treated as depending only on material parameters and loading rate; the transverse geometric scale is itself a constitutive part of the prediction.

3. Deterministic geometric fragmentation and occupied volume

A distinct usage appears in the ordered model for fragmented granular matter. Here geometric fragmentation is a purely deterministic process: a very long rectangular prism with square cross section $\dot\varepsilon_\varphi = \frac{V_R}{R},$5 and length

$\dot\varepsilon_\varphi = \frac{V_R}{R},$6

is cut first into $\dot\varepsilon_\varphi = \frac{V_R}{R},$7 equal pieces and then recursively into $\dot\varepsilon_\varphi = \frac{V_R}{R},$8 equal pieces at each stage, producing fragments of length

$\dot\varepsilon_\varphi = \frac{V_R}{R},$9

At each stage the fragments are reassembled into a highly ordered square-cross-section tower with a single square-prismatic central cavity, chosen to maximize enclosed volume (Meladze, 13 Feb 2026).

The resulting occupied volume is not the solid volume alone but the volume of solid plus cavity. Because the solid volume remains

$2h$0

the entire evolution is determined by cavity geometry. For stage $2h$1,

$2h$2

This implies non-monotonic volume evolution: there is an initial increase above the original prism volume, followed by monotone decrease under further fragmentation, with a universal terminal value

$2h$3

independent of $2h$4 (Meladze, 13 Feb 2026). When mapped to grain geometry with grain length $2h$5 and square cross-sectional size $2h$6, the upper bound becomes

$2h$7

and the minimum at $2h$8 is the same $2h$9 limit (Meladze, 13 Feb 2026).

This formulation uses geometric fragmentation in a stricter sense than the dynamic-fracture models: the process is defined by cuts and reassembly alone, with no forces, disorder, or constitutive law. The significance of the model is not fracture mechanics but the derivation of sharp geometric upper bounds on occupied volume, intended as ordered reference states for disordered granular assemblies.

4. Geometry as a record of fragmentation history

A third major theme is that fragmentation leaves a morphological record in fragment shape. In the polyhedral reconstruction framework, real rock fragments are approximated by ideal convex polyhedra whose combinatorics reflect the stress field under which fragmentation occurred. The central claim is that, under ideal fragmentation conditions, fragments are convex polyhedra, and the numbers of faces $2h \ll R$0 and vertices $2h \ll R$1 are diagnostic: hydrostatic stress is associated with Voronoi-like cells with average $2h \ll R$2, whereas multiple successive shear events lead to $2h \ll R$3 (Torok et al., 11 Apr 2025).

The reconstruction algorithm begins from a 3D scan, computes its convex hull, maps face normals to a Fibonacci-lattice spherical histogram, smooths that histogram with Gaussian kernels over

$2h \ll R$4

in $2h \ll R$5 equidistant steps, and identifies local maxima as dominant face orientations. Plane offsets are then estimated by a weighted distance histogram, and the ideal polyhedron is reconstructed as the bounded cell containing the reference point $2h \ll R$6 in the induced hyperplane mosaic (Torok et al., 11 Apr 2025). Because the reconstructed objects are simple polyhedra, Euler’s formula yields

$2h \ll R$7

The method is benchmarked on $2h \ll R$8 fragments and compared to hand counts; the two variants trade exact face-count agreement against volume fidelity, with algorithm (b) giving volume ratio $2h \ll R$9 relative to the original convex hull and algorithm (a) giving $Y$0 (Torok et al., 11 Apr 2025).

A related but more global morphological perspective comes from convex-mosaic theory. There, fragmented solids are treated as convex mosaics, and the average combinatorics of fragments exhibit two attractors in 2D—Platonic quadrangles and Voronoi hexagons—and a dominant Platonic attractor in 3D, where average fragment geometry is cuboid (Domokos et al., 2019). In 2D, regular primitive mosaics have $Y$1, while isotropic Voronoi mosaics sit near $Y$2. In 3D, primitive mosaics yield

$Y$3

whereas Poisson–Voronoi mosaics have approximately

$Y$4

The paper argues that generic binary breakup drives mosaics toward the Platonic attractor, explaining why natural rock fragments exhibit average combinatorics close to cuboids (Domokos et al., 2019).

Taken together, these approaches define geometric fragmentation as the study of how fragment geometry encodes formative stress, crack topology, and fracture sequence. A plausible implication is that fragment morphology can function as a compressed descriptor of fragmentation history even when direct dynamical data are unavailable.

5. Statistical and kinetic formulations of geometric fragmentation

Several cited works treat fragmentation as a geometric stochastic process and focus on scaling laws of fragment sizes. In interval fragmentation, the object is a one-dimensional element of size $Y$5 that can fragment into $Y$6 pieces with probabilities $Y$7, conserving length in each event. With the logarithmic variable

$Y$8

the distribution can be solved exactly through a generating function $Y$9, and for power-law fragmentation probabilities

$A_p = \frac{4}{3}Y h^2,$0

the asymptotic small-$A_p = \frac{4}{3}Y h^2,$1 behavior depends on whether $A_p = \frac{4}{3}Y h^2,$2 or $A_p = \frac{4}{3}Y h^2,$3 (Fortin et al., 2013). For $A_p = \frac{4}{3}Y h^2,$4,

$A_p = \frac{4}{3}Y h^2,$5

whereas for $A_p = \frac{4}{3}Y h^2,$6,

$A_p = \frac{4}{3}Y h^2,$7

This is a geometric fragmentation model in the literal sense of recursively partitioning an interval (Fortin et al., 2013).

Minimal fragmentation of regular polygonal plates studies the opposite extreme: every plate is cut once into exactly two pieces. For regular $A_p = \frac{4}{3}Y h^2,$8-gons under isotropic random cracks, the accumulated probability for the normalized mass $A_p = \frac{4}{3}Y h^2,$9 of the smaller fragment obeys, for any finite $P/T$0,

$P/T$1

whereas the disk limit exhibits

$P/T$2

The crossover between polygonal and disk behavior occurs at

$P/T$3

in the isotropic model (Dias et al., 2014). The same paper also reports a second power-law regime with exponent $P/T$4 for an anisotropic model (Dias et al., 2014).

For rectangular fragmentation with discrete side lengths, the process jams when all fragments become sticks of minimal width. The average number of sticks in the jammed state scales as

$P/T$5

for large rectangle area $P/T$6, independent of aspect ratio, while the stick-length distribution has an exact tail

$P/T$7

hence $P/T$8 (Ben-Naim et al., 2019). This is another explicitly geometric model: the fragmentation law is a stochastic process on lattice rectangles rather than a constitutive failure model.

Across these statistical formulations, the common structure is that fragment distributions are derived from geometric rules of partition rather than from explicit elastodynamic simulation. This suggests that geometric fragmentation is often best viewed as a theory of admissible partitions and their induced statistics.

6. Extensions beyond ordinary material fracture

The term also appears in systems where “fragmentation” no longer means disconnected pieces of solid matter. In crumpled thin sheets, fragmentation refers to the partition of a connected sheet into flat facets separated by ridges. The state variable is the facet-area distribution $P/T$9, which evolves by a fragmentation rate equation

$$2a = \left(16\sqrt{3}\,\frac{Y h}{\rho\,\dot\varepsilon_\varphi^2}\right)^{1/3}, \qquad N = \pi R\left(\frac{\rho\,\dot\varepsilon_\varphi^2}{16\sqrt{3}\,Y h}\right)^{1/3}.$$0

with breakup rate $$2a = \left(16\sqrt{3}\,\frac{Y h}{\rho\,\dot\varepsilon_\varphi^2}\right)^{1/3}, \qquad N = \pi R\left(\frac{\rho\,\dot\varepsilon_\varphi^2}{16\sqrt{3}\,Y h}\right)^{1/3}.$$1, $$2a = \left(16\sqrt{3}\,\frac{Y h}{\rho\,\dot\varepsilon_\varphi^2}\right)^{1/3}, \qquad N = \pi R\left(\frac{\rho\,\dot\varepsilon_\varphi^2}{16\sqrt{3}\,Y h}\right)^{1/3}.$$2, and scale-invariant daughter law

$$2a = \left(16\sqrt{3}\,\frac{Y h}{\rho\,\dot\varepsilon_\varphi^2}\right)^{1/3}, \qquad N = \pi R\left(\frac{\rho\,\dot\varepsilon_\varphi^2}{16\sqrt{3}\,Y h}\right)^{1/3}.$$3

The resulting facet-area distribution, ridge-length gamma law, and logarithmic growth of total crease length are interpreted as consequences of geometric frustration under confinement (Andrejevic et al., 2020).

In fractal-like agglomerates, fragmentation is defined by random bond removal in a loopless contact graph of equal-size spheres. The fragment-size distribution depends only on morphology, especially the fractal dimension $$2a = \left(16\sqrt{3}\,\frac{Y h}{\rho\,\dot\varepsilon_\varphi^2}\right)^{1/3}, \qquad N = \pi R\left(\frac{\rho\,\dot\varepsilon_\varphi^2}{16\sqrt{3}\,Y h}\right)^{1/3}.$$4, and is approximated by a symmetric beta distribution with exponent $$2a = \left(16\sqrt{3}\,\frac{Y h}{\rho\,\dot\varepsilon_\varphi^2}\right)^{1/3}, \qquad N = \pi R\left(\frac{\rho\,\dot\varepsilon_\varphi^2}{16\sqrt{3}\,Y h}\right)^{1/3}.$$5. A universal morphology-dependent density

$$2a = \left(16\sqrt{3}\,\frac{Y h}{\rho\,\dot\varepsilon_\varphi^2}\right)^{1/3}, \qquad N = \pi R\left(\frac{\rho\,\dot\varepsilon_\varphi^2}{16\sqrt{3}\,Y h}\right)^{1/3}.$$6

is proposed, with the straight-chain limit $$2a = \left(16\sqrt{3}\,\frac{Y h}{\rho\,\dot\varepsilon_\varphi^2}\right)^{1/3}, \qquad N = \pi R\left(\frac{\rho\,\dot\varepsilon_\varphi^2}{16\sqrt{3}\,Y h}\right)^{1/3}.$$7 (Drossinos et al., 2019). Here the geometry is the topology and fractal morphology of the aggregate.

In quantum many-body physics, geometric fragmentation denotes fragmentation of Hilbert space organized by geometry. In the cubic U(1) quantum dimer model, maximal winding in one direction freezes motion on $$2a = \left(16\sqrt{3}\,\frac{Y h}{\rho\,\dot\varepsilon_\varphi^2}\right)^{1/3}, \qquad N = \pi R\left(\frac{\rho\,\dot\varepsilon_\varphi^2}{16\sqrt{3}\,Y h}\right)^{1/3}.$$8 and $$2a = \left(16\sqrt{3}\,\frac{Y h}{\rho\,\dot\varepsilon_\varphi^2}\right)^{1/3}, \qquad N = \pi R\left(\frac{\rho\,\dot\varepsilon_\varphi^2}{16\sqrt{3}\,Y h}\right)^{1/3}.$$9 plaquettes and reduces dynamics to decoupled $2/3$0-plane sectors. The Hilbert space then fragments according to the distribution of plane-wise winding numbers, and the number of fragments grows exponentially in linear system size, producing weak fragmentation (Steinegger et al., 5 Aug 2025). In one-dimensional integer-spin chains, “peak-valley fragmentation” labels disconnected Krylov sectors by the heights and depths of alternating peaks and valleys in a geometric height representation; the local PV condition preserves

$2/3$1

yielding exponentially many sectors and strong Hilbert space fragmentation (Fu et al., 26 Apr 2026).

These examples broaden the concept substantially. This suggests that geometric fragmentation can function as a unifying idea whenever geometry defines the admissible decomposition of a system—whether the decomposed objects are physical fragments, facets, graph components, or dynamical subspaces.

7. Scope, limitations, and recurring themes

Despite its breadth, the term does not refer to a single universal formalism. In expanding cylinders and rings, it denotes geometry-dependent energy balance under ductile necking (Goloveshkin et al., 2013). In granular towers, it denotes a deterministic cutting-and-reassembly construction (Meladze, 13 Feb 2026). In rock-fragment morphology, it denotes the inference of stress regime from convex polyhedral geometry (Torok et al., 11 Apr 2025). In convex mosaics, it denotes the combinatorial attractors of crack-generated tessellations (Domokos et al., 2019). In constrained quantum systems, it denotes geometry-defined fragmentation of state-space connectivity (Steinegger et al., 5 Aug 2025).

Several recurrent limitations also appear. Many fracture-mechanics models are two-dimensional, plane-strain, or thin-wall approximations and predict only average fragment size rather than full distributions (Goloveshkin et al., 2013). Ordered geometric models for granular matter intentionally ignore friction, gravity, cohesion, and disorder, so they provide geometric envelopes rather than typical packings (Meladze, 13 Feb 2026). Polyhedral reconstruction assumes convexity and may fail on very rounded pebbles or where small faces are ambiguous between primary fracture and later chipping (Torok et al., 11 Apr 2025). Convex-mosaic theory idealizes cracks as flat and fragments as convex polytopes (Domokos et al., 2019). Hilbert-space applications depend on highly constrained sectors or specific local rules (Steinegger et al., 5 Aug 2025).

A cautious synthesis is therefore appropriate. Geometric fragmentation is best understood not as one model class but as a research program centered on the proposition that fragmentation outcomes are often controlled, organized, or remembered by geometry. In some settings geometry enters as a quantitative scale in the governing law; in others it is the object being optimized, classified, or decoded. What unifies these usages is the claim that fragmentation cannot be fully characterized by material strength or stochastic branching alone: geometry is itself a dynamical variable, a statistical constraint, and a repository of process information.