Why it matters

Eshelby’s solution is a foundation of micromechanics. Ellipsoids are used to represent coherent precipitates, phase-transforming regions, inclusions and inhomogeneities, and they enter many homogenization and mean-field models.

Real three-dimensional microstructures can be faceted, cuboidal, necked, interconnected, or locally rough. In practice, a complicated particle is often replaced by a best-fit ellipsoid because the resulting model is tractable. What is usually missing is error control: how much mechanics is lost when the geometry is simplified?

A quantitative stability theory would turn the ellipsoidal approximation from a geometric convenience into a controlled mechanical approximation. Given a measured shape, it could tell us whether the Eshelby idealization should remain accurate and which geometric features most strongly destroy uniformity.

What we know

1957 — Exact ellipsoidal uniformity.
Eshelby, The determination of the elastic field of an ellipsoidal inclusion, and related problems showed that a uniform eigenstrain in an ellipsoidal region embedded in an infinite homogeneous elastic solid generates a uniform interior elastic strain in the classical setting.

1996–1998 — Non-ellipsoidal exclusions and local geometry.
Rodin (1996) and Lubarda & Markenscoff (1998) excluded broad non-ellipsoidal classes. Most directly relevant to EL01, Markenscoff, Inclusions with constant eigenstress (1998) showed locally that infinitesimal perturbations preserving constant eigenstress are precisely the perturbations moving within the ellipsoid family. This identifies the ellipsoid family as the natural local null directions.

2008–2010 — Major progress on the inverse problem.
Kang & Milton (2008) proved the classical weak three-dimensional Eshelby conjecture. Liu (2008) clarified distinct formulations and their connection to potential theory. Ammari et al. (2010) established important regimes of the stronger one-loading formulation.

2011 — Fields for weakly non-spherical inclusions.
Huang et al. (2011) derived explicit Eshelby tensors for arbitrary weakly non-spherical three-dimensional inclusions and validated the perturbative expressions against exact solutions and FEM.

2023 — Further constrained progress on the strong conjecture.
Yuan, Huang & Wang (2023) addressed part of the remaining repeated-eigenvalue structure under additional constraints. The exact inverse problem continues to narrow, but it is distinct from the quantitative stability question considered here.

2025 — Adjacent extensions of uniformity theory.
Ye (2025) developed field solutions and uniformity conditions for linear multiphysical inclusions. This is relevant recent activity around inclusion uniformity, but it does not provide the two-sided quantitative near-ellipsoid stability law posed in EL01.

The frontier: from exactness to stability

The classical question is qualitative:

Which shapes give exactly uniform interior strain?

EL01 asks for a quantitative version.

Let

$$q(\Omega) = \frac{\|\boldsymbol{\varepsilon} -\langle\boldsymbol{\varepsilon}\rangle_\Omega\|_{L^2(\Omega)}} {\text{natural strain scale}\,\sqrt{|\Omega|}}$$

measure whole-interior strain heterogeneity, so that $q=0$ exactly for a uniform interior strain.

Let $\mathcal E$ denote the ellipsoid family and define

$$\rho(\Omega,\mathcal E) = \inf_{E\in\mathcal E} d_{\mathrm{shape}}(\Omega,E).$$

The target is a two-sided stability law of the form

$$\boxed{ c\,\rho(\Omega,\mathcal E) \;\le\; q(\Omega) \;\le\; C\,\rho(\Omega,\mathcal E) }$$

under clearly stated mechanical and geometric assumptions.

The two sides answer different questions. The upper bound asks whether a nearly ellipsoidal inclusion is necessarily nearly uniform. The lower bound is a rigidity statement: if the interior strain is nearly uniform, must the shape itself be near some ellipsoid?

The correct shape distance is part of the mechanics. A metric based only on maximum radial amplitude can miss rapidly varying roughness that has a disproportionately large elastic effect.

Recent progress: quantitative local stability and finite-shape bounds

Ni (2026), Quantitative Eshelby stability: local and finite-shape theory through AI-assisted mechanics, is available as an arXiv preprint. It establishes three complementary results in an infinite homogeneous isotropic linear-elastic solid with uniform dilatational eigenstrain, identical inclusion and matrix moduli, perfect bonding, and no remote loading.

Nonlinear two-sided local stability. Near any fixed nondegenerate ellipsoid, whole-interior strain heterogeneity is bounded above and below by a geometry-only distance to the nearby ellipsoid family. The admissible shapes lie in a sufficiently small smooth neighborhood. Changes that merely translate, rotate, rescale, or otherwise produce another ellipsoid are removed from the distance. The theorem controls both how shape imperfections generate heterogeneity and how near-uniformity constrains the shape.

Critical half-order geometric scaling. The half-order Sobolev scale captures both imperfection amplitude and morphological wavelength. It is the critical order for frequency-independent two-sided control: a weaker order loses the uniform upper bound, while a stronger order loses the positive uniform lower bound. For a sphere, the preprint derives the exact linear spectrum and sharp all-mode bounds after the ellipsoidal directions are removed.

Convergent bounds for prescribed finite shapes. Subtracting a volume-balanced ellipsoidal source gives an exact residual-field budget. Partial interior and exterior field information yields lower and upper bounds on the heterogeneity. With exact integration, these bounds converge to the exact response as the resolved regions exhaust the respective domains. This construction applies to prescribed finite geometries beyond the local theorem's neighborhood.

Numerical evidence. The finite-shape bounds are evaluated for protruded, indented, and necked inclusions. Independently assembled displacement-based finite-element calculations give heterogeneity values within the reported intervals.

What remains open

The local theorem supplies a geometry-only two-sided relation near ellipsoids. The finite-shape construction supplies convergent bounds for a specified geometry using residual-field information. A global geometry-only two-sided stability law beyond the local neighborhood remains open, including the appropriate shape class, distance, constants, and possible counterexamples.

Further questions concern general deviatoric and mixed eigenstrain modes, elastic modulus mismatch, anisotropic constitutive response, inclusions near free surfaces or interfaces, periodic domains, and interacting multiple inclusions. Both the geometry-only stability relation and the finite-shape bounds require extension to these physical settings.

A complete theory may not use one universal constant or one universal geometry metric. Part of the problem is to determine which quantities remain stable across these different physical settings.

What would count as progress?

• Prove a geometry-only two-sided stability relation beyond the local neighborhood, or extend the relation to a more general eigenstrain, material or boundary setting, with the shape class and norm stated precisely.
• A sharp counterexample showing that a proposed global metric or stability statement cannot hold would materially settle part of the problem.
• Obtain effective local constants for general nonspherical ellipsoids, including their dependence on aspect ratio and the admissible neighborhood.
• Extend computable geometry-to-field error bounds and independent validation to additional material, loading or boundary settings relevant to real inclusions.

Scientific scope

This page does not claim that the classical Eshelby conjecture is simply open. The weak three-dimensional conjecture is solved in its standard setting, and strong formulations have substantial partial results.

Experimental precipitate morphologies motivate the need for finite shape deviations and best-fit ellipsoidal reductions; they are not, by themselves, experimental validation of a stability theorem.

The literature audit of 30 September 2026, preceding this preprint, did not identify a published result matching the full combination of a whole-interior heterogeneity measure, two-sided quantitative control, distance to the full ellipsoid family, a morphology-sensitive half-order geometry scale, and nonlinear validity near a general nondegenerate ellipsoid. That search result should not be interpreted as proof of novelty.

Selected milestones

• J. D. Eshelby, The determination of the elastic field of an ellipsoidal inclusion, and related problems (1957) — exact ellipsoidal uniformity.
• X. Markenscoff, Inclusions with constant eigenstress (1998) — local ellipsoid tangent structure for constant eigenstress.
• H. Kang & G. W. Milton, Solutions to the Pólya–Szegö Conjecture and the Weak Eshelby Conjecture (2008) — proof of the classical weak three-dimensional Eshelby conjecture.
• H. Ammari et al., Progress on the strong Eshelby’s conjecture and extremal structures for the elastic moment tensor (2010) — major progress on the strong three-dimensional conjecture.
• M. Huang et al., Explicit expressions of the Eshelby tensor for an arbitrary 3D weakly non-spherical inclusion (2011) — perturbative fields for weakly non-spherical inclusions.
• T. Yuan, K. Huang & J. Wang, A constrained proof of the strong version of the Eshelby conjecture for three-dimensional isotropic media (2023) — constrained progress on the remaining strong-conjecture structure.
• Bo Ni, Quantitative Eshelby stability: local and finite-shape theory through AI-assisted mechanics, arXiv:2610.13400 (2026, preprint) — nonlinear local stability, critical half-order scaling, and convergent finite-shape bounds.

Progress log

11 October 2026 — Preprint incorporated; entry v1.1.
The arXiv preprint supplies a public reference for the local theorem, critical geometric scaling and sphere spectrum, and convergent finite-shape construction. This update states the model assumptions, records the independent numerical comparisons, and distinguishes the remaining global geometry-only and broader physical-setting questions. Scientific status remains partial progress.

30 September 2026 — Problem statement v1.0.
Quantitative stability is separated explicitly from the classical exact Eshelby conjectures. The current manuscript is recorded as partial progress in a local isotropic setting; broader material, loading, boundary and finite-shape regimes remain open.

Why it matters

Dynamic fracture controls catastrophic failure under impact, shock, blast, rapid loading, fragmentation, and many high-rate manufacturing and structural events. Predicting the onset of fracture is not enough when a crack can accelerate, change morphology, branch, or transition to a different propagation regime within microseconds.

The engineering question is not merely how fast a crack could move in an idealized model. It is which dynamic state a real material can actually sustain before an instability changes the fracture process.

A predictive theory would connect measurable material properties and geometry to:

• the attainable crack-speed range;
• the first instability encountered as driving increases;
• the post-instability fracture mode;
• the role of specimen thickness, disorder, and process-zone physics;
• and the conditions under which trans-Rayleigh or supershear tensile cracks become possible.

What we know

1921 — Energy balance provides the organizing principle.
Griffith, The Phenomena of Rupture and Flow in Solids established the energetic foundation of fracture mechanics.

1951 — The moving-crack elastic field.
Yoffe, The moving Griffith crack analyzed a steadily propagating crack and helped establish the classical dynamic picture in which wave speeds strongly constrain the near-tip elastic field.

1980s–1990s — Experiments reveal instability well below the classical endpoint.
The high-speed experiments of Ravi-Chandar & Knauss, followed by Fineberg et al. (1991) and Sharon, Gross & Fineberg (1995), showed that real brittle cracks can destabilize through microbranching before reaching the Rayleigh-wave speed predicted as the distinguished endpoint of the classical smooth Mode-I LEFM branch.

2007–2017 — Suppressing microbranching exposes a different instability.
In thin brittle gels, Livne, Ben-David & Fineberg (2007) observed a high-speed oscillatory instability. Goldman et al. (2012) connected its wavelength to an intrinsic nonlinear near-tip scale, and Chen, Bouchbinder & Karma (2017) developed a nonlinear theory of the corresponding path instability.

Three-dimensional crack fronts introduce additional physics.
Crack-front waves, microbranches, and disorder make the 3D problem qualitatively harder. Direct crack-front experiments such as Kolvin, Cohen & Fineberg (2015) show how front geometry couples to local instability.

2023 — Supershear Mode-I fracture is directly observed.
Wang, Shi & Fineberg, Tensile cracks can shatter classical speed limits demonstrated tensile cracks propagating beyond the classical Rayleigh/shear-wave range in brittle neo-Hookean materials. This does not invalidate dynamic LEFM within its domain; it demonstrates that the classical sub-Rayleigh branch is not a universal description of all constitutive regimes.

Recent progress

2024 — Geometric nonlinearity and 3D disorder sharpen the problem.
Pundir, Adda-Bedia & Kammer showed that geometric nonlinearity can admit transonic and supershear Mode-I propagation in a continuum model with the crack confined to a weak plane. Separately, Lubomirsky & Bouchbinder showed in 3D simulations that quenched disorder and branching can generate an effective disorder-dependent limiting velocity.

2025 — Dynamic failure can switch among distinct modes.
Wang, Fineberg & Needleman studied transitions among crack-like, supershear-like, and spall-like separation modes under tensile loading, reinforcing that the modern problem is one of mode selection, not one universal velocity ceiling.

2026 — New experiments and computations continue to expose material-dependent regimes.
Dynamic fracture and catastrophic crack branching in highly entangled hydrogels reports strongly rate- and material-dependent high-speed fracture with unconventional branching behavior. Ignatev et al. reproduce crack-speed trends and microbranching in a peridynamic framework using an incubation-time fracture criterion.

These advances narrow the problem but do not provide a general theory that predicts which dynamic branch a given material will select.

The frontier: branch selection and stability

For a smooth Mode-I crack, a useful starting point is dynamic energy balance,

$$G(v;\mathcal G,\mathcal C) = \Gamma(v;\mathcal P),$$

where

• $G$ is the dynamic energy-release rate,
• $v$ is crack speed,
• $\mathcal G$ represents geometry and loading,
• $\mathcal C$ represents the bulk constitutive response,
• and $\mathcal P$ represents fracture/process-zone physics.

Satisfying energy balance is necessary, but it is not sufficient. A dynamically admissible branch must also be stable to perturbations of the path, crack front, and fracture process.

Schematically, define

$$v_{\mathrm{att}} = \sup\left\{ v: \text{a dynamically admissible and stable tensile-crack state exists} \right\},$$

and

$$v_{\mathrm{inst}} = \inf\left\{ v: \text{the selected smooth crack state loses stability} \right\}.$$

The open problem is to derive

$$\boxed{ v_{\mathrm{att}}, \qquad v_{\mathrm{inst}}, \qquad \text{and the selected post-instability mode} }$$

from mechanics for a clearly specified constitutive law, fracture/process-zone law, geometry, dimensionality, and disorder.

Why the Rayleigh-wave speed is not the whole answer

For a smooth Mode-I crack in classical linear elastic fracture mechanics, the Rayleigh-wave speed $c_R$ is a distinguished endpoint of the conventional sub-Rayleigh branch.

But three different statements must not be conflated:

1. The classical LEFM branch has a Rayleigh-speed endpoint.
2. A real crack may become unstable before reaching that endpoint.
3. A different constitutive or fracture regime may support a different dynamic branch, including trans-Rayleigh or supershear propagation.

The open problem concerns the selection and stability of these physically different branches.

What remains open

A general theory should predict, for a stated material and fracture model:

• which tensile-crack branches exist;
• which of those branches are dynamically stable;
• which branch is selected from a given loading history;
• the first instability encountered as driving increases;
• whether the instability is microbranching, oscillation, crack-front roughening, spall-like separation, or another mode;
• how the thresholds depend on nonlinear elasticity, fracture energy $\Gamma(v)$, cohesive/process-zone strength and length scales, material disorder, specimen thickness, and geometry;
• and how two-dimensional and three-dimensional fracture regimes connect.

The difficulty is not simply obtaining a faster numerical crack. The theory must explain why a particular material and geometry select one observed regime rather than another.

What would count as progress?

• A stability theory that predicts the first dynamic instability from independently measurable material and fracture parameters.
• A mechanics-based criterion distinguishing sub-Rayleigh, trans-Rayleigh, supershear, branching, and spall-like regimes.
• A predictive 3D theory linking crack-front disorder and branching to attainable crack speed.
• A controlled asymptotic connection between near-tip nonlinear/process-zone physics and macroscopic dynamic instability.
• Experiments capable of discriminating competing theories using material, thickness, or loading changes specified in advance.

Scientific scope

This page does not claim that the Rayleigh-wave speed is irrelevant. It is fundamental to the classical smooth Mode-I solution in linear elastic fracture mechanics.

Nor does the observation of supershear tensile cracks imply that LEFM is generally invalid. Classical dynamic LEFM remains quantitatively successful in important sub-Rayleigh regimes.

The problem is also not identical in two and three dimensions. Suppressing three-dimensional microbranching can expose a two-dimensional oscillatory instability that would otherwise not be the first observed instability.

Finally, calculations in which the crack is constrained to a weak plane should not be interpreted as unrestricted predictions of freely branching three-dimensional fracture.

Key papers and milestones

• A. A. Griffith, The Phenomena of Rupture and Flow in Solids (1921) — energetic foundation of fracture mechanics.
• E. H. Yoffe, The moving Griffith crack (1951) — classical moving-crack elastic field.
• K. Ravi-Chandar & W. G. Knauss, dynamic-fracture experiments (1984) — foundational high-speed fracture observations.
• J. Fineberg et al., Instability in dynamic fracture (1991) — experimental critical-velocity/instability behavior.
• E. Sharon, G. Gross & J. Fineberg, Local crack branching as a mechanism for instability in dynamic fracture (1995) — microbranching as a dynamic instability mechanism.
• A. Livne, O. Ben-David & J. Fineberg, Oscillations in Rapid Fracture (2007) — high-speed oscillatory instability in thin brittle gels.
• T. Goldman et al., Intrinsic Nonlinear Scale Governs Oscillations in Rapid Fracture (2012) — nonlinear near-tip length scale linked to oscillation wavelength.
• C.-H. Chen, E. Bouchbinder & A. Karma, Instability in dynamic fracture and the failure of the classical theory of cracks (2017) — nonlinear theory of the oscillatory instability.
• M. Wang, S. Shi & J. Fineberg, Tensile cracks can shatter classical speed limits (2023) — direct supershear Mode-I fracture in brittle neo-Hookean materials.
• M. Pundir, M. Adda-Bedia & D. S. Kammer, Transonic and Supershear Crack Propagation Driven by Geometric Nonlinearities (2024) — nonlinear continuum supershear branch under path confinement.
• Y. Lubomirsky & E. Bouchbinder, Quenched disorder and instability control dynamic fracture in three dimensions (2024) — 3D disorder, branching, and effective limiting velocity.
• M. Wang, J. Fineberg & A. Needleman, crack–supershear–spall transition (2025) — selection among distinct tensile separation modes.
• Dynamic fracture and catastrophic crack branching in highly entangled hydrogels (2026) — recent high-speed fracture experiments in a distinctmaterial regime.

Progress log

30 September 2026 — Problem statement v1.0.
The problem is framed as selection and stability of dynamic tensile-fracture branches rather than as a universal scalar crack-speed limit. Recent supershear, 3D disorder, branching, and hydrogel results are incorporated into the known/open boundary.

Why it matters

Plasticity models used in engineering routinely contain constitutive ingredients such as yield conditions, slip laws, hardening variables, gradient terms, and history dependence. Many of these quantities are known to originate from dislocation motion and interactions, but the connection is usually made through phenomenology, fitted parameters, or model-specific bridges.

A first-principles scale hierarchy would do something more demanding. It would tell us:

• which microscopic information survives coarse-graining;
• which internal variables are sufficient at a given observation scale;
• which constitutive terms emerge from dislocation correlations and reactions;
• when local plasticity is adequate;
• when gradients, nonlocality, stochasticity, or memory are unavoidable;
• and how to decide whether MD, DDD, continuum dislocation dynamics, crystal plasticity, or a macroscopic constitutive law is the appropriate description.

This is important not only for accuracy. It goes to the foundations of continuum plasticity: what is the smallest set of state variables that makes plastic deformation predictable?

The multiscale hierarchy already exists

A useful schematic is

$$\boxed{ \text{atomic dynamics} \rightarrow \text{dislocation-line dynamics} \rightarrow \text{kinetic / statistical dislocation theory} \rightarrow \text{continuum dislocation / crystal-scale plasticity} \rightarrow \text{macroscopic continuum plasticity} }$$

with representative computational descriptions

$$\text{MD} \rightarrow \text{DDD} \rightarrow \text{kinetic theory / CDD} \rightarrow \text{FDM / CDD-CP / crystal plasticity} \rightarrow \text{polycrystal and engineering-scale constitutive models}.$$

The open problem is not to invent these levels. They already exist and many successful bridges have been demonstrated.

The problem is to determine when the arrows between them are controlled, predictive, and transferable.

What we know

1953 — Dislocation density becomes a continuum geometric quantity.
Nye, Some geometrical relations in dislocated crystals showed that a three-dimensional dislocation content can be represented through a second-rank tensor related to lattice curvature. This established a fundamental bridge between discrete line defects and continuum incompatibility.

1997 — A BBGKY-like hierarchy appears in dislocation mechanics.
Groma, Link between the microscopic and mesoscopic length-scale description of the collective behavior of dislocations started from an $N$-dislocation evolution equation and derived a hierarchy for one-, two-, three-, and higher-order distribution functions. Closure required an approximation to higher-order correlations.

This is an important conceptual precursor to the present problem: the loss of information under coarse-graining appears explicitly as a closure problem.

2000 — Statistical dislocation kinetics is formulated for single crystals.
El-Azab developed Liouville-type kinetic equations for orientation-resolved dislocation distributions, including long-range interactions, short-range reactions, and cross-slip.

2001 — Dislocation mechanics already generates continuum crystal plasticity.
Acharya, A model of crystal plasticity based on the theory of continuously distributed dislocations constructed a continuum theory in which dislocation velocity, dislocation generation, elasticity, and natural constitutive nonlocality enter explicitly.

This is why DI02 should not be posed simply as “derive crystal plasticity from dislocations.”

2002 — DDD can generate hardening laws.
Madec, Devincre & Kubin, From Dislocation Junctions to Forest Hardening used large-scale 3D DDD to obtain a parameter-free estimate of forest strength and its scaling, showing how a macroscopic hardening ingredient can emerge from line-defect interactions.

2009 — Explicit MD → DDD → crystal-plasticity bridging is demonstrated.
Groh et al., Multiscale modeling of the plasticity in an aluminum single crystal used MD to determine dislocation mobility, DDD to determine work hardening, and those results to parameterize a crystal-plasticity law.

2015 — Curved-dislocation kinematics is systematized.
Hochrainer, Multipole expansion of continuum dislocations dynamics in terms of alignment tensors separated two problems: representing dislocation kinematics with manageable continuum variables, and deriving averaged dynamics in those variables. The alignment-tensor hierarchy substantially advances the first problem; closure of the dynamics remains harder.

2016 — Discrete-to-density coarse-graining reveals scale-dependent correlation stresses.
Valdenaire et al., Density-based crystal plasticity: From the discrete to the continuum showed that spatial and statistical averaging of discrete dislocations produces friction- and back-stress terms associated with correlations, and that these contributions depend on coarse-graining scale.

What is already bridged — and what is not

The hierarchy should not be described as a sequence of completely disconnected theories.

Atomistics → dislocation lines

Atomistic calculations routinely provide dislocation core structures, Peierls barriers, mobilities, cross-slip mechanisms, and reaction energetics. Recent machine-learning approaches can transfer large MD datasets into DDD mobility laws.

But a general dynamic derivation of DDD—with mobility, nucleation, junctions, annihilation, and reactions emerging under a controlled limit from atomic dynamics—is not available.

DDD → continuum dislocation dynamics

Statistical and density-based coarse-graining already exist. Kinematics of curved line ensembles is highly developed.

The central difficulty is closure: which density, orientation, curvature, topology, and correlation variables must be retained so that the coarse dynamics predicts its own future?

Dislocation continuum → crystal plasticity

Orowan-type kinematics,

$$\dot\gamma^\alpha = b^\alpha \rho_m^\alpha v^\alpha,$$

and dislocation-density-based hardening already provide strong bridges. Crystal plasticity is therefore an important intermediate continuum description, not the final unresolved endpoint.

Crystal plasticity → macroscopic response

Mean-field, self-consistent, finite-element, and FFT homogenization can map crystal-scale behavior to polycrystal response. These methods successfully predict texture, anisotropy, strength, and heterogeneous fields.

The deeper unresolved question is whether the resulting macroscopic behavior admits a closed reduced continuum law whose state variables and evolution are themselves derived rather than selected phenomenologically.

Recent rigorous progress

2024–2025 — A three-dimensional lattice model converges to line-dislocation energy.
Conti, Garroni & Ortiz, A discrete crystal model in three dimensions: The line-tension limit for dislocations proved, under a dilute scaling, convergence from a discrete three-dimensional crystal energy to a line-tension theory supported on Volterra dislocations.

This is a major controlled atomistic/discrete-to-line-defect result on the energetic side.

2024 — Discrete dislocation evolution can be homogenized in a restricted setting.
Bonicatto & Rindler, Homogenization of elasto-plastic evolutions driven by the flow of dislocations rigorously obtain continuously distributed dislocation-field evolutions as limits of discrete dislocation evolutions in a prototypical small-strain, quasistatic setting.

2026 — Volterra dislocations converge to strain-gradient plasticity in 2D.
Kupferman & Maor, From Volterra dislocations to strain-gradient plasticity derived a finite-strain strain-gradient plasticity energy as a $\Gamma$-limit of two-dimensional bodies containing edge dislocations.

These results demonstrate that theorem-level multiscale passage is possible. They also show clearly how far the realistic three-dimensional dynamic problem still extends beyond current proofs.

Recent progress: correlations, scale, and network state

2024 — Short-range correlations enter continuum kinetics explicitly.
Anderson & El-Azab, Dislocation correlations and the continuum dynamics of the weak line bundle ensemble derived additional continuum driving forces from coarse-grained dislocation correlations, directly confronting information lost by density-only descriptions.

2026 — Pair correlations are extended to curved dislocation systems.
Marx, Sandfeld & Hochrainer developed continuum approximations of pair correlations for curved dislocations and explicitly framed continuum plasticity as a statistical theory of interacting flexible line defects.

2026 — The correct closure depends on observation scale.
Anderson & El-Azab, Scale-dependence and the line bundle regime in continuum dislocation dynamics showed that coarse-graining length changes the amount of line-orientation cancellation and therefore the appropriate continuum representation. A proposed line-bundle closure works only within a finite resolution regime, while a standard maximum-entropy closure performs poorly across the tested scales.

This result sharpens DI02 considerably:

The problem may not be finding one universal closure. The state variables and closure laws may themselves have to transform with scale.

2026 — Network statistics expose candidate hidden variables.
Akhondzadeh et al., Link statistics of dislocation network during strain hardening showed that link-length distributions carry systematic information about active and inactive slip systems and the evolving network.

A complementary 2026 study of 3D DDD networks under tension, shear, bending, and torsion identified link length, pinning, and proximity to free surfaces as useful descriptors and embedded them in a probabilistic continuum plasticity model.

The deeper question is whether these are merely useful empirical descriptors or the correct slow variables selected by coarse-graining.

The frontier: what information must survive coarse-graining?

Let a discrete three-dimensional network be

$$\Gamma^\varepsilon(t) = \bigcup_i \Gamma_i^\varepsilon(t),$$

where the curves carry Burgers vectors, line directions, curvature, and reaction rules.

One may define an empirical line measure on a richer phase space,

$$\mu_t^\varepsilon = \varepsilon^\eta \sum_i \int_{\Gamma_i^\varepsilon(t)} \delta_{ (\mathbf x,\boldsymbol\ell,\kappa,\mathbf b,\ldots)} \,ds .$$

A first controlled goal is

$$\mu_t^\varepsilon \rightharpoonup f( \mathbf x, \boldsymbol\ell, \kappa, \mathbf b, t ),$$

with $f$ satisfying a closed kinetic equation such as

$$\partial_t f + \nabla_{\mathbf x}\!\cdot(\mathbf v f) + \nabla_{\boldsymbol\ell}\!\cdot(\boldsymbol\omega f) + \partial_\kappa(\dot\kappa f) = \mathcal C[f] + \mathcal R[f].$$

Here

• $\mathcal C[f]$ represents correlation effects and coarse-grained interactions;
• $\mathcal R[f]$ represents topology-changing processes such as annihilation, junction formation, multiplication, or cross-slip.

Continuum variables then appear as moments,

$$\rho = \int f, \qquad \boldsymbol\alpha = \int \mathbf b\otimes\boldsymbol\ell\,f, \qquad q = \int \kappa f,$$

together with orientation tensors, junction variables, network statistics, or other quantities if needed.

The crucial question is:

$$\boxed{ \text{Is there a finite set of continuum state variables that closes the dynamics?} }$$

If yes, identify it and derive its equations.

If not, determine what additional orientation distribution, memory, nonlocality, stochasticity, or hierarchy is irreducible.

From dislocation continuum to continuum plasticity

At a coarser level, the target is not necessarily a predetermined crystal-plasticity law.

The emergent continuum response might have the structure

$$\boldsymbol\sigma = \mathcal S ( \boldsymbol\varepsilon, \boldsymbol\varepsilon^p, Z, \nabla Z, \text{history} ),$$

with

$$\dot{\boldsymbol\varepsilon}^p = \mathcal G( \boldsymbol\sigma, Z, \nabla Z,\ldots ), \qquad \dot Z = \mathcal H( \boldsymbol\sigma, Z, \nabla Z,\ldots ).$$

The state $Z$ might contain dislocation density, Nye tensor, slip-system densities, curvature, junctions, texture, network statistics, or other emergent variables.

Or the dynamics may fail to be Markovian in any small state set and require memory,

$$\dot Z(t) = \mathcal H[Z(t)] + \int_0^t K(t-s)Z(s)\,ds+\cdots .$$

A successful derivation should tell us which of these structures is necessary rather than assuming it beforehand.

A useful analogy: kinetic theory

A recent landmark in mathematical physics provides a useful structural analogy. Yu Deng’s 2026 Fields Medal citation includes the rigorous derivation of the Boltzmann equation from hard-sphere dynamics. Deng, Hani & Ma extend the hard-sphere-to-Boltzmann limit to the lifespan of the regular Boltzmann solution, connecting microscopic Newtonian dynamics to a kinetic equation and, through hydrodynamic limits, to continuum fluid equations.

The analogous aspiration here is

$$\text{discrete defect dynamics} \rightarrow \text{kinetic theory of line defects} \rightarrow \text{continuum plasticity}.$$

The analogy is not literal. Dislocations are connected, oriented, long-range-interacting curves that multiply, annihilate, cross-slip, form junctions, and retain strong history.

Those differences are precisely what make DI02 difficult.

What remains open

A complete multiscale theory would need to determine:

• the dynamic limit from atomistic crystal mechanics to moving dislocation lines;
• which DDD mobility and reaction laws emerge rather than being fitted;
• a controlled 3D DDD-to-kinetic or DDD-to-CDD limit for curved multi-slip networks;
• how pair and higher correlations evolve over finite strain;
• how annihilation, multiplication, junction formation, and cross-slip enter as effective reaction operators;
• whether a finite state set can close the continuum dynamics;
• how the state and closure transform with coarse-graining scale;
• which network variables carry plastic memory and hardening;
• how crystal-scale continuum theories reduce to engineering-scale plasticity;
• and whether gradients, nonlocality, stochasticity, or memory are mathematically unavoidable in specific regimes.

The aim is not one universal equation valid at all scales.

The aim is a controlled hierarchy with explicit validity regimes.

What would count as progress?

• A rigorous 3D dynamic DDD-to-continuum limit for curved glide-only dislocation networks.
• A theorem propagating pair correlations over finite deformation rather than imposing an ad hoc closure.
• Identification of a minimal sufficient state for a stated scale regime.
• A no-go theorem proving that a proposed finite state set cannot generate autonomous plastic dynamics.
• A scale-transformation law connecting continuum dislocation variables at different coarse-graining lengths.
• Derivation of junction, annihilation, multiplication, or cross-slip source terms from lower-scale dynamics.
• A dynamic atomistic-to-DDD limit producing mobility and reaction rules with controlled validity.
• Derivation of macroscopic yield, hardening, or flow laws rather than fitting them.
• Quantitative error estimates indicating when MD, DDD, CDD, crystal plasticity, or a macroscopic constitutive description is adequate.

Scientific scope

This page does not claim that continuum dislocation dynamics, field dislocation mechanics, or dislocation-based crystal plasticity are missing. They are mature and active theories.

It also does not claim that no multiscale bridges exist. MD-to-DDD mobility transfer, DDD-derived hardening, DDD-to-density coarse-graining, and crystal-plasticity homogenization are all established in important settings.

Nor is crystal plasticity itself the unresolved endpoint. It is one important continuum representation within a larger scale hierarchy.

The open problem is the controlled derivation and closure of that hierarchy, especially for realistic evolving three-dimensional dislocation networks, and the emergence of a reduced continuum plasticity theory whose state variables and constitutive structure are determined by the underlying defect mechanics.

Key papers and milestones

• J. F. Nye, Some geometrical relations in dislocated crystals (1953) — continuum dislocation-density tensor and lattice curvature.
• I. Groma, Link between the microscopic and mesoscopic length-scale description of the collective behavior of dislocations (1997) — BBGKY-like hierarchy and explicit statistical closure problem.
• A. El-Azab, Statistical mechanics treatment of the evolution of dislocation distributions in single crystals (2000) — kinetic equations for orientation-resolved dislocation distributions.
• A. Acharya, A model of crystal plasticity based on the theory of continuously distributed dislocations (2001) — dislocation-based continuum crystal plasticity and natural nonlocality.
• R. Madec, B. Devincre & L. P. Kubin, From Dislocation Junctions to Forest Hardening (2002) — DDD derivation of forest hardening.
• S. Groh et al., Multiscale modeling of the plasticity in an aluminum single crystal (2009) — explicit MD → DDD → crystal-plasticity bridge.
• T. Hochrainer, Multipole expansion of continuum dislocations dynamics in terms of alignment tensors (2015) — systematic continuum representation of curved-line orientation and curvature.
• P.-L. Valdenaire et al., Density-based crystal plasticity: From the discrete to the continuum (2016) — coarse-graining and scale-dependent correlation stresses.
• N. Bertin, V. V. Bulatov & F. Zhou, Learning dislocation dynamics mobility laws from large-scale MD simulations (2024) — data-driven MD → DDD constitutive bridge.
• J. P. Anderson & A. El-Azab, Dislocation correlations and the continuum dynamics of the weak line bundle ensemble (2024) — correlation-dependent continuum kinetics.
• S. Conti, A. Garroni & M. Ortiz, A discrete crystal model in three dimensions: The line-tension limit for dislocations (2025) — rigorous 3D discrete-crystal to line-dislocation energetic limit.
• R. Kupferman & C. Maor, From Volterra dislocations to strain-gradient plasticity (2026) — rigorous 2D dislocation-to-finite-strain-gradient-plasticity limit.
• A. J. Marx, S. Sandfeld & T. Hochrainer, Continuum approximation of dislocation correlations for systems of curved dislocations (2026) — pair correlations for curved line systems.
• J. P. Anderson & A. El-Azab, Scale-dependence and the line bundle regime in continuum dislocation dynamics (2026) — resolution-dependent state representation and closure.
• Sh. Akhondzadeh et al., Link statistics of dislocation network during strain hardening (2026) — candidate statistical descriptors of evolving network state.

Progress log

30 September 2026 — Problem statement v1.0.
The problem is formulated around the emergence of continuum plasticity rather than the derivation of crystal plasticity alone. Crystal plasticity is treated as an intermediate continuum level. The primary unresolved core is the controlled coarse-graining and closure of realistic three-dimensional dislocation-network dynamics, together with identification of the state variables, scale dependence, memory, and constitutive structure required at larger scales.

Why it matters

Localization is where a continuum constitutive law is tested most severely.

Before localization, a model may reproduce stress–strain curves and homogeneous deformation accurately. After localization, the same local softening model can produce:

• vanishing band thickness under mesh refinement;
• mesh-dependent dissipation;
• nonunique post-bifurcation solutions;
• different failure paths under arbitrarily small perturbations;
• and no intrinsic rule for band propagation or arrest.

These are not merely numerical inconveniences. They signal that information needed to determine the physical continuation is absent from the local theory.

A predictive post-localization theory would have to explain, rather than prescribe:

• the finite band width;
• the jump or internal profile across the band;
• its propagation velocity;
• its dissipation;
• whether one or many bands form;
• whether a band arrests, broadens, or narrows;
• and when localization evolves into damage or fracture.

This matters in high-rate metal failure, machining, impact, geological localization, granular media, amorphous solids, porous materials, and many other inelastic continua.

What we know

Classical continuum instability has a mature theory.
The modern localization framework grew from Hadamard’s ideas on loss of ellipticity and was developed in plasticity by Hill, Mandel, Rice, and others.

Rudnicki & Rice, Conditions for the localization of deformation in pressure-sensitive dilatant materials (1975) derived a general planar localization criterion. In modern notation, an incipient discontinuity with normal $\mathbf n$ becomes admissible when the acoustic tensor loses invertibility,

$$\det \mathbf Q(\mathbf n)=0, \qquad Q_{ik} = n_j C^{\mathrm{ep}}_{ijkl} n_l.$$

The condition identifies a loss of ellipticity of the incremental equilibrium equations and can predict the onset and orientation of localization for broad classes of constitutive models.

But loss of ellipticity is not a post-localization theory.
Once a local softening model loses ellipticity, arbitrarily short localization wavelengths can become admissible. Numerical solutions may then collapse toward the mesh scale and dissipate an amount of energy that changes with discretization.

Modern reviews of elastoplastic localization still emphasize this connection between loss of ellipticity, vanishing localization width, and mesh sensitivity.

Thus a classical criterion answers

When can the homogeneous state cease to be stable?

It does not generally answer

Which finite localized state does the body choose next?

An important caution: loss of ellipticity is not itself the whole answer

It is tempting to equate

$$\det \mathbf Q=0$$

with the inevitable appearance of a particular shear band.

That is too strong.

Triantafyllidis and collaborators showed in microstructured solids that loss of macroscopic ellipticity and the structure of the post-bifurcated solution need not be related in a trivial one-to-one way. Whether localization actually appears, and what form it takes, depends on the nonlinear post-bifurcation mechanics.

PL01 therefore begins after the classical onset calculation.

Why simply adding a length scale is not the solution

Many extended continuum theories can regularize localization.

A schematic gradient model might contain

$$\boldsymbol\sigma = \boldsymbol\sigma_{\mathrm{local}} - \ell^2 \mathcal L(\nabla \boldsymbol\varepsilon^p),$$

while a thermoviscoplastic model may add heat conduction and rate dependence,

$$\rho c\,\dot T = \beta\, \boldsymbol\sigma:\dot{\boldsymbol\varepsilon}^{p} + k\nabla^2T.$$

Other models introduce:

• nonlocal averaging;
• strain gradients;
• dislocation/GND density;
• micromorphic or Cosserat structure;
• viscosity;
• microinertia;
• damage or porosity;
• phase-field variables;
• shear-transformation-zone physics;
• or explicit strong discontinuities.

Any one of these can make the equations better posed and can generate a finite localization scale.

That does not establish that it is the correct physical continuation.

The difficult question is:

$$\boxed{ \text{Which regularized theory is selected by the underlying material physics?} }$$

and, more deeply,

$$\boxed{ \text{Do different physically justified regularizations converge to the same macroscopic post-localization law?} }$$

A useful precedent: kinetic relations for phase boundaries

There is an instructive analogy in nonlinear elasticity with phase transformation.

For nonconvex elastic materials, balance laws and entropy inequalities can admit multiple weak solutions containing moving phase boundaries. Abeyaratne & Knowles showed that a kinetic relation together with a nucleation criterion can select a unique physical evolution from among multiple admissible solutions.

More importantly, they showed that adding small viscosity and strain-gradient effects and then taking the singular limit can generate a particular kinetic relation for the moving interface. Their 1991 analysis therefore connects small-scale regularizing physics to an effective macroscopic interface law.

This does not prove that shear bands obey the same theory.

It suggests a deeper question for plastic localization:

Can the unresolved small-scale physics inside a shear band be replaced, in an appropriate limit, by an effective macroscopic admissibility or kinetic law for the localized interface?

If so, a shear band might eventually be represented not by an arbitrary regularization term but by a lower-dimensional object whose nucleation, motion, jump conditions, and dissipation are inherited from the underlying physics.

A possible mathematical formulation

Consider a family of physically regularized plasticity theories

$$\mathcal P_\varepsilon ,$$

where $\varepsilon$ represents one or more small physical scales:

$$\varepsilon \sim \frac{\ell_{\mathrm{micro}}}{L}, \qquad \frac{\ell_T}{L}, \qquad \frac{\ell_{\mathrm{visc}}}{L}, \ldots .$$

Let the corresponding solutions be

$$u_\varepsilon,\qquad \boldsymbol\varepsilon^p_\varepsilon, \qquad T_\varepsilon,\ldots .$$

Before localization, one expects the regularized and classical theories to agree asymptotically.

After loss of ellipticity, ask whether

$$(u_\varepsilon,\boldsymbol\varepsilon^p_\varepsilon,\ldots) \longrightarrow (u_0,\boldsymbol\varepsilon^p_0,\ldots) \qquad \text{as }\varepsilon\to0$$

in an appropriate weak sense.

The central questions are:

1. Does the limit exist?
2. Is it unique?
3. Does it depend on the type of regularization?
4. What jump, kinetic, or admissibility law does the limit satisfy?
5. Which observables remain finite—band dissipation, propagation speed, strain jump, temperature rise?

A useful reduced description might involve a moving localization surface $S(t)$ with normal velocity $V_n$, jump variables $\llbracket \mathbf v\rrbracket$, and a driving force $\mathcal G_{\mathrm{band}}$, linked by an emergent law such as

$$\boxed{ \mathcal G_{\mathrm{band}} = \Phi( V_n, \llbracket \mathbf v\rrbracket, Z_{\mathrm{band}}, \ldots ) }$$

together with a nucleation criterion.

The form of $\Phi$ should be derived from a small-scale continuum or microstructural limit, not imposed phenomenologically.

Three levels of the problem

I. Onset — comparatively mature

Determine when the homogeneous solution loses stability:

$$\det \mathbf Q(\mathbf n)=0.$$

This level determines a critical state and possible localization directions.

II. Selection — the central theoretical gap

Determine which physical regularization or limiting mechanism selects the post-bifurcation state.

Questions include:

• finite-width band or true discontinuity?
• one band or several?
• stationary or propagating?
• unique continuation or history-dependent selection?
• universal limit or regularization-dependent limit?

III. Evolution — the engineering consequence

Determine the nonlinear fate of the selected localized state:

$$\text{localized plasticity} \rightarrow \begin{cases} \text{stable finite band},\\ \text{traveling band},\\ \text{band interaction},\\ \text{voiding / damage},\\ \text{fracture}. \end{cases}$$

The three levels should not be conflated.

What recent theories already accomplish

Recent work makes the selection question sharper because it demonstrates that different regularizations can each generate plausible finite localized states.

Gradient plasticity can select a finite post-localization width.
Fleck & Hutchinson (2026) studied finite-strain localization with a strain-gradient length and obtained detailed bifurcation and post-localization solutions; in their model class, band width is roughly $10$–$15$ times the material length parameter.

A variational defect energy can select the plastic support itself.
Comella et al. (2026) derived exact solutions in rank-one gradient plasticity in which the extent of the plastic zone is not prescribed. The theory produces distinct nucleation regimes and geometry-dependent scaling of the localized support.

Dislocation mechanics can regularize adiabatic shear without heat conduction.
Singh et al. (2026, preprint) obtained finite-width, mesh-converged shear bands and GND patterning in mesoscale polycrystalline simulations even without heat conduction, through competition between dislocation hardening and thermal softening.

Viscosity and defects can select process-zone width.
Clayton (2025) showed analytically how nonlinear viscous dissipation can prevent an infinitesimal-width band while concentrated defects narrow localization.

These are not competing proofs of one universal mechanism. They demonstrate why PL01 must ask about selection among physically distinct regularized limits.

The frontier: a theory of post-localization selection

The strongest form of PL01 is:

$$\boxed{ \text{classical continuum} \overset{\text{loss of ellipticity}}{\longrightarrow} \text{nonunique / ill-posed local continuation} \overset{\text{small-scale physics}}{\longrightarrow} \text{selected macroscopic evolution}. }$$

A successful theory should tell us when different microscale mechanisms lie in the same effective class.

For example:

$$\text{viscosity} + \text{softening} \quad\stackrel{\varepsilon\to0}{\longrightarrow}\quad \text{kinetic law A},$$

while

$$\text{gradient energy} + \text{softening} \quad\stackrel{\varepsilon\to0}{\longrightarrow}\quad \text{kinetic law B}.$$

If A and B coincide over a regime, that would reveal a form of universality.

If they do not, then the material’s small-scale physics remains irreducibly part of the continuum selection law.

Either outcome would be important.

What remains open

A general post-localization theory still needs to determine:

• whether physically different regularizations select the same or different macroscopic limits;
• which small-scale parameters survive the limit as effective kinetic or interface quantities;
• whether a unique admissibility criterion exists after loss of ellipticity;
• when the limiting object is a finite-width band versus a displacement/strain discontinuity;
• how nucleation should be distinguished from subsequent propagation;
• how band propagation velocity and dissipation are determined;
• whether post-localization evolution can be represented by a lower-dimensional kinetic law;
• how material memory and pre-existing imperfections enter the selection;
• how the selected plastic band transitions into damage or fracture;
• and which parts of this structure are universal across constitutive classes.

The central issue is not to predict the exact width of every shear band in every material.

It is to determine the continuum selection principle that replaces the classical local theory once that theory ceases to be sufficient.

What would count as progress?

• A proof that a physically motivated family of regularized plasticity models converges to a unique post-localization evolution.
• Derivation of an effective kinetic relation for a propagating shear band from a finite-width thermoviscous, gradient, or dislocation-based model.
• A classification showing which different regularizations belong to the same post-localization universality class.
• A counterexample showing that two equally admissible small-scale regularizations select genuinely different macroscopic continuations.
• A no-go theorem proving that balance laws, the local constitutive law, and the dissipation inequality are insufficient to select the post-localization solution.
• A theory separating nucleation, propagation, arrest, and transition to fracture within one consistent framework.
• Controlled asymptotic matching between a finite-width internal band structure and an effective lower-dimensional discontinuity law.
• Experimental tests designed to distinguish competing kinetic/admissibility laws rather than merely fit band width.

Scientific scope

This page does not claim that plastic localization onset is unsolved. Loss-of-ellipticity and bifurcation criteria are mature and powerful tools.

It also does not claim that every loss of ellipticity necessarily produces a particular localized post-bifurcation mode.

Nor does it claim that gradient, nonlocal, viscous,dislocation-based, or damage regularizations are missing. Many such theories already generate well-posed finite-width localization.

The open problem is one level deeper:

Which of those regularizations represents the relevant unresolved physics, and what macroscopic selection law remains when the small regularizing scales are eliminated?

The analogy to phase-boundary kinetics is structural, not an assertion that a plastic shear band is a thermodynamic phase boundary.

Finally, material dependence is expected. The theoretical goal is not to remove material parameters, but to determine which material information is relevant to the limiting post-localization law and which details disappear under coarse-graining.

Key papers and milestones

• J. W. Rudnicki & J. R. Rice, Conditions for the localization of deformation in pressure-sensitive dilatant materials (1975) — general localization/bifurcation criterion and loss of ellipticity.
• J. R. Rice, The localization of plastic deformation (1976) — broader continuum framework for plastic localization.
• R. Abeyaratne & J. K. Knowles, Kinetic relations and the propagation of phase boundaries in solids (1991) — kinetic relation and nucleation criterion as additional selection information for nonunique continuum evolution.
• R. Abeyaratne & J. K. Knowles, Implications of Viscosity and Strain-Gradient Effects for the Kinetics of Propagating Phase Boundaries in Solids (1991) — small-scale regularization generates an effective kinetic relation.
• M. P. Santisi d’Avila, N. Triantafyllidis & G. Wen, Localization of deformation and loss of macroscopic ellipticity in microstructured solids (2016) — clarifies the nontrivial relation between loss of ellipticity and actual post-bifurcation localization.
• K. Viswanathan, S. Yadav & D. Sagapuram, Shear Bands in Materials Processing: Understanding the Mechanics of Flow Localization From Zener’s Time to the Present (2020) — modern review of shear-band mechanics and the post-localization challenge.
• J. D. Clayton, Analytical Shear-Band Process Zone Model Incorporating Nonlinear Viscous Effects and Initial Defects (2025) — viscous/defect regularization of finite-width bands.
• N. A. Fleck & J. W. Hutchinson, Shear band localization in finite strain deformation theory with strain gradient effects (2026) — finite-width post-localization selected by a gradient length.
• S. Singh et al., Finite-width adiabatic shear banding and dislocation patterning in mesoscale polycrystalline aggregates (2026, preprint) — dislocation/GND-based regularization and finite-width dynamic bands.
• M. C. Comella et al., Nucleation in rank-one gradient plasticity: Exact solutions and geometry-dependent regimes (2026) — variational selection of finite plastic support and geometry-dependent regimes.

Progress log

30 September 2026 — Problem statement v2.0.
PL01 is reformulated from a material-specific band-width question into a continuum-selection problem. The classical loss-of-ellipticity criterion is treated as the starting point, not the unresolved target. The open core is whether physically motivated small-scale regularizations select a unique effective post-localization evolution, and whether that evolution can be represented by an emergent admissibility or kinetic law for localized plastic flow.

Why it matters

A finite deformation is the actual map

$$\mathbf y:\Omega\rightarrow\mathbb R^3$$

that tells us where every material point goes.

A physically acceptable theory must therefore control not only stress and strain, but also:

• orientation;
• injectivity and non-interpenetration;
• creation of cavities or new surfaces;
• topology of the deformed body;
• regularity of the inverse deformation;
• and concentration of energy into lower-dimensional structures.

A singularity can therefore mean very different things:

$$\boxed{ \begin{array}{ll} \textbf{Physical singularity:} & \text{an idealized defect or instability with meaningful mechanics;} \\[2mm] \textbf{Pathological singular limit:} & \text{a mathematically accessible weak limit that escapes the intended kinematics of matter.} \end{array}}$$

The mechanics problem is to distinguish them.

Linear elasticity provides the model example

For a crack, notch, interface corner, or point load, linear elasticity often produces

$$\mathbf u(r,\theta)\sim r^\lambda \mathbf U(\theta), \qquad \boldsymbol\sigma(r,\theta)\sim r^{\lambda-1}\boldsymbol\Sigma(\theta).$$

For a classical crack,

$$\lambda=\frac12, \qquad \sigma\sim r^{-1/2}.$$

The stress is singular, but the near-tip elastic energy is locally integrable. Mechanics keeps the singular outer solution and supplements it with additional physics at smaller scales.

The intellectual pattern is

$$\boxed{ \text{bulk elasticity} + \text{identified defect} + \text{defect law}. }$$

For fracture, that means elasticity plus crack geometry plus fracture/surface energy.

Finite elasticity raises a harder possibility: the singular structure itself may emerge from the variational problem rather than being prescribed beforehand.

Why finite elasticity is harder

In finite elasticity,

$$\mathbf F=\nabla\mathbf y, \qquad \det\mathbf F>0,$$

and the deformation map determines the current geometry.

Local orientation is not enough

The condition

$$\det\nabla \mathbf y>0$$

prevents local orientation reversal almost everywhere, but does not by itself encode all global/topological constraints on matter.

Weak convergence can lose geometric information

Energy bounds often provide only weak compactness:

$$\mathbf y_k\rightharpoonup\mathbf y.$$

Under this limit, inverse regularity, cofactors, distributional determinants, and topology can develop singular parts.

Thus

$$\boxed{ \text{regular admissible sequence} \not\Rightarrow \text{regular admissible weak limit}. }$$

Singularities can emerge from minimization

In linear fracture mechanics, one normally specifies a crack and asks for the field around it.

In nonlinear elasticity one may instead have

$$\text{regular/invertible deformations} \rightarrow \text{increasing concentration} \rightarrow \text{singular limiting deformation},$$

even though no singular defect was prescribed initially.

This makes interpretation of the singularity part of the theory itself.

Singularities with established physical meaning: cavitation

Finite elasticity already contains physically meaningful singular solutions.

In Ball’s 1982 cavitation theory, a sufficiently stretched nonlinear-elastic solid can develop a cavity. The deformation becomes singular at the cavity center, yet the phenomenon has a clear mechanics interpretation and is connected to internal rupture of rubber-like materials.

This is an important precedent:

$$\boxed{ \text{singular finite deformation} \neq \text{automatically unphysical}. }$$

Modern variational theories can explicitly assign energetic cost to newly created surfaces.

For example, Bresciani & Friedrich (2026) study singular cavitating minimizers with an energy that penalizes both cavity volume and perimeter and prove a core-radius $\Gamma$-convergence result.

So the open problem is not to explain every singularity in finite elasticity from scratch.

The unresolved frontier is what to do when a singular weak limit appears without an established physical defect interpretation.

Canonical benchmark: the 3D neo-Hookean problem

Consider

$$E[\mathbf y] = \int_\Omega \left[ |\nabla\mathbf y|^2 + H(\det\nabla\mathbf y) \right]\,d\mathbf X,$$

with strong penalization of extreme volume change and orientation preservation.

Finite energy gives only

$$\mathbf y\in W^{1,2}(\Omega;\mathbb R^3).$$

In three dimensions,

$$p=2=n-1$$

is a critical regularity regime for important topological and inverse-map properties.

The direct-method question becomes

$$\begin{array}{c} \text{regular, noncavitating/invertible minimizing sequence} \\ \downarrow \\ H^1\text{ weak limit} \\ \downarrow \\ \boxed{\text{does the limit still represent the intended physical class?}} \end{array}$$

In general, not necessarily.

The neo-Hookean dipole

Conti & De Lellis (2003) constructed a pathological three-dimensional weak-limit mechanism involving a dipole-type singularity.

The issue should not simply be described as “the weak limit self-interpenetrates.”

Recent results show that weak limits of Sobolev homeomorphisms can remain one-to-one almost everywhere under very weak assumptions when the limiting Jacobian is positive. Bouchala, Hencl & Zhu (2025) establish this in dimensions two and three at very low Sobolev exponents.

Therefore

$$\boxed{ \text{a.e. injectivity} \neq \text{full physical/topological admissibility}. }$$

The dipole pathology concerns stronger information: condition (INV), divergence identities, inverse regularity, and the correct relation between a material region and its deformed boundary.

A weak map can therefore retain positive Jacobian and strong measure-theoretic injectivity properties while losing topological information carried by the approximating regular deformations.

A hierarchy of admissibility

Local orientation preservation

$$\det\nabla y>0 \quad\text{a.e.}$$

Necessary, but not sufficient.

Almost-everywhere injectivity

Most target points should not possess multiple material preimages.

Recent mathematics shows that this can survive surprisingly weak limits.

Global/topological admissibility

Conditions such as the Ciarlet–Nečas condition and Müller–Spector condition (INV) encode stronger forms of non-interpenetration and the relation between regions and their images.

Regular inverse

If

$$y^{-1}$$

belongs to an appropriate Sobolev or $BV$ class, the inverse retains geometric information that the forward map may lose under weak convergence.

The neo-Hookean relaxation problem sits in the gap between these levels.

What relaxation theory has discovered

2023 — Dipoles acquire an explicit energetic interpretation.
Barchiesi, Henao, Mora-Corral & Rodiac studied harmonic dipoles and the relaxation of the 3D neo-Hookean problem and derived an explicit energetic penalty for the singular structure.

2024 — A modified energy restores compactness and existence.
A Relaxation Approach to the Minimization of the Neo-Hookean Energy in 3D enlarges the minimization space and introduces a functional that penalizes singularities of the inverse deformation.

A concrete expression arising in this program is

$$\boxed{ F[y] = \int_\Omega \left( |\nabla y|^2 + H(|\det\nabla y|) \right)\,dX + 2\|D^s y^{-1}\|. }$$

Here $D^s y^{-1}$ is the singular part of the distributional derivative of the inverse map.

Within the developed framework, the coefficient $2$ is not a fitted constitutive parameter: it arises from the concentration mechanism and is sharp for the Conti–De Lellis dipole construction.

The weak limit appears to have lost memory of energy concentrated in increasingly fine regular deformations; the singular inverse term restores that energetic information.

This raises a central mechanics question:

Is the inverse-map singular energy a physically meaningful defect energy, or mathematical bookkeeping needed to penalize a nonphysical relaxation pathway?

Recent progress in admissibility

2025 — Weak limits can remain one-to-one almost everywhere.
Bouchala, Hencl & Zhu prove that a.e. injectivity can survive weak limits of Sobolev homeomorphisms under very weak assumptions, provided the limiting Jacobian is positive.

This rules out an overly simple interpretation of the pathology as mere loss of injectivity.

2025 — Minimizers exist in important weakly closed noncavitating classes.
Kalayanamit proves existence and inverse-regularity results for neo-Hookean minimizers satisfying condition (INV) in important weak-closure settings.

These results clarify what weak limits can preserve, but do not identify the original regular class with its weak closure.

2026: the Lavrentiev gap makes the distinction unavoidable

Barchiesi, Henao, Mora-Corral & Rodiac, A Lavrentiev phenomenon in the neo-Hookean model construct boundary data for which

$$\boxed{ \inf_{y\in\mathcal A_{\mathrm{reg}}}E[y] > \inf_{y\in\overline{\mathcal A_{\mathrm{reg}}}^{\,H^1_{\mathrm{weak}}}}E[y]. }$$

The lower energy in the weak closure is produced by a dipole-type singular mechanism.

This is stronger than saying regularity is difficult to prove.

It shows that the enlarged weak problem can genuinely prefer a singular state whose energy lies below every deformation in the original regular class.

Therefore

$$\text{enlarge the space} \rightarrow \text{find a minimizer} \rightarrow \text{prove that it was regular after all}$$

cannot succeed for every boundary condition.

The distinction between regular physical deformation and relaxed singular state is structural.

The deeper mechanics question

The crack-tip example teaches

$$\text{a singularity can be useful if we understand what it represents}.$$

The neo-Hookean dipole asks the converse:

$$\boxed{ \text{What evidence is required before a mathematical singularity should be admitted as a physical state?} }$$

A candidate singularity should be tested against questions such as:

• Is total energy finite?
• Is local orientation preserved?
• Is the deformation one-to-one almost everywhere?
• Does it preserve the stronger topology expected of matter?
• Can it be approached by physically admissible finite-scale states?
• What energy concentrates in that approximating sequence?
• Does the singular state require an additional surface/line/point or inverse-map energy?
• Does a microscopic or higher-gradient model preserve the singularity or eliminate it?
• Does it correspond to a recognizable defect such as a cavity, crack, interface, or phase boundary?

No single criterion is enough.

The target is a coherent admissibility theory.

The frontier

The broad mechanics problem is

$$\boxed{ \text{Which finite-energy singularities are admissible states of a finite-elastic body?} }$$

The canonical benchmark is

$$\boxed{ \begin{array}{c} \text{3D neo-Hookean energy at }W^{1,2} \\ \downarrow\\ \text{regular admissible class not weakly compact} \\ \downarrow\\ \text{dipole singularities in weak closure} \\ \downarrow\\ \text{possible Lavrentiev gap} \\ \downarrow\\ \text{determine the correct admissibility/relaxation theory}. \end{array}}$$

The eventual continuum completion may be

$$\text{bulk elasticity}+\text{defect energy},$$

as in fracture and cavitation,

or instead

$$\text{bulk elasticity}+\text{stronger admissibility constraint}$$

that excludes a pathological pathway.

Different singularity classes may require different answers.

What remains open

For the 3D neo-Hookean benchmark

• determine the exact relaxed energy in general three dimensions;
• classify singular structures generated by bounded-energy regular sequences;
• determine whether dipoles exhaust the relevant noncompactness mechanisms;
• characterize when the Lavrentiev gap vanishes and when it is positive;
• identify boundary data and geometries that trigger or exclude singular relaxation;
• clarify the relation among positive Jacobian, a.e. injectivity, condition (INV), inverse regularity, and full physical admissibility;
• determine whether the singular inverse energy has a direct mechanical interpretation;
• construct numerical schemes that converge to the intended regular or relaxed theory rather than a discretization-selected state.

For the broader singularity question

• establish criteria distinguishing physical defects from pathological weak limits;
• determine which singularities require new surface, line, point, or inverse-map energies;
• identify which singularities disappear when unresolved small-scale physics is introduced;
• and determine whether different singularities fall into recognizable classes with common mechanics.

What would count as progress?

• A general 3D characterization of the relaxed neo-Hookean energy.
• Classification of finite-energy singular defect measures generated by regular minimizing sequences.
• Necessary and sufficient conditions for closure of the physical admissible class.
• A criterion predicting whether a given singularity should be admitted, penalized, or excluded.
• Derivation of the singular defect term from a microscopic or higher-order continuum model.
• A proof that the dipole term represents a genuine physical defect—or that physically consistent regularizations eliminate it.
• Classification of boundary conditions for which a Lavrentiev gap occurs.
• A convergent computational formulation that distinguishes a regular minimizer, cavitating solution, and relaxed dipole state.
• A finite-elastic analogue of the successful fracture-mechanics logic: $$\text{identify singular field} \rightarrow \text{extract finite defect quantity} \rightarrow \text{supply physical defect law}.$$

Scientific scope

This page does not claim that singularities in finite elasticity are generally unexplained.

Cavitation and fracture already possess substantial physical and mathematical theories.

It does not claim that the neo-Hookean dipole is simply self-interpenetration. Positive Jacobian and even almost-everywhere injectivity can coexist with failure of stronger topological admissibility.

It also does not claim that every weak singularity is physical merely because it has finite or relaxed energy. Determining the additional admissibility criteria is precisely the open problem.

The canonical energy

$$|F|^2+H(\det F)$$

is a mathematical neo-Hookean-type benchmark; engineering models called “neo-Hookean” need not use exactly this form.

Finally, NL01 is a variational/quasistatic problem. Dynamic blow-up phenomena in other PDEs are only conceptual analogies and are not part of the problem statement.

Key papers and milestones

• J. M. Ball, Convexity conditions and existence theorems in nonlinear elasticity (1977) — polyconvexity and the direct-method framework.
• J. M. Ball, Global invertibility of Sobolev functions and the interpenetration of matter (1981) — global invertibility under stronger Sobolev regularity.
• J. M. Ball, Discontinuous equilibrium solutions and cavitation in nonlinear elasticity (1982) — physically interpretable cavitation singularities.
• P. G. Ciarlet & J. Nečas, Injectivity and self-contact in nonlinear elasticity (1987) — global injectivity/non-interpenetration.
• S. Müller & S. J. Spector, An existence theory for nonlinear elasticity that allows for cavitation (1995) — condition (INV) and lower-regularity admissibility.
• S. Conti & C. De Lellis, Some remarks on the theory of elasticity for compressible Neohookean materials (2003) — dipole-type lack of compactness.
• M. Barchiesi et al., Harmonic dipoles and the relaxation of the neo-Hookean energy in 3D elasticity (2023) — dipole energy and relaxation structure.
• M. Barchiesi et al., A Relaxation Approach to the Minimization of the Neo-Hookean Energy in 3D (2024) — enlarged admissible space and inverse-singularity energy.
• O. Bouchala, S. Hencl & Z. Zhu, Weak limits of Sobolev homeomorphisms are one to one almost everywhere (2025) — preservation of a.e. injectivity under weak convergence.
• P. Kalayanamit, Sobolev regularity of the inverse for minimizers of the neo-Hookean energy satisfying condition INV (2025) — existence and inverse regularity in important weakly closed classes.
• M. Bresciani & M. Friedrich, Core-Radius Approximation of Singular Minimizers in Nonlinear Elasticity (2026) — cavitation with volume/perimeter cost and $\Gamma$-convergent core regularization.
• M. Barchiesi et al., A Lavrentiev phenomenon in the neo-Hookean model (2026, preprint) — strict gap between a regular class and its weak closure for suitable boundary data.

Progress log

30 September 2026 — Problem statement v2.0.
NL01 is reframed from a narrow neo-Hookean minimizer-existence question into a mechanics problem about physical admissibility of singularities in finite elasticity. Crack-tip mechanics and cavitation provide examples of singularities with established physical interpretation. The three-dimensional neo-Hookean dipole/Lavrentiev problem is retained as the precise unresolved benchmark. The page now distinguishes positive Jacobian and a.e. injectivity from stronger topological admissibility, and treats the inverse-map defect term as a mathematically established relaxation mechanism whose physical interpretation remains part of the open problem.

Why it matters

Thin shells achieve exceptionally high stiffness and load capacity at low mass. That is why they appear in:

• launch-vehicle tanks and interstages;
• aircraft fuselages;
• pressure vessels;
• submarine and marine structures;
• storage tanks and silos;
• lightweight civil structures.

But the ideal cylinder is one of the most imperfection-sensitive structures in mechanics.

Historically, experimental collapse loads could lie dramatically below the perfect-shell bifurcation prediction. The gap was sufficiently large that aerospace design adopted empirical lower-bound curves rather than trusting the classical theoretical load directly.

Modern manufacturing, metrology, experiments, and nonlinear finite-element analysis have improved prediction enormously. NASA’s current SP-8007 Rev. 2 incorporates the results of the Shell Buckling Knockdown Factor program and modern design practice.

Yet an underlying theoretical question remains:

$$\boxed{ \text{How should a manufacturing tolerance be converted into a guaranteed stability margin?} }$$

What we know: the perfect-shell problem is not the open question

For a long thin isotropic cylinder, classical shell theory gives a clean ideal buckling stress of the form

$$\sigma_{\mathrm{cl}} \sim \frac{E}{\sqrt{3(1-\nu^2)}} \frac{t}{R},$$

under the ideal assumptions associated with the classical result.

The historical difficulty was that real cylinders often buckled at much smaller stresses.

von Kármán & Tsien (1941) recognized that nonlinear post-buckling mechanics was essential to understanding the discrepancy.

Koiter’s 1945 stability theory provided the fundamental explanation: axially compressed cylinders have a strongly unstable initial post-buckling response. Small imperfections can therefore convert the ideal bifurcation into a substantially lower limit point.

Thus the classical theoretical lesson is already clear:

$$\boxed{ \text{unstable post-buckling} \Rightarrow \text{strong imperfection sensitivity}. }$$

SH01 begins one step later.

From “imperfection sensitivity” to “which imperfection?”

An initial radial imperfection can be written as

$$r(\theta,z) = R+\eta(\theta,z).$$

For a particular known $\eta$, high-fidelity nonlinear analysis can often predict the corresponding collapse load,

$$P_{\mathrm{cr}}=\mathcal P[\eta].$$

The difficulty appears before that calculation:

Which $\eta$ must the structure be safe against?

Imperfections differ not only in amplitude, but also in:

• axial and circumferential wavelength;
• localization;
• slope and curvature;
• spectral content;
• correlation length;
• location relative to edges;
• interaction with welds, stiffeners, thickness variation, and load introduction;
• and combinations of multiple defects.

Therefore

$$\|\eta\|_\infty$$

alone does not define severity.

Two imperfections with the same maximum radial deviation can trigger completely different nonlinear instability paths.

The robust lower-bound formulation

Let

$$\mathcal A_\delta$$

be a physically admissible class of imperfections consistent with a declared manufacturing and assembly tolerance.

Define

$$\boxed{ P_*(\delta) = \inf_{\eta\in\mathcal A_\delta} P_{\mathrm{cr}}[\eta] }$$

and the corresponding mechanics-based knockdown factor

$$\boxed{ \kappa_*(\delta) = \frac{P_*(\delta)}{P_{\mathrm{cl}}}. }$$

This formulation exposes the real theoretical problem.

A complete theory must determine:

1. what belongs to $\mathcal A_\delta$;
2. whether the infimum exists;
3. what imperfection approaches or attains it;
4. which nonlinear equilibrium/bifurcation causes collapse;
5. how the result scales with shell geometry and tolerance.

A “worst imperfection” is meaningful only after the admissible class has been defined physically.

Why the imperfection metric matters

Suppose we impose only

$$\|\eta\|_\infty\le\delta.$$

This bounds the maximum radial deviation, but says nothing about wavelength or curvature.

A narrow sharp defect and a broad smooth dimple can have the same amplitude while producing very different shell mechanics.

An $L^2$ or rms bound has the opposite problem: a very localized severe defect may contribute little to the global norm.

A physically meaningful admissible set may therefore require several descriptors, for example

$$\frac{\|\eta\|_\infty}{t}\le a, \qquad \|\nabla\eta\|\le s_{\max}, \qquad \ell_{\min} \le \ell_{\mathrm{corr}} \le \ell_{\max},$$

together with spectral constraints or statistics inferred from measured manufacturing data.

The correct metric is not a mathematical afterthought.

It is part of the mechanics.

The nonlinear energy landscape

A major advance came from studying unstable localized equilibria of the perfect shell.

2006 — A single dimple appears as a mountain-pass state.
Horák, Lord & Peletier examined the perfect-cylinder energy landscape and identified a mountain-pass equilibrium separating the unbuckled state from lower-energy post-buckling states. Numerically, the lowest-energy mountain-pass state is a localized single dimple.

This introduces a different picture of imperfection sensitivity:

$$\text{imperfection} \quad\text{does not create the dangerous state from nothing;}$$

it can steer the shell toward a nonlinear localized state already embedded in the perfect-shell landscape.

2017 — Fully localized edge states.
Kreilos & Schneider computed localized equilibria for a finite clamped cylinder. A single dimple lies on the stability boundary of the unbuckled state, and continuation reveals homoclinic snaking into multi-dimple states.

This gives the lower-bound problem a possible mechanistic basis:

$$\boxed{ \text{worst imperfection} \;\leftrightarrow\; \text{most accessible dangerous localized equilibrium?} }$$

That connection remains incomplete.

Why numerical worst-case optimization does not close the problem

There is already a substantial literature on finding severe imperfection shapes.

Deml & Wunderlich (1997) developed a direct nonlinear finite-element method for computing a worst imperfection.

Modern methods include:

• linear-mode combinations;
• single perturbation load approach;
• multiple/worst multiple perturbation load approach;
• Monte Carlo measured-imperfection models;
• Koiter reduced-order optimization;
• surrogate-assisted robust optimization.

For a selected finite-dimensional family, one can solve problems such as

$$\min_{\eta\in V_N,\;\|\eta\|_\infty\le\delta} P_{\mathrm{cr}}[\eta].$$

But this determines the worst imperfection inside $V_N$.

It does not establish that $V_N$ contains every physically relevant imperfection.

Conversely, an unconstrained mathematical optimization may find a shape that manufacturing could never produce.

The open problem is therefore not the optimization algorithm.

It is defining and understanding the physical uncertainty set over which the optimization should be performed.

Recent progress: defect geometry can switch the buckling mechanism

September 2026 — The origin of sensitivity is traced to bifurcation switching.
Yang & Schneider, The Origin of Imperfection Sensitivity in the Buckling of Cylindrical Shells systematically vary the depth and size of one localized Gaussian defect using continuation and bifurcation analysis.

They find that localized equilibria already present in the perfect shell become pinned by the defect.

Small changes in defect geometry can change which bifurcation triggers buckling. As a result,

$$P_{\mathrm{cr}}$$

can vary non-monotonically and discontinuously even when the imperfection geometry changes smoothly.

The key implication is

$$\boxed{ \text{imperfection sensitivity} \neq \text{sensitivity to one scalar imperfection amplitude}. }$$

The topology of the nonlinear equilibrium branches matters.

Recent progress: local defects interact nonlocally

July 2026 — The “butterfly defect” effect.
Ubamanyu, Blond, Hutchinson & Reis study multiple localized imperfections and their interaction with shell boundaries.

A local dimple produces a spatially extended butterfly-like deformation pattern. Consequently:

• nearby defects interact even when their cores are localized;
• defects interact with clamped edges;
• the deepest defect need not trigger collapse;
• increasing the defect population produces a statistical size effect;
• the shell length-to-thickness ratio $H/t$ becomes important in addition to $R/t$.

Their calculated knockdown factors still remain above some historical experimental values.

That result is informative: a Gaussian dimple is a natural severe imperfection, but it is not established as the universal worst-case defect for cylinders.

Recent progress: engineering lower envelopes are becoming explicit

2026 — Numerical lower-bound KDF envelopes.
Wang et al., A numerical framework for lower-bound knockdown factor prediction of compressed cylindrical shells combine the Worst Multiple Perturbation Load Approach with surrogate-assisted FE evaluation and multi-gene genetic programming.

They construct explicit tolerance-aware lower-bound relations over ranges of

$$R/t,\qquad L/R,\qquad \text{imperfection amplitude}.$$

This is substantial progress toward practical prediction.

But the framework still defines its lower envelope through a selected perturbation representation.

The remaining theoretical question is whether that representation spans the full physically admissible imperfection class relevant to the design guarantee.

What remains open

A complete mechanics-based knockdown theory should answer:

• What geometric/loading descriptors define a realistic admissible imperfection class?
• Which norms, correlation lengths, spectral constraints, or manufacturing statistics are physically necessary?
• Does a worst-case imperfection exist for that class?
• Is it localized, distributed, or multi-defect?
• Is it related systematically to a mountain-pass/edge state of the perfect shell?
• When does the identity of the triggering bifurcation change?
• Can those switches produce non-smooth or discontinuous lower-envelope curves?
• How should geometric, loading, boundary, thickness, and material imperfections be combined?
• How does the lower envelope scale as $$t/R\rightarrow0$$ and with $$\delta/t,\quad L/R?$$
• How can a statistical manufacturing description be converted into a deterministic or reliability-based design margin without excessive conservatism?

A target theory

The desired result is not one universal number.

It is something like

$$\boxed{ \kappa_* = \Phi \left( \frac{t}{R}, \frac{L}{R}, \frac{\delta}{t}, \mathcal S_{\mathrm{imp}}, \mathcal B \right), }$$

where

• $\mathcal S_{\mathrm{imp}}$ describes the physically admissible spatial/spectral structure of imperfections;
• $\mathcal B$ describes boundary and loading conditions.

Ideally the theory would also identify

$$\eta_* \in\mathcal A_\delta$$

or a family of asymptotically optimal imperfections and connect $\eta_*$to the nonlinear localized state that controls collapse.

What would count as progress?

• A manufacturing-derived imperfection metric rather than an assumed modal/dimple family.
• Proof or asymptotic characterization of the robust lower envelope.
• Identification of the minimizing imperfection and its nonlinear triggering branch.
• A theorem relating dangerous imperfections to localized mountain-pass or edge states.
• A classification of bifurcation switches that create discontinuous changes in buckling load.
• A theory of nonlocal defect-defect and defect-edge interaction.
• A controlled thin-shell scaling law for the lower envelope.
• A connection between measured imperfection statistics and guaranteed/reliability-based KDFs.
• Validation across different shell radii, thicknesses, lengths, and manufacturing routes without redefining the imperfection family for every experiment.

Why this is not just an empirical or materials problem

The issue survives for an isotropic linearly elastic material.

Its difficulty comes from structural nonlinearities:

• strongly subcritical post-buckling;
• many nearly degenerate modes;
• symmetry breaking;
• localized unstable equilibria;
• nonlocal interactions;
• boundary sensitivity;
• switching between nonlinear bifurcation routes.

The material law can remain simple while the stability landscape is extraordinarily complex.

That makes the cylinder a canonical problem of nonlinear structural mechanics.

Scientific scope

This page does not claim that modern shell buckling loads cannot be predicted accurately. When measured geometry, loading, and boundary details are available, high-fidelity FE models often correlate well with tests.

It does not claim that NASA SP-8007 is obsolete. Rev. 2 explicitly updates the historical guidance using modern NASA/industry knowledge.

It also does not claim that no worst-case methods exist. Many deterministic and stochastic methods do.

The unresolved issue is what makes a worst-case calculation physically complete rather than basis-dependent.

A Gaussian dimple, linear buckling eigenmode, or perturbation-load pattern is therefore a model of an imperfection family—not automatically the universal worst imperfection.

Finally, a probabilistic low quantile and a deterministic infimum are different objects and should not be conflated.

Key papers and milestones

• T. von Kármán & H.-S. Tsien, The Buckling of Thin Cylindrical Shells Under Axial Compression (1941) — nonlinear post-buckling explanation of the classical discrepancy.
• W. T. Koiter, Over de stabiliteit van het elastisch evenwicht (1945) — general nonlinear stability and imperfection-sensitivity theory.
• N. Yamaki, Elastic Stability of Circular Cylindrical Shells (1984) — comprehensive theory/experiment synthesis.
• M. Deml & W. Wunderlich, Direct evaluation of the ‘worst’ imperfection shape in shell buckling (1997) — direct worst-imperfection optimization.
• J. Horák, G. J. Lord & M. A. Peletier, Cylinder Buckling: The Mountain Pass as an Organizing Center (2006) — energy barrier and localized single-dimple mountain pass.
• J. W. Hutchinson, Knockdown factors for buckling of cylindrical and spherical shells subject to reduced biaxial membrane stress (2010) — mechanics of loading-state dependence of knockdown.
• T. Kreilos & T. M. Schneider, Fully localized post-buckling states of cylindrical shells under axial compression (2017) — localized edge states and homoclinic snaking.
• H. N. R. Wagner et al., Robust knockdown factors for the design of cylindrical shells under axial compression (2018) — modern deterministic/probabilistic robust KDF design.
• NASA/SP-8007-2020/REV 2, Buckling of Thin-Walled Circular Cylinders (2020) — current NASA design guidance.
• T. Yang & T. M. Schneider, The Origin of Imperfection Sensitivity in the Buckling of Cylindrical Shells (2026, preprint) — defect pinning and bifurcation switching.
• U. K. Ubamanyu et al., The Butterfly Defect Effect in Cylindrical Shell Buckling (2026, preprint) — nonlocal defect interactions, edge coupling, and statistical size effect.
• F. Wang et al., A numerical framework for lower-bound knockdown factor prediction of compressed cylindrical shells (2026) — explicit numerical lower-bound envelope construction.

Progress log

30 September 2026 — Problem statement v1.0.
SH01 is formulated as a robust nonlinear-stability problem rather than a request for another empirical knockdown curve. Koiter theory explains why the cylinder is imperfection sensitive; modern localized-state theory identifies dangerous equilibria; 2026 continuation results show that defect geometry can switch the triggering bifurcation; and modern numerical frameworks construct practical lower envelopes. The unresolved core is to define the physically admissible imperfection class and derive/validate the lower buckling envelope over that class.

Why it matters

Dry friction is among the simplest and oldest constitutive laws in mechanics. Elasticity is equally fundamental.

Their coupling controls:

• brakes and friction-induced vibration;
• bolted and mechanical joints;
• press fits and seals;
• tires and sliding interfaces;
• forming processes;
• particulate contacts;
• robotic grasping and manipulation.

Yet the continuum problem is mathematically subtle because the state of the interface is itself unknown.

At each contact point one must determine simultaneously:

$$\boxed{ \text{contact or separation} \quad+\quad \text{normal pressure} \quad+\quad \text{stick or slip} \quad+\quad \text{friction traction}. }$$

The problem is therefore a canonical example of how elementary local mechanical laws can produce a difficult global continuum theory.

The Signorini–Coulomb problem

Let $\mathbf u$ be the displacement and let the contact traction be decomposed into

$$\mathbf t = t_n\mathbf n+\mathbf t_t.$$

On the possible contact boundary $\Gamma_C$, unilateral Signorini contact requires

$$u_n-g\le 0, \qquad t_n\le 0, \qquad t_n(u_n-g)=0.$$

These conditions mean that the body cannot penetrate the obstacle, the obstacle can only push on the body, and contact pressure acts only where the gap is closed.

Coulomb friction adds

$$|\mathbf t_t| \le -f\,t_n.$$

Inside this friction cone the interface may stick. On its boundary it may slide, with tangential traction opposing the slip direction.

The difficulty is visible immediately:

$$\boxed{ \text{friction threshold} = f\times \text{unknown contact pressure}. }$$

Why Coulomb friction is harder than prescribed friction

If the tangential friction threshold is prescribed independently of the normal traction, one obtains the classical Tresca problem. Its variational structure is comparatively well behaved.

Coulomb friction instead feeds the unknown normal solution back into the tangential constitutive law.

Duvaut and Lions placed these contact laws within the framework of variational inequalities in the foundational theory of unilateral mechanics:

• G. Duvaut & J.-L. Lions, Inequalities in Mechanics and Physics.

The Coulomb coupling turns the problem into a nonlinear/quasi-variational system rather than a standard convex minimization problem.

What classical theory already established

CO01 is not the claim that three-dimensional Coulomb contact lacks existence results.

Eck & Jarušek proved existence for the static continuum contact problem using penalization, smoothing, and additional regularity, provided the friction coefficient is sufficiently small.

Andersson subsequently proved existence for a quasistatic time-dependent Signorini problem with local Coulomb friction, within a framework that inherits a small-friction restriction and additional regularity assumptions.

Thus

$$\boxed{ \text{restricted 3D existence theory is established.} }$$

The long-standing question is how far those restrictions are genuinely necessary.

Existence and uniqueness are different questions

Even when a solution exists, it need not be unique.

Hild (2005) constructed continuum examples with multiple Coulomb-contact solutions, including stick and separation states.

Renard (2006) obtained a conditional uniqueness criterion under additional assumptions and sufficiently small friction.

Therefore the fundamental question is not:

$$\text{"Does every Coulomb problem have one unique solution?"}$$

It is to determine the correct existence and evolution regimes, including where multiplicity or nonexistence can occur.

Why computation does not close the theoretical problem

Finite-element and contact algorithms routinely solve large three-dimensional frictional-contact problems.

That computational success is important but answers a different question.

A spatial discretization produces a finite-dimensional nonsmooth or complementarity problem. Finite-dimensional compactness and fixed-point tools can exist even when the corresponding passage to the continuum limit is unresolved.

Hence

$$\boxed{ \text{a computed discrete contact state} \not\Rightarrow \text{a general continuum existence theorem}. }$$

The analytical difficulty appears when normal pressure and frictional response converge only weakly and their nonlinear product must still have a meaningful continuum limit.

2024: the two-dimensional incremental problem becomes sharp

A major advance came from

• P. Ballard & F. Iurlano, Existence results for the time-incremental elastic contact problem with Coulomb friction in 2D (2024).

They consider the general incremental Signorini–Coulomb problem in two-dimensional linear elasticity with arbitrary geometry and elasticity tensor.

The formulation uses the normal traction as a principal unknown and reduces the problem to a boundary variational inequality involving a nonlinear operator

$$A:t_n\mapsto u_n$$

that incorporates both elasticity and dry friction.

Isotropic elasticity

For general 2D isotropic elasticity, including heterogeneous isotropic materials, a solution exists for

$$\boxed{\text{arbitrarily large Coulomb friction coefficient}.}$$

This removes the classical small-friction restriction for the incremental 2D isotropic problem.

Anisotropic elasticity

For 2D anisotropic elasticity, the situation is genuinely different.

The authors identify a critical friction condition for existence and exhibit a case with no solution when that condition is violated.

Thus large-friction nonexistence is not merely a weakness of older proof techniques.

Why the sharp result stops at two dimensions

The 2024 analysis reduces the contact problem to a nonlinear boundary operator in the Leray–Lions/pseudomonotone framework.

Most required properties can be established independently of spatial dimension.

The delicate step is a compactness property needed to pass to the limit in a product of weakly convergent quantities. Ballard and Iurlano prove it using a fine property of the elastic Neumann-to-Dirichlet operator.

That boundary-operator property is established for the two-dimensional problem.

The corresponding general three-dimensional step is not currently available.

This is the precise mathematical boundary behind CO01:

$$\boxed{ \text{sharp 2D incremental existence theory} \quad\text{vs.}\quad \text{incomplete sharp 3D theory}. }$$

2025: continuous quasistatic evolution in 2D

The next step was studied in

• P. Ballard & F. Iurlano, Transition from Continuous to Jumping Solutions in 2D Quasi-static Elastic Contact Problems with Coulomb Friction (2025 preprint).

They formulate the general two-dimensional quasistatic evolution allowing jumps in load and solution.

For an optimal condition on friction, absolutely continuous loading admits an absolutely continuous quasistatic solution.

When the condition is violated, they give examples in which the solution undergoes spontaneous jumps in time even under continuously varying load.

This distinguishes two questions:

$$\boxed{ \text{Does an incremental equilibrium exist?} }$$

and

$$\boxed{ \text{Can those equilibria form a continuous quasistatic evolution?} }$$

Both questions remain less complete in three dimensions.

What exactly remains open

For a general three-dimensional elastic body with Signorini contact and local Coulomb friction, the sharp continuum theory remains incomplete.

The unresolved state of the problem includes:

• whether general 3D isotropic incremental contact admits solutions for arbitrary friction magnitude, as proved in 2D;
• the sharp existence condition for general 3D anisotropic elasticity;
• the corresponding regimes of multiplicity or nonexistence;
• the passage from incremental solutions to a time-continuous quasistatic evolution;
• and the conditions under which that evolution remains continuous or must develop jumps.

These statements define the current frontier without assuming how it will be resolved.

Why three dimensions change the problem

The difference is not simply the size of the numerical model.

In two-dimensional elasticity, contact has a single scalar tangential direction.

In three dimensions, the tangential traction and slip are vectors in the two-dimensional tangent plane:

$$\mathbf t_t\in T_x\Gamma_C.$$

The contact surface itself is two-dimensional.

The fine boundary-operator property used in the current sharp existence proof is presently available only in the bidimensional setting.

Thus the dimensional gap is tied to the analytical structure of elastic boundary response and weak compactness.

Known / open boundary

What is established

• Variational-inequality formulations of unilateral elastic contact are classical.
• Coulomb friction is harder than Tresca friction because its threshold depends on the unknown normal pressure.
• Static and quasistatic 3D existence is known under small-friction and regularity assumptions.
• Solutions can be nonunique.
• The general 2D incremental problem now has sharp existence results.
• In 2D isotropic elasticity, incremental solutions exist for arbitrarily large friction.
• In 2D anisotropic elasticity, large friction can genuinely destroy existence.
• Recent 2D theory also distinguishes continuous quasistatic response from spontaneous jumps.

What remains unresolved

The corresponding sharp theory for general three-dimensional continuum contact remains incomplete.

Scientific scope

This page does not claim that Coulomb frictional contact cannot be computed in 3D. It routinely is.

It does not claim that no 3D existence theorem exists. Restricted continuum results have been known for decades.

It does not assume uniqueness.

It also does not treat the full elastodynamic contact problem as the main open question. CO01 concerns the incremental and quasistatic elastic Signorini–Coulomb problem.

Finally, “arbitrarily large friction” is already known to behave differently for isotropic and anisotropic elasticity in two dimensions; the website should not suggest that one universal unrestricted theorem is necessarily true.

Key papers and milestones

• G. Duvaut & J.-L. Lions, Inequalities in Mechanics and Physics (1976) — foundational variational-inequality treatment of unilateral mechanics and friction.
• C. Eck & J. Jarušek, Existence results for the static contact problem with Coulomb friction (1998) — continuum existence under a small-friction condition.
• P. Ballard, A counter-example to uniqueness in quasi-static elastic contact problems with small friction (1999) — limitations of simple small-friction uniqueness expectations.
• L.-E. Andersson, Existence Results for Quasistatic Contact Problems with Coulomb Friction (2000) — quasistatic continuum existence.
• P. Hild, Multiple solutions of stick and separation type in the Signorini model with Coulomb friction (2005) — explicit continuum multiplicity.
• Y. Renard, A Uniqueness Criterion for the Signorini Problem with Coulomb Friction (2006) — conditional uniqueness.
• P. Ballard & F. Iurlano, Homogenization of friction in a 2D linearly elastic contact problem (2022) — 2D friction homogenization and effective Coulomb law.
• P. Ballard & F. Iurlano, Existence results for the time-incremental elastic contact problem with Coulomb friction in 2D (2024) — sharp general 2D incremental existence theory.
• P. Ballard & F. Iurlano, Transition from Continuous to Jumping Solutions in 2D Quasi-static Elastic Contact Problems with Coulomb Friction (2025, preprint) — sharp 2D continuous-versus-jumping quasistatic evolution.

Progress log

30 September 2026 — Problem statement v1.0.
CO01 is formulated around the sharp 3D existence/evolution boundary rather than the broad claim that Coulomb contact is unsolved. Classical 3D small-friction existence results are acknowledged. The main recent benchmark is the 2024 sharp 2D incremental theory, followed by the 2025 2D continuous-versus-jumping quasistatic result. No comparable general 3D sharp theorem was identified in the literature audit through 30 September 2026.

Why it matters

The question is directly relevant to concrete, rocks, ceramics, composites, porous solids, additively manufactured materials, and other heterogeneous systems in which reliability depends on tails and sample-to-sample variability rather than only mean stiffness.

A deterministic effective fracture energy may be sufficient for one observable and insufficient for another. Peak strength, crack-initiation load, crack path, process-zone morphology, and failure probability can have different self-averaging properties.

The engineering issue is therefore not simply whether the constitutive energy homogenizes. It is which uncertainty survives into the observable used for design or reliability assessment.

What we know

Representative-volume and statistical-volume concepts are mature for random elastic media, but softening and localization complicate their extension to failure. Work on quasi-brittle materials has long shown that an ordinary elastic-style RVE may cease to be representative once damage localizes.

Static stochastic fracture energies can homogenize rigorously. Cagnetti, Dal Maso, Scardia and Zeppieri established stochastic homogenization of broad free-discontinuity functionals, with deterministic effective densities under ergodicity.

Effective fracture resistance can also be derived in specific disordered brittle systems. Lebihain and co-workers developed an effective-toughness framework by homogenizing crack-front interactions with toughness disorder.

These results establish that disorder can disappear from some macroscopic energetic descriptions. They do not imply that every fracture observable self-averages.

Finite-size failure can remain strongly random

Statistical-volume studies report that fracture strength can remain random, anisotropic, and size-dependent at scales where elastic response has already become effectively homogeneous.

This is not a contradiction of energy homogenization. It reflects that

in general.

Rare weak regions and localization can make failure depend on extreme statistics rather than on a simple spatial average.

Evolutionary homogenization is possible in special settings

The gap is not absolute.

Negri proved homogenization of Griffith evolution for periodic brittle laminates with a prescribed crack path. This is an important positive result: under sufficiently structured geometry and evolution constraints, an effective deterministic crack evolution can emerge.

The open problem is therefore not to show that fracture evolution can never homogenize. It is to determine when it does and when it does not.

Recent rigorous progress

2025 — Stochastic surface-energy homogenization broadens.
Donnarumma and Friedrich proved almost-sure stochastic homogenization for free-discontinuity surface functionals on asymptotically piecewise-rigid deformations; under ergodicity the effective density is deterministic.

2024 — Griffith evolution homogenizes in a periodic laminate.
Negri obtained an effective Griffith evolution for a periodic brittle laminate with a prescribed crack path.

Together these results sharpen the frontier. Static and even some evolutionary fracture quantities can homogenize. The unresolved issue is the fate of strength, nucleation, unrestricted paths, full random evolution, and fluctuations around the effective limit.

The frontier: observable-by-observable self-averaging

Let a stationary random brittle medium be indexed by $\omega$, with microstructural scale $\varepsilon$. For the fracture energy,

may hold under ergodicity.

Now let $Q_\varepsilon^\omega$ denote a fracture observable, for example:

• crack-initiation load;
• peak strength;
• effective toughness;
• crack geometry or path;
• size of the fracture process zone;
• dissipated energy at instability;
• or a quasistatic crack evolution.

The problem is to determine whether

with deterministic $Q_{\mathrm{eff}}$, or whether a nontrivial probability law survives.

If a deterministic limit exists, one should also ask for the fluctuation scale,

and identify the scaling $a_\varepsilon$ and limiting random variable or field $X$.

What remains open

A general theory should determine, for a declared random ensemble and fracture model:

• which observables self-average;
• the rate at which their fluctuations decay;
• whether rare-event or extreme-value statistics control initiation or peak strength;
• when crack paths converge geometrically and when macroscopic path randomness survives;
• when quasistatic random fracture evolution converges to an effective deterministic evolution;
• whether the relevant stochastic continuum state is scalar, field-valued, or history-dependent;
• and how the answers depend on disorder correlation length, tail statistics, spatial dimension, process-zone scale, and specimen size.

The theoretical target is not one universal law of randomness. Different fracture observables may belong to different stochastic scaling classes.

What would count as progress?

• A theorem proving self-averaging of a declared fracture observable under a nontrivial random ensemble.
• A theorem or counterexample showing that a specific observable retains nonzero macroscopic randomness.
• Quantitative fluctuation bounds or a central-limit / extreme-value scaling law for strength, initiation, or effective toughness.
• Homogenization of unrestricted crack evolution beyond prescribed or periodic paths.
• A stochastic continuum fracture model derived as the correct macroscopic limit when deterministic closure fails.
• Experiments or simulations designed to discriminate deterministic homogenization from persistent randomness over controlled scale ratios.

Scientific scope

This page does not claim that stochastic fracture energy fails to homogenize. Rigorous deterministic effective energies are already known under important assumptions.

It also does not claim that finite-size variability disproves homogenization. A deterministic infinite-scale limit can coexist with very slow convergence or large pre-asymptotic fluctuations.

The initial scope is quasistatic brittle or quasi-brittle fracture with a well-defined random ensemble and clear scale separation. Dynamic fragmentation, fatigue, corrosion-assisted fracture, and uncertainty from uncontrolled processing histories require separate formulations unless explicitly incorporated into the random model.

Key papers and milestones

• Ostoja-Starzewski, Material spatial randomness: From statistical to representative volume element (2006) — RVE/statistical-volume baseline.
• Gitman, Askes & Sluys, Representative volume: Existence and size determination (2007) — representative-volume issues for failure.
• Nguyen et al., On the existence of representative volumes for softening quasi-brittle materials (2010) — localization and failure-zone averaging.
• Cagnetti, Dal Maso, Scardia & Zeppieri, Stochastic Homogenisation of Free-Discontinuity Problems (2019) — rigorous stochastic fracture-energy homogenization.
• Lebihain, Ponson, Kondo & Leblond, Effective toughness of disordered brittle solids: A homogenization framework (2021) — disorder and effective toughness.
• Cagnetti et al., A global method for deterministic and stochastic homogenisation in BV (2022) — general framework.
• Abedi et al. (2022) — finite-size fracture statistics and anisotropy.
• Negri, Homogenization of Griffith’s Criterion for Brittle Laminates (2024) — effective crack evolution in a periodic setting.
• Donnarumma & Friedrich, stochastic homogenisation for asymptotically piecewise-rigid fracture functionals (2025) — recent rigorous progress.

Progress log

30 September 2026 — Problem statement v1.0.
S01 is formulated around the distinction between homogenization of fracture energy and self-averaging of physically observable failure quantities. The residual target is an observable-by-observable theory of deterministic versus stochastic macroscopic fracture, including fluctuation scaling when convergence occurs.

Theoretical question

Consider the terminal mushy zone after a load-bearing solid skeleton has formed. Let the body consist of evolving solid and liquid regions, with a crack or connected separation set Γ(t), and let z(t) collect the mechanically relevant state. The question is whether this evolving medium admits a Griffith-type fracture closure:

Here B(t) denotes the evolving solid-bridge/load-bearing network and H denotes history. The open target is to determine when a scalar crack-driving force and fracture resistance can govern crack advance and arrest, or to show that no such Griffith-type reduction is possible and identify the additional state or history variables that are irreducible.

Why it matters

Solidification cracking limits casting, welding, and fusion-based additive manufacturing of high-performance alloys. It can render otherwise attractive alloy compositions unmanufacturable or leave crack networks that compromise structural integrity.

The mechanics is unusual: the crack advances while the material carrying load is itself being created. Solid bridges form and strengthen, liquid channels feed or disconnect, thermal and solidification shrinkage generate loading, and the topology of the skeleton evolves on the same time scale as fracture.

The question is now experimentally testable in ways that were previously impossible. In-situ synchrotron imaging can resolve crack initiation and propagation during welding and laser powder-bed fusion at micrometre and sub-millisecond scales.

Model and scope

Primary setting: terminal solidification of metallic alloys after mechanical coherency of the solid skeleton, with casting and welding as mature physical foundations and additive manufacturing as a high-rate benchmark.

The first target is crack advance, arrest, and the transition from local cavity/bridge failure to a connected crack. Nucleation of the first void is included only insofar as it supplies the initial crack-like defect or affects the subsequent driving state.

Liquid feeding, permeability, interfacial energy, bridge topology, temperature, solid fraction, and strain rate may enter the state; the page does not assume in advance that they can be collapsed into one conventional K_IC or G_c.

Liquation cracking, ductility-dip cracking, cold cracking, fatigue of the finished component, and fracture entirely below the solidus are outside the core statement.

Known and open boundary

Hot tearing has been studied for many decades through strain-, stress-, feeding-, and liquid-film ideas. A 2007 review explicitly argued that a modern criterion should include both crack nucleation and propagation and should be based on fracture mechanics. The present problem therefore has a long engineering history rather than arising only from additive manufacturing.[R1]

The RDG model provides a classical mechanics description of deformation, solidification shrinkage, interdendritic feeding, permeability, liquid-pressure drop, and cavitation. Kou’s later criterion gives a compact solidification-path measure of susceptibility. These theories are important for when separation becomes likely, but neither is a general crack-driving-force theory for an evolving semisolid crack.[R2][R4]

Fracture mechanics has already entered the field. The 2009 SKK model lets cavities form when feeding is insufficient and uses a Griffith-type critical cavity dimension to distinguish microporosity from a hot tear. The open problem is therefore not simply to ‘apply Griffith’ to hot tearing.[R3]

Direct experiments have separated stages of damage evolution. Synchrotron radiography during steel welding resolved isolated microcavities, coalescence into microcracks, and subsequent crack propagation, including measured propagation speeds. This establishes that initiation and propagation are experimentally distinct observables.[R5]

Continuum fracture formalisms can reproduce hot-crack patterns in process simulations. A thermodynamically consistent thermo-mechanical phase-field fracture model has been applied to additive-manufacturing hot cracking, including thermal strain and solidification shrinkage. Such models demonstrate computational feasibility but take fracture properties and degradation laws as constitutive inputs rather than deriving a universal mushy-zone fracture resistance from the evolving solid-liquid network.[R6]

Two 2024 developments moved directly into propagation mechanics. Lin et al. proposed an energy-balance propagation model and in-situ test for steel. Liu, Li and Lu formulated solidification cracking as competition between thermal-stress accumulation and solid-bridge strength and found that their crack-like solidifying structure has no classical LEFM stress singularity. Together these results show both that propagation can be modeled energetically and that the ordinary K-field cannot be assumed a priori.[R7][R8]

Work in 2025 resolved additional pieces of the evolving state. A strain-rate/solidification-rate criterion was validated for AZ91D casting; an Al-Li extension of SKK incorporated semisolid high-temperature toughness; and an interfacial-energy/microstructure model predicted the bridging liquid fraction. These advances make bridge formation and semisolid resistance more quantitative, but they produce model- or alloy-specific descriptors rather than a general crack-driving-force/resistance pair.[R9][R10][R11]

In 2026, fracture-style numerical prediction advanced further. A thermo-mechanical peridynamic model with a temperature-dependent critical-stretch law reproduced initiation and propagation of multiple hot cracks in a semisolid AA7050 quenching experiment, while a new analytical L-DED cracking criterion for Ni superalloys improved solidification-susceptibility prediction. These are important predictive closures in declared settings, not evidence that a unique Griffith-type law exists for the underlying evolving two-phase medium.[R12][R14]

Also in September 2026, high-speed synchrotron imaging of LPBF Al6061 quantified crack evolution at sub-millisecond resolution: cracking began shortly before melt-pool disappearance and developed over roughly a millisecond. The propagation dynamics needed to test a fracture law are therefore increasingly observable rather than hidden inside a post-mortem crack count.[R13]

Through the literature audit of 30 September 2026, no general result was identified that derives a process-independent crack-driving force and resistance for a solid-liquid load-bearing network whose bridges, liquid connectivity, and topology evolve during crack growth, together with a controlled reduction to ordinary solid fracture as f_s approaches one. That residual closure—not the absence of hot-cracking criteria—is the conservative open core.[R3][R7][R8][R9][R10][R11][R12][R13][R14]

What would settle it

For a clearly specified semisolid continuum or homogenization limit, derive a thermodynamically and mechanically consistent crack-evolution law and determine whether a scalar Griffith/J-type driving force and state-dependent resistance exist. The result should state the admissible state variables and history dependence, characterize the relevant crack-tip or process-zone mechanics, and recover the appropriate fully solid fracture law in a controlled limit.

Mechanically, the proposed closure should predict crack advance versus arrest and at least one propagation observable from independently characterized material/state variables across more than one loading or manufacturing configuration without refitting a new fracture parameter for every test.

A negative resolution is equally meaningful: prove that no state-local scalar resistance can close the problem for the stated class and identify the minimal additional internal variables, memory, or nonlocal information required.

Recent progress

2026 — Crack dynamics become directly measurable and fracture-style simulations mature

Sub-millisecond LPBF synchrotron imaging now quantifies crack growth during solidification, while semisolid peridynamic simulations reproduce initiation and propagation with a declared critical-stretch law. These advances make the residual fracture closure experimentally and computationally testable.[R12][R13][R14]

2025 — Bridge state and semisolid resistance become more quantitative

Rate-versus-solidification criteria, semisolid toughness measurements, and an interfacial-energy model for the bridging liquid fraction turn several previously empirical ingredients into measurable or computable state variables.[R9][R10][R11]

2024 — Propagation and solid-bridge fracture become explicit mechanics targets

Energy-balance propagation theory and solid-bridge fracture modeling move beyond susceptibility and initiation; the latter also shows that a classical crack-tip singularity need not exist in a solidifying crack.[R7][R8]

2007–2009 — Fracture-mechanics formulation enters the hot-tearing literature

The field explicitly called for a nucleation-and-propagation fracture framework, and the SKK model subsequently introduced a Griffith-type critical cavity size. This establishes the historical baseline that the modern open problem must go beyond.[R1][R3]

References

• historical fracture-mechanics framing[R1] Eskin & Katgerman, A Quest for a New Hot Tearing Criterion (2007)
• feeding/cavitation foundation[R2] Rappaz, Drezet & Gremaud, A new hot-tearing criterion (1999)
• Griffith-type cavity criterion[R3] Suyitno, Kool & Katgerman, Integrated Approach for Prediction of Hot Tearing (2009)
• solidification-path susceptibility[R4] Kou, A criterion for cracking during solidification (2015)
• direct initiation and propagation kinetics[R5] Aucott et al., Initiation and growth kinetics of solidification cracking during welding of steel (2017)
• thermodynamic phase-field fracture[R6] Ruan et al., A thermo-mechanical phase-field fracture model: Application to hot cracking simulations in additive manufacturing (2023)
• propagation energy model[R7] Lin et al., A hot tearing propagation model for steel (2024)
• solid-bridge fracture mechanics[R8] Liu, Li & Lu, Modeling solidification cracking: A new perspective on solid bridge fracture (2024)
• rate/solidification criterion[R9] Li et al., A method for predicting hot tearing during alloy solidification and its application in AZ91D magnesium alloy (2025)
• semisolid toughness extension[R10] Guo et al., An extended criterion for hot cracking susceptibility of Al-Li alloys (2025)
• bridge-state prediction[R11] Sheikhi et al., On the development of an interfacial energy and microstructure dependent model to predict bridging phenomenon and solidification cracking behavior (2025)
• semisolid fracture simulation[R12] Remani & Nallathambi, Ordinary state based peridynamic thermomechanical simulation of hot crack occurrence during semi-solid plate quenching experiment (2026)
• latest direct AM crack dynamics[R13] Ali et al., Study of crack formation dynamics in aluminum alloy by using laser powder bed fusion and in-situ synchrotron X-ray characterization (2026)
• latest AM susceptibility criterion[R14] Guan et al., A new cracking criterion for additive manufacturing and repair of Ni-based superalloys via laser directed energy deposition (2026)

References document the known boundary or reported progress. Their inclusion does not by itself certify that the broader target is globally open.

Theoretical question

Let a material contain structural levels k=0,…,N with characteristic scale ell_k, architecture A_k, constituent/interface mechanics C_k, and local failure physics Phi_k. For an evolving crack with active mechanism set M(a), the central mechanics question is

Here ell_p is a fracture/process-zone scale and eta_k indicates when fracture interacts directly with structural level k. The open target is not generic multiscale simulation or simple coarse-graining: it is to identify the mechanics that selects failure modes across interacting levels and thereby determines crack path and the macroscopic R-curve, while establishing when some levels may legitimately be replaced by an effective fracture law.

Why it matters

Many biological materials achieve combinations of strength and damage tolerance using only a small set of relatively ordinary constituents arranged hierarchically. Bone, nacre, enamel, dentin, tendon, wood and arthropod cuticle provide different physical realizations of this strategy; engineered hierarchical composites increasingly reproduce selected features.

Fracture is where structural hierarchy becomes mechanically nontrivial. A crack can be deflected at one level, activate interface sliding or distributed damage at another, and leave load-bearing bridges at a larger level. The relevant fracture/process-zone scale can therefore interact with several structural scales during one event.

A predictive theory would connect architecture and local failure physics to quantities used at engineering scale—crack path, initiation versus stable growth, rising R-curves and final instability—without fitting a new macroscopic fracture law separately for each specimen or hierarchy.

Model and scope

Primary setting: passive quasistatic fracture of hierarchical solids and composites with two or more mechanically distinct structural scales. Biological materials provide the richest physical examples, but the mechanics target also includes engineered bio-inspired and other hierarchical composites.

System-specific physics is retained through the local laws Phi_k: interface sliding, viscoelasticity, sacrificial bonding, anisotropy, gradients, hydration dependence, or other mechanisms may be essential in a given material. The problem is to understand how those local mechanisms interact with architecture across levels, not to erase them.

The initial target is structural hierarchy and crack/damage evolution. Active healing, remodeling, growth, cellular mechanobiology and whole-organ adaptation are outside the core statement; dynamic fracture and fatigue are later extensions unless explicitly included.

Known and open boundary

Hierarchical biological materials have long been understood as mechanically organized across many length scales rather than as homogeneous solids. Classical reviews and fracture-mechanics analyses established characteristic-size, interface and hierarchy effects as central to their strength and toughness.[R1][R2]

Mechanics already explains important routes by which hierarchy can transmit flaw tolerance. Gao’s hierarchical fracture treatment and Yao & Gao’s multiscale cohesive laws provide explicit bottom-up relations for idealized biological architectures. The open problem is therefore not simply whether hierarchy can toughen a material.[R2][R3]

Hierarchy is not universally beneficial or monotonic. Zhang, Zhang and Gao predicted an optimal number of hierarchy levels from competing stiffness, strength, flaw-tolerance and stress-transfer requirements, while Sen and Buehler showed that hierarchy alone can generate stable crack growth and strongly rising R-curves in an idealized brittle system.[R4][R5]

General multiscale fracture mechanics can already derive effective strength, fracture energy and traction-separation relations from constituent properties and microstructure in specific composite settings. Thus a two-scale effective fracture law is not itself the missing theory.[R6]

A deeper closure issue appears even in energetically consistent scale transitions. Turteltaub and Suárez-Millán constructed two distinct Hill-Mandel-consistent multiscale fracture transitions that dissipate the same energy but generally produce different effective traction-separation laws. Energy consistency alone therefore does not guarantee a unique effective fracture description.[R7]

Recent biological-materials literature shows that different systems realize toughening through different combinations of characteristic size, interfaces, structural hierarchy, gradients and dissipation mechanisms. A 2025 review synthesizes these common mechanics ingredients across bone, nacre, horns, silks and other systems rather than identifying one universal biological fracture mechanism.[R8]

System-specific theory is becoming increasingly quantitative. A 2025 hierarchical crack-bridging model couples nacre microstructure to macroscopic T-stress and shell morphology, showing how distinct hierarchy levels can act synergistically on strength and toughness.[R9]

Recent work also shows why a single intrinsic toughness is often insufficient. In 2026, fracture energy in hierarchical biological composites was shown to depend on hydration and structural scale because those variables change the active crack-tip dissipation mechanisms and process-zone development.[R10]

Other 2026 studies demonstrate additional forms of mechanism selection: microstructural randomness in nacreous materials can produce a transition to distributed fractal cracking, while a dual-damage phase-field model reproduces different tensile and compressive failure modes through the interaction of brittle tablets, compliant interfaces and architecture.[R11][R12]

Through the literature audit of 30 September 2026, no general mechanics theory was identified that takes a declared structural hierarchy and independently characterized local failure physics and predicts, across multiple interacting levels, the sequence of active failure mechanisms, crack trajectory and macroscopic R-curve while also characterizing when lower levels admit a unique effective reduction. That cross-scale mechanism-selection and fracture-resistance problem is the conservative open core.[R3][R5][R6][R7][R8][R9][R10][R11][R12]

What would settle it

For a declared hierarchy and loading class, predict the active sequence or competition of fracture mechanisms, crack path and macroscopic growth resistance from independently characterized architecture, constituent/interface mechanics and lower-scale failure laws, without fitting an independent macroscale fracture law to the target response.

A theoretical resolution should identify the structural and fracture length-scale information required for that prediction and state the regimes in which selected hierarchy levels can be replaced by a closed effective fracture law. If such a reduction is claimed, its uniqueness or admissible nonuniqueness should be characterized.

A negative result would also be meaningful if it proves that a proposed reduced state—such as a scalar toughness or one traction-separation law—is insufficient, and identifies the additional history, internal variables or nonlocal information required.

Recent progress

2026 — Scale, environment and disorder sharpen the fracture-state problem

Hydration-dependent fracture energy, random-architecture fractal cracking and loading-dependent nacre failure modes show that the active dissipation/process zone and mechanism set can change with structural scale, environment and microstructural statistics.[R10][R11][R12]

2025 — Cross-level synergy becomes quantitatively modelled

A broad review consolidates hierarchy, interfaces and characteristic size as recurring ingredients across biological materials, while a hierarchical nacre model explicitly couples microstructural crack bridging to macroscale stress and morphology.[R8][R9]

2018–2020 — Effective fracture scale transitions established—and shown nonunique

Computational homogenization derives effective traction-separation relations from microscale fracture, but later work shows that distinct energy-consistent transition rules can generate different effective laws from the same microscale response.[R6][R7]

2006–2011 — Hierarchical fracture mechanics establishes the foundation

Flaw-tolerance theory, recursive cohesive laws, optimal-hierarchy analysis and hierarchy-generated R-curves establish that architecture can transform local failure across scales; the modern problem is to predict that transformation for general interacting hierarchies.[R2][R3][R4][R5]

References

• hierarchical-materials foundation[R1] Fratzl & Weinkamer, Nature’s hierarchical materials (2007)
• hierarchical fracture foundation[R2] Gao, Application of Fracture Mechanics Concepts to Hierarchical Biomechanics of Bone and Bone-like Materials (2006)
• recursive cohesive mechanics[R3] Yao & Gao, Multi-scale cohesive laws in hierarchical materials (2007)
• optimal hierarchy[R4] Zhang, Zhang & Gao, On optimal hierarchy of load-bearing biological materials (2011)
• hierarchy-generated R-curve[R5] Sen & Buehler, Structural hierarchies define toughness and defect-tolerance despite simple and mechanically inferior brittle building blocks (2011)
• effective fracture homogenization[R6] Turteltaub et al., Multiscale analysis of mixed-mode fracture and effective traction-separation relations for composite materials (2018)
• scale-transition nonuniqueness[R7] Turteltaub & Suárez-Millán, Energetically-consistent multiscale analysis of fracture in composites materials (2020)
• cross-system biological mechanics review[R8] Toughening and strengthening mechanisms of biological materials: A review (2025)
• hierarchical crack bridging[R9] Yan & Feng, Synergistic toughening mechanisms of macro- and micro-structures in nacre: Effects of T-stresses (2025)
• scale-dependent fracture resistance[R10] Kellersztein, Hydration- and scale-dependent fracture energy in hierarchical biological composites (2026)
• disorder and emergent fracture mode[R11] Yan et al., Fractal cracking: An emergent failure mode of nacreous materials with microstructural randomness (2026)
• loading-dependent mechanism selection[R12] Vigliotti, Multiscale phase-field analysis of nacre’s asymmetric mechanical strength (2026)

References document the known boundary or reported progress. Their inclusion does not by itself certify that the broader target is globally open.