Saint-Venant's Compatibility Condition¶

The differential integrability test for whether a symmetric small-strain field can be the symmetrized gradient of one displacement field.

Version: v1 · 2026-10-03 · History
Domain-specific #: 13587
Domain group: Formal Sciences
Origin domain: Mathematics
Subdomains: Continuum Mechanics, Elasticity Complex → Mathematics
Aliases: Saint Venant Compatibility Equations, Strain Compatibility Equations

Core Idea¶

Saint-Venant's compatibility condition asks whether a symmetric infinitesimal-strain tensor field can come from one continuous, single-valued displacement field. If $\mathbf u$ is a small-displacement vector field, its symmetric gradient is $e_{ij}=\tfrac12(\partial_i u_j+\partial_j u_i)$. The Saint-Venant operator takes certain second derivatives of a proposed $e$; every field of the displayed form makes that operator vanish. Conversely, on a suitable simply connected Euclidean domain, vanishing of the operator is sufficient to recover a displacement field under the theorem's regularity assumptions.^[1]

This is a precise integrability statement, not the vague assertion that neighboring material elements “fit.” A local tensor prescription may look plausible point by point but be impossible to assemble into a single displacement over the whole body. The condition detects that obstruction before a material law or force equation is used. Even when a displacement exists, it is only determined up to an infinitesimal rigid translation and rotation, since these alter no symmetric strain.^[1]

In a smooth two-dimensional specialization, the condition is

\[ \frac{\partial^2 e_{xx}}{\partial y^2} +\frac{\partial^2 e_{yy}}{\partial x^2} -2\frac{\partial^2 e_{xy}}{\partial x\,\partial y}=0. \]

Three dimensions have a tensor system of related second-derivative equations rather than just this one scalar equation. The zero condition is necessary whenever $e=\operatorname{sym}\nabla\mathbf u$; calling it sufficient requires topology and regularity to be stated. A multiply connected body may need additional loop conditions even though the local operator is zero.^[1]

Structural Signature¶

Sig role-phrases: symmetric strain-like tensor field → displacement-potential question → second-derivative incompatibility operator → domain and regularity hypothesis → existence verdict with rigid-motion freedom → defect-source qualification.

Symmetric strain-like tensor field. The proposed data are $e_{ij}=e_{ji}$ over a region, with enough regularity for the chosen classical or weak derivative formulation. A scalar, skew tensor or finite-strain metric is a different integrability problem.^[1]
Displacement-potential question. The target is one vector field $\mathbf u$ whose symmetric gradient equals $e$. This target separates the condition from merely calculating derivatives of a tensor.^[1]
Second-derivative incompatibility operator. The Saint-Venant combination cancels for every symmetric gradient because mixed partial derivatives commute in the smooth case. A nonzero result excludes an ordinary displacement potential for that candidate field.^[1]
Domain and regularity hypothesis. A simply connected domain and the theorem's functional setting turn local cancellation into a global existence result. Simple connectivity is a clean sufficient topological assumption, not a statement that every compatible field on an annulus is impossible.^[1]
Existence verdict with rigid-motion freedom. Passing the test under those hypotheses establishes some $\mathbf u$, not a uniquely positioned body or a stress value. Adding rigid translation/infinitesimal rotation leaves $e$ unchanged.^[1]
Defect-source qualification. A dislocation-containing model can have elastic-strain incompatibility constrained by dislocation density. The physical material is not thereby “impossible”; the homogeneous zero-source condition has been replaced by a source-consistent one.^[2]

What It Is Not¶

It is not Hooke's law. A constitutive law relates stress to strain; compatibility asks whether a proposed small-strain field can arise from one displacement, independent of the stiffness tensor.^[1]^[3]
It is not static or dynamic force balance. Equilibrium equations constrain stresses and loads; a field can satisfy a stress relation yet fail kinematic compatibility.
It is not a finite-strain compatibility theorem. Large deformations require nonlinear measures and different integrability conditions; the symmetrized-gradient criterion is linearized.^[1]
It is not globally sufficient on every domain solely because the local operator vanishes. A hole can permit local potentials that fail to glue into a single-valued global displacement without extra period conditions.^[1]
It is not a claim that all physically meaningful dislocation strains must have zero incompatibility. The elastic part can obey a defect-sourced condition rather than the homogeneous equation.^[2]
It is not a measurement of stress. A nonzero operator diagnoses a kinematic obstruction or source requirement; material response and boundary data are needed to calculate a stress field.^[2]

Scope of Application¶

The classical home is small-strain continuum mechanics. When strain is computed from a known displacement, the equations give a consistency check; when strain is chosen as a primary unknown, they become a constraint that keeps the strain field within the image of the symmetrized-gradient operation. Ciarlet and Ciarlet Jr. explicitly use weak Saint-Venant conditions in a strain-primary formulation of three-dimensional linearized elasticity.^[1]^[3]

The same formal test also matters in tensor analysis independent of any particular elastic material. It is an operator on a symmetric field over a Euclidean domain, with a theorem about potential existence. In dislocation-containing solids, the zero-source formulation is a reference point for a modified incompatibility equation. Gröger, Lookman and Saxena analyze that distinction in three dimensions and illustrate it with an edge-dislocation stress model.^[2]

The smooth planar expression is a useful teaching and diagnostic specialization, but the full three-dimensional theorem has more components and can be formulated weakly for lower-regularity fields. Its exact sufficiency statement travels with its domain assumptions; it does not transfer automatically to arbitrary curved manifolds or large-strain kinematics.^[1]

Clarity¶

Three questions are often conflated. Symmetry asks whether $e_{ij}=e_{ji}$. Compatibility asks whether such an $e$ is a symmetric gradient. Material response asks which stresses a compatible deformation would produce. Symmetry alone is not compatibility, and compatibility alone says nothing about stiffness, loading or boundary traction.^[1]^[3]

The theorem also distinguishes local from global information. Computing zero second-derivative incompatibility is a local check. A global displacement requires the right topology or supplementary loop data. Thus “$W(e)=0$” and “one single-valued displacement exists on this entire region” are equivalent only after the hidden global hypotheses have been made visible.^[1]

Manages Complexity¶

Instead of guessing three displacement components and solving many derivative equations directly, the condition compresses the existence problem into a test on the proposed symmetric tensor field. In two dimensions one scalar expression detects the local obstruction; in three dimensions a structured tensor system does. A passing test with appropriate topology licenses reconstruction, but still leaves rigid translation and rotation unfixed.^[1]

The compression is especially valuable when strain is the natural data or numerical unknown. Without it, an arbitrary collection of pointwise strain components might be mistaken for a realizable deformation. With it, kinematic admissibility is checked separately from constitutive and equilibrium equations. In defect models, a source term keeps this discipline rather than abandoning compatibility when the homogeneous test fails.^[3]^[2]

Abstract Reasoning¶

Start with a symmetric small-strain field $e$ and specify the region. Apply the appropriate classical or weak Saint-Venant operator. If the result is nonzero, $e$ cannot by itself be the symmetric gradient of one smooth displacement under the homogeneous kinematic model. If it vanishes, ask whether the region and field meet a global existence theorem—simply connectedness is a standard sufficient assumption. Only then infer a displacement field, unique up to infinitesimal rigid motion.^[1]

For the planar equation, take $e_{xx}=y$, $e_{xy}=x/2$, and $e_{yy}=0$. Its compatibility expression is zero, and $\mathbf u=(xy,0)$ is an explicit potential. By contrast, $e_{xx}=y^2$ with $e_{xy}=e_{yy}=0$ gives the value $2$ and cannot be that symmetric gradient on any open planar patch. These are exact calculus checks, not experimentally measured strains. In a dislocation model, do not classify every nonzero elastic incompatibility as model failure; ask whether a defect source supplies the modified balance.^[1]^[2]

Knowledge Transfer¶

The condition transfers literally from analytical small-strain elasticity to strain-based numerical formulations: in both, a symmetric tensor must lie in the image of $\operatorname{sym}\nabla$ and the same local operator tests it. A defect model uses the same incompatibility language but changes the right-hand side or field decomposition to account for dislocations. The transfer therefore preserves the operator's diagnostic role while refusing to carry a zero-source equation into a source-bearing case unchanged.^[3]^[2]

Outside that tensor/displacement setting, “compatibility” may only be an analogy. The live Prime Compatibility captures broad coexistence or composability, but it does not supply these second-derivative equations or the topology theorem. A portable “local integrability versus global potential” skeleton is a future-prime question, not evidence that this named condition itself is prime.

Examples¶

Displacement-generated plane strain¶

On a simply connected planar patch, let $\mathbf u(x,y)=(xy,0)$. Its symmetric gradient has $e_{xx}=y$, $e_{xy}=x/2$ and $e_{yy}=0$. The planar Saint-Venant expression vanishes: $\partial_{yy}y+\partial_{xx}0-2\partial_{xy}(x/2)=0$. A single-valued displacement is already displayed, so this example shows both the necessity of the differential condition and the kind of potential its sufficiency theorem promises. Adding a small rigid translation or rotation would change $\mathbf u$ without changing $e$.^[1]

Mapped back: symmetric strain-like tensor field = the three displayed components; displacement-potential question = whether one $\mathbf u$ generates them; second-derivative incompatibility operator = zero planar expression; domain and regularity hypothesis = smooth field on simply connected patch; existence verdict with rigid-motion freedom = explicit $\mathbf u$ exists but is not unique; defect-source qualification = none is needed in this ordinary compatible case.

Elastic incompatibility around a dislocation¶

Gröger, Lookman and Saxena formulate three-dimensional strain-based mechanics in a medium with a dislocation network. Their original study distinguishes a homogeneous Saint-Venant constraint for a crack-free body without such sources from an incompatibility constraint consistent with dislocation density, and illustrates the latter with an edge-dislocation stress calculation. The point is not that the entire body lacks any kinematics: it is that the elastic strain part cannot simply be classified by an unmodified zero-source compatibility rule.^[2]

Mapped back: symmetric strain-like tensor field = the modeled elastic strain; displacement-potential question = whether that elastic part alone is an ordinary global symmetric gradient; second-derivative incompatibility operator = the three-dimensional incompatibility operator; domain and regularity hypothesis = continuum model with a defect source, not an unqualified source-free domain; existence verdict with rigid-motion freedom = homogeneous reconstruction is not licensed from the elastic part alone; defect-source qualification = a nonzero source must match the modeled dislocation distribution.

Boundary: an impossible prescribed planar field¶

Set $e_{xx}=y^2$ and $e_{xy}=e_{yy}=0$ on an open planar patch. The compatibility expression is $2$, not zero. No smooth $\mathbf u$ can have precisely this symmetric gradient there. This is a negative diagnostic of missing operator cancellation, not a third positive case or a claim about a real material.^[1]

Structural Tensions¶

T1 — Local test versus global reconstruction. The differential operator efficiently detects local failure, but vanishing may not settle whether potentials glue around holes. Requiring simple connectivity gives a clean theorem but excludes compatible fields that happen to exist on multiply connected regions; ignoring topology risks false certification. Diagnostic: Is the domain simply connected, or have the needed loop conditions been checked?^[1]

T2 — Strain-primary modeling versus automatic kinematic admissibility. A strain-primary formulation can directly express measured or optimized strain fields, but must enforce compatibility and later recover displacement only up to rigid motion. Displacement-primary modeling builds compatibility in but can obscure the strain-focused structure of a problem. Diagnostic: Which variable is primary, and where is the integrability obligation discharged?^[3]

T3 — Homogeneous compatibility versus sourced defects. Requiring $W(e)=0$ everywhere would wrongly exclude elastic incompatibility associated with dislocations. Allowing arbitrary $W(e)\ne0$ would lose the constraint entirely. A defect-sourced relation keeps the tensor field accountable to a specified dislocation distribution. Diagnostic: Is a nonzero incompatibility supported by a modeled defect source?^[2]

Structural–Framed Character¶

This condition is strongly structural within its formal mechanics domain. Evaluative weight: the zero/nonzero verdict is mathematical, though engineers may prefer one case for a design. Human-practice dependence: low for the theorem; choosing a material model or data field is a practice-dependent application. Institutional origin: its historical name does not constitute the operator or proof. Vocabulary travel: “compatibility” travels broadly, but the Saint-Venant equations travel literally only with symmetric small-strain tensors and displacement gradients. Import versus recognition: applying the condition recognizes an integrability property already implied by the field and domain, rather than imposing a social or institutional criterion.^[1]

Its character: a domain-specific mathematical integrability condition with an explicit topology boundary and a defect-source extension. A more general local-to-global potential pattern may be a future-prime candidate, but that possibility does not change the named condition's typed carrier.

Structural Core vs. Domain Accent¶

The structural core is the chain potential field → symmetric derivative → vanishing second-derivative obstruction, with a qualified converse from vanishing obstruction to global potential. Its asserted DAG parent is Tensor Field by composition/presupposition: a symmetric rank-two field over a region is the required carrier. The condition is not a subtype of a tensor field.

The domain accent is the infinitesimal displacement/strain relation, the Saint-Venant derivative combination, the topology assumption and the rigid-motion gauge. Remove these and “compatible” becomes an uninformative broad label. Prime Compatibility is therefore related but not a strict parent; a portable local-to-global integrability skeleton remains an explicitly marked future-prime question rather than a fabricated live node.

This entry presupposes Tensor field.

Asserted prerequisite parent — Tensor Field. The condition requires a symmetric tensor field whose spatial derivatives can be tested; this is a composition edge, not subsumption.
Related, not asserted parent — Compatibility. The prime addresses broad relational coexistence; it does not entail symmetric-gradient integrability.
Related use — Linear Elasticity. That material-response model often needs the condition, but the mathematical test does not require a Hookean constitutive law.
Related contrast — Finite Strain Theory. Nonlinear deformation geometry has different compatibility questions.
Related sourced case — Dislocation. Its defect density can appear in a modified incompatibility equation; this does not make every elastic strain field homogeneously compatible.

Relationships to Other Abstractions¶

Current abstraction Saint-Venant's Compatibility Condition Domain-specific

Parents (1) — more general patterns this builds on

Saint-Venant's Compatibility Condition presupposes Tensor field Domain-specific

The condition acts on a symmetric rank-two tensor field over a region.

Hierarchy path (1) — routes to 1 parentless root

Saint-Venant's Compatibility Condition → Tensor field → Local-to-Global Aggregation

Neighborhood in Abstraction Space¶

Saint-Venant's Compatibility Condition sits in a sparse region of the domain-specific corpus (83^rd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (2551 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-08

Not to Be Confused With¶

Saint-Venant's principle. Tell: that principle concerns decay or localization of load effects away from an application region, not this symmetric-gradient integrability test.
Stress equilibrium. Tell: forces and stress divergence enter equilibrium; second derivatives of proposed strain and existence of $\mathbf u$ enter compatibility.
Constitutive law. Tell: stress–strain response requires material coefficients; compatibility does not.^[3]
Finite-deformation compatibility. Tell: a nonlinear deformation gradient or finite strain cannot be screened solely with these linearized equations.
Generic Prime Compatibility. Tell: coexistence of two systems says nothing about the Saint-Venant operator or a displacement potential.
Defect-free versus dislocation-sourced conditions. Tell: the latter has a nonzero source tied to defect density, not arbitrary violation of the homogeneous equation.^[2]

References¶

[1] Philippe G. Ciarlet, Patrick Ciarlet Jr., Giuseppe Geymonat and Françoise Krasucki, “Characterization of the kernel of the operator CURL CURL,” Comptes Rendus Mathématique 344 (2007), 305–308, original theorem on symmetric tensor fields and linearized strains in simply connected domains. https://www.numdam.org/articles/10.1016/j.crma.2007.01.001/ . registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j ↩k ↩l ↩m ↩n ↩o ↩p ↩q ↩r ↩s ↩t ↩u ↩v

[2] R. Gröger, T. Lookman and A. Saxena, “Incompatibility of strains and its application to mesoscopic studies of plasticity,” Physical Review B 82, 144104 (2010), original abstract. https://journals.aps.org/prb/abstract/10.1103/PhysRevB.82.144104 . registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g ↩h ↩i ↩j

[3] Philippe G. Ciarlet and Patrick Ciarlet Jr., “Another approach to linearized elasticity and Korn's inequality,” Comptes Rendus Mathématique 339 (2004), 307–312, original article. https://www.numdam.org/articles/10.1016/j.crma.2004.06.021/ . registry ↩a ↩b ↩c ↩d ↩e ↩f ↩g