Fractional Korn Inequality in Nonlocal Analysis

• Fractional Korn inequality is defined as the equivalence, up to rigid motions, between standard fractional Sobolev seminorms and nonlocal projected-difference seminorms.
• It applies to various domains—including C1, Lipschitz, and fractal boundaries—thereby underpinning models in peridynamics and fractional PDE theory.
• Analytical techniques such as localization, Hardy-type inequalities, and Fourier analysis are crucial for proving these inequalities and understanding their limitations.

A fractional Korn inequality is an analogue of the classical Korn inequality in the framework of fractional Sobolev spaces and nonlocal vector calculus. It establishes the equivalence, up to rigid motions, between the standard fractional Sobolev seminorm and certain nonlocal, projected-difference (directional) seminorms for vector fields. These inequalities are central in nonlocal mechanics (notably peridynamics), fractional PDE theory, and the study of the regularity properties of nonlocal systems on domains with various geometric properties. The fractional Korn inequalities can be classified by boundary behavior constraints, domain geometry, and the order of differentiability and integrability exponents.

1. Fractional Sobolev Spaces and Projected-Difference Seminorms

Let $\Omega \subset \mathbb{R}^n$ be an open set, $n \geq 2$, $1 < p < \infty$, and $s \in (0,1)$. The classical fractional Sobolev space $W^{s,p}(\Omega; \mathbb{R}^n)$ is defined as the space of vector fields $u$ with finite norm

$$\|u\|_{W^{s,p}(\Omega)} = \|u\|_{L^p(\Omega)} + |u|_{W^{s,p}(\Omega)},$$

where

$$|u|_{W^{s,p}(\Omega)}^p = \int_\Omega \int_\Omega \frac{|u(x) - u(y)|^p}{|x - y|^{n + ps}} \, dy \, dx.$$

Motivated by nonlocal elasticity, the "fractional symmetric gradient" or projected-difference seminorm is defined for $u : \Omega \to \mathbb{R}^n$ as

$$[u]_{\mathcal X^{s,p}(\Omega)}^p = \int_\Omega \int_\Omega \frac{\left[(u(y) - u(x)) \cdot \frac{y-x}{|y-x|} \right]^p}{|y-x|^{n+ps}} \, dy \, dx.$$

The kernel of $[\cdot]_{\mathcal X^{s,p}}$ is the space of infinitesimal rigid motions $$\mathcal{R} = \{ r(x) = A x + b : A^\top + A = 0, b \in \mathbb{R}^n \}$$.

2. Main Forms of the Fractional Korn Inequality

The fractional Korn inequalities relate the above seminorms. There are two principal types:

• First Fractional Korn Inequality (Rigid-motion removal):

$$\inf_{r \in \mathcal{R}} |u-r|_{W^{s,p}(\Omega)}^p \leq C [u]_{\mathcal X^{s,p}(\Omega)}^p$$

for all $u \in W^{s,p}(\Omega; \mathbb{R}^n)$, with no boundary condition required on domains with $C^1$ or small Lipschitz constant boundary (Harutyunyan et al., 2023). This is the unconstrained case. In earlier literature, validity required $u|_{\partial\Omega}=0$ and $ps > 1$ (Harutyunyan et al., 2022).

• Second Fractional Korn Inequality:

$$|u|_{W^{s,p}(\Omega)}^p \leq C \left( [u]_{\mathcal X^{s,p}(\Omega)}^p + \|u\|_{L^p(\Omega)}^p \right)$$

for all $u \in W^{s,p}(\Omega; \mathbb{R}^n)$ (Harutyunyan et al., 2023).

For fields with compact support or zero trace ($u \in W^{s,p}_0(\Omega)$), the two seminorms are equivalent (Mengesha et al., 2020): $$|u|_{W^{s,p}(\Omega)} \approx [u]_{\mathcal X^{s,p}(\Omega)}$$ when $ps>1$ and $\Omega$ is $C^1$.

3. Domain Geometry and Boundary Conditions

The range of validity for the fractional Korn inequalities depends crucially on the regularity and geometric properties of $\Omega$, and on the values of $p,s$.

• On $C^1$ domains and Lipschitz domains with sufficiently small Lipschitz constant, both Korn inequalities hold for all $1 < p < \infty$, $0 < s < 1$, with no boundary condition imposed (Harutyunyan et al., 2023, Rutkowski, 2021).
• On planar convex polygonal domains, both inequalities are established regardless of the Lipschitz constant (Harutyunyan et al., 2023).
• On uniform domains (which strictly contain Lipschitz domains), the unconstrained (second) fractional Korn inequality holds, extending previous results (Acosta et al., 13 Jan 2026).
• John domains (which strictly contain uniform domains) admit fractional Korn-type inequalities for truncated and weighted seminorms (Acosta et al., 13 Jan 2026).
• In the presence of fractal boundaries (domains where the Assouad dimension of $\partial\Omega$ enters into estimates), weighted versions of the Korn inequalities remain valid provided the blowup exponent stays below the codimension threshold (Acosta et al., 13 Jan 2026).
• For bounded domains and $ps<1$, there exist explicit counterexamples showing that the unconstrained fractional Korn inequality fails (Harutyunyan et al., 2022). This pathology does not occur in the whole space or on epigraphs.

4. Analytical and Proof Techniques

The proofs of fractional Korn inequalities combine several advanced methodologies:

• Localization and Partition of Unity: Domains are covered by a finite number of patches, each diffeomorphic to epigraphs with small Lipschitz constant (Harutyunyan et al., 2023, Mengesha et al., 2020).
• Epigraph and Wedge Reflection Extensions: Nitsche-type extension operators are constructed on the epigraphs or wedges near corners, enabling precise control over boundary strips (Harutyunyan et al., 2023).
• Fractional Hardy-Type Inequalities: These control the behavior of fields near the boundary (Mengesha et al., 2020, Mengesha, 2018, Harutyunyan et al., 2022). For $\Omega$ admitting such an inequality,

$$\int_\Omega \frac{|u(x)|^p}{\mathrm{dist}(x, \partial\Omega)^{ps}} dx \leq C |u|_{\mathcal X^{s,p}(\Omega)}^p,$$

one can deduce the Korn inequality without chart-based arguments (Rutkowski, 2021).

• Discrete Poincaré and Whitney Tree Decomposition: On uniform or John domains, the global estimate is obtained by piecing together cube-local estimates using a tree structure and discrete Hardy/Poincaré inequalities to handle mean-value constraints (Acosta et al., 13 Jan 2026).
• Truncation and Weighted Seminorms: These allow extension to generic John domains and control boundary singularities, with the range of weights linked to the Assouad dimension (Acosta et al., 13 Jan 2026).
• Fourier Analysis: On the whole space, the equivalence between $W^{s,p}$ and the projected-difference space $\mathcal X^{s,p}$ is established via matrix-valued Fourier symbols. In particular, for $p=2$, this reduces to the positive-definiteness of the associated matrix (Mengesha, 2018, Scott et al., 2018).
• Rigidity Arguments: The kernel of $[\cdot]_{\mathcal X^{s,p}}$ is the space of infinitesimal rigid motions; Hardy's inequality shows that only the zero motion survives in bounded settings (Harutyunyan et al., 2022).

5. Sharpness, Limitations, and Counterexamples

The dichotomy $ps>1$ vs.\ $ps<1$ for bounded domains is sharp (Harutyunyan et al., 2022):

• For $ps>1$, Korn's first inequality holds for $u \in W^{s,p}_0(\Omega)$.
• For $ps<1$, explicit counterexamples exist: by constructing vector fields that interpolate between a rigid motion deep inside $\Omega$ and zero near the boundary, one demonstrates failure of the inequality (Harutyunyan et al., 2022).
• In the whole space or epigraph domains, Korn-type inequalities are valid for all $ps \neq 1$.
• The case $ps=1$ remains delicate and unresolved.

Table: Validity of Fractional Korn Inequality (First Type)

Domain Type	Boundary Condition	$ps>1$	$ps<1$	Reference
Bounded $C^1$	$u=0$ on $\partial$	Yes	No	(Harutyunyan et al., 2022)
Bounded $C^1$	None	Yes	Yes	(Harutyunyan et al., 2023) (small Lipschitz constant and all $0$1<p<\infty$)
Uniform / John domain	None	Yes	Yes	(Acosta et al., 13 Jan 2026)
Whole space	None	Yes	Yes	(Scott et al., 2018)

A plausible implication is that for applications demanding unconstrained (i.e., no trace vanishing) settings, fractional Korn inequalities should be used only under the regime of suitable domain regularity and, if necessary, adjusted seminorms or weighted measures.

6. Connections to Nonlocal Elasticity and PDE Theory

Fractional Korn inequalities are foundational in the analysis of nonlocal models, particularly peridynamics and fractional elasticity, due to their role in identifying energy spaces and controlling symmetrized gradients via nonlocal difference quotients (Scott et al., 2018, Mengesha et al., 2020).

• The equivalence $W^{s,p} = \mathcal X^{s,p}$ allows importing classical results: compactness, Sobolev embeddings, Poincaré and Caccioppoli inequalities, and regularity theory into nonlocal, fractional settings (Scott et al., 2018, Mengesha et al., 2020).
• Recent work establishes self-improving (“higher fractional differentiability”) properties for solutions to nonlinear, strongly coupled nonlocal systems (Mengesha et al., 2020). The fractional Korn inequality is an essential step in proving such regularity gains.

7. Generalizations and Open Problems

Key directions and unresolved issues:

• Domain Regularity: While small Lipschitz constant suffices, full generality for arbitrary bounded Lipschitz domains is conjectured (Harutyunyan et al., 2023).
• Critical Exponent: Korn-type inequalities at $ps=1$ are not presently established.
• Weighted Kernels and Fractal Boundaries: The impact of boundary Assouad dimensions and kernel singularity order is an active area (Acosta et al., 13 Jan 2026).
• Metric Spaces and Variable Exponents: Research into extensions for metric measure spaces with doubling or Poincaré properties and for variable exponent/Orlicz frameworks is suggested (Acosta et al., 13 Jan 2026).
• Nonlocal Trace Theory: The equivalence of projected-difference and Sobolev spaces without vanishing trace is generally open (Mengesha et al., 2020).

An implication is that the landscape of fractional Korn inequalities remains dynamic, with ongoing research refining conditions for validity, boundary regularity, and compatibility with generalized nonlocal models.

---