Dual Sobolev Identity in Analysis

• Dual Sobolev Identity is a duality principle connecting Sobolev energies and deficits with dual objects, preserving the variational structure across classical and fractional frameworks.
• It employs tools such as inverse elliptic operators, Legendre transforms, and divergence representations to bridge analytical gaps between different Sobolev and Hardy–Littlewood–Sobolev settings.
• The identity underpins stability inequalities and bubble decompositions, providing a quantitative mechanism for analyzing critical points and transferring residual information.

Dual Sobolev Identity denotes a family of duality principles in analysis that identify a Sobolev quantity with a dual object, often through an inverse elliptic operator, a Legendre transform, a divergence representation, or a boundary trace form. In the modern fractional critical-point setting, the term refers most specifically to the operator identity introduced by Chen–Lu–Tang, which links the residual of the fractional Sobolev Euler–Lagrange equation to the residual of the Hardy–Littlewood–Sobolev equation through the map $f=|g|^{\frac{4s}{n-2s}}g$, namely

$R(f)=(-\Delta)^{-s}F(g),$

with $F(g)=(-\Delta)^s g-|g|^{\frac{4s}{n-2s}}g$ and $R(f)=|f|^{-\frac{4s}{n+2s}}f-(-\Delta)^{-s}f$ (Chen et al., 19 May 2026). In a broader historical sense, the same expression has also been used for the Euclidean Sobolev–Hardy–Littlewood–Sobolev correspondence, for dual energy formulas on metric measure spaces, for weak$^*$ differentiation identities of Banach-valued Sobolev maps, and for exact bulk–boundary identities in nonlocal extension theory (Dolbeault, 2011).

1. Terminological scope

The phrase has no single universal meaning across all branches of analysis. Rather, it recurs in settings where a Sobolev energy, Sobolev deficit, or Sobolev derivative can be represented by a dual quantity with the same sharp scale, the same extremals, or the same variational structure. This suggests that the expression is best understood as a class of duality mechanisms rather than a single theorem.

Setting	Core dual identity	Source
Euclidean Sobolev–HLS duality	$\int |\nabla u|^2\,dx=\int f\,(-\Delta)^{-1}f\,dx$ when $-\Delta u=f$	(Dolbeault, 2011)
Fractional critical points	$R(f)=(-\Delta)^{-s}F(g)$	(Chen et al., 19 May 2026)
Metric measure spaces	${\sf F}_p(\mu)={\sf D}_q(\mu)$	(Ambrosio et al., 14 Oct 2025)
Dual Banach-valued Sobolev maps	$$\partial_i\langle u(\cdot),y\rangle=\langle \partial_i^{w^*}u(\cdot),y\rangle$$	(Creutz et al., 2023)
Nonlocal extension theory	$R(f)=(-\Delta)^{-s}F(g),$0	(Bogdan et al., 2020)

Among these formulations, the Sobolev–Hardy–Littlewood–Sobolev line is historically central, while the Chen–Lu–Tang framework is the first one in the provided literature to use the identity directly at the level of critical-point stability and Struwe-type decomposition in the fractional regime (Chen et al., 19 May 2026).

2. Classical Sobolev–Hardy–Littlewood–Sobolev duality

In the Euclidean theory for $R(f)=(-\Delta)^{-s}F(g),$1, the sharp Sobolev inequality and the sharp Hardy–Littlewood–Sobolev inequality are dual statements with the same optimal constant. With

$R(f)=(-\Delta)^{-s}F(g),$2

the sharp Sobolev inequality is

$R(f)=(-\Delta)^{-s}F(g),$3

while the dual HLS inequality can be written as

$R(f)=(-\Delta)^{-s}F(g),$4

The exact bridge is the energy identity

$R(f)=(-\Delta)^{-s}F(g),$5

valid when $R(f)=(-\Delta)^{-s}F(g),$6 and $R(f)=(-\Delta)^{-s}F(g),$7 (Dolbeault, 2011).

The variational formulation expresses the same correspondence through Legendre transforms. Writing

$R(f)=(-\Delta)^{-s}F(g),$8

one has

$R(f)=(-\Delta)^{-s}F(g),$9

Thus $F(g)=(-\Delta)^s g-|g|^{\frac{4s}{n-2s}}g$0 is equivalent to $F(g)=(-\Delta)^s g-|g|^{\frac{4s}{n-2s}}g$1, so the Sobolev inequality and the HLS inequality are Legendre dual with matching sharp constant (Dolbeault, 2011).

A stronger quantitative form appears in the completion-of-squares framework. For $F(g)=(-\Delta)^s g-|g|^{\frac{4s}{n-2s}}g$2, $F(g)=(-\Delta)^s g-|g|^{\frac{4s}{n-2s}}g$3, and $F(g)=(-\Delta)^s g-|g|^{\frac{4s}{n-2s}}g$4, one has

$F(g)=(-\Delta)^s g-|g|^{\frac{4s}{n-2s}}g$5

with

$F(g)=(-\Delta)^s g-|g|^{\frac{4s}{n-2s}}g$6

This inequality measures the HLS gap by the Sobolev gap and singles out the Aubin–Talenti family as the common extremal manifold (Dolbeault et al., 2013).

The dynamic version of the same duality uses the fast diffusion equation

$F(g)=(-\Delta)^s g-|g|^{\frac{4s}{n-2s}}g$7

for which the functional

$F(g)=(-\Delta)^s g-|g|^{\frac{4s}{n-2s}}g$8

has nonnegative derivative equal to a positive factor times the Sobolev deficit of $F(g)=(-\Delta)^s g-|g|^{\frac{4s}{n-2s}}g$9. In dimension $R(f)=|f|^{-\frac{4s}{n+2s}}f-(-\Delta)^{-s}f$0, Onofri’s inequality and logarithmic HLS replace the Sobolev–HLS pair, with a corresponding super-fast diffusion flow (Dolbeault, 2011).

3. Fractional critical-point formulation

The most explicit modern use of the expression “Dual Sobolev Identity” is the fractional critical-point framework of Chen–Lu–Tang. Here $R(f)=|f|^{-\frac{4s}{n+2s}}f-(-\Delta)^{-s}f$1, $R(f)=|f|^{-\frac{4s}{n+2s}}f-(-\Delta)^{-s}f$2,

$R(f)=|f|^{-\frac{4s}{n+2s}}f-(-\Delta)^{-s}f$3

and the sharp Hardy–Littlewood–Sobolev and fractional Sobolev inequalities are

$R(f)=|f|^{-\frac{4s}{n+2s}}f-(-\Delta)^{-s}f$4

and

$R(f)=|f|^{-\frac{4s}{n+2s}}f-(-\Delta)^{-s}f$5

The HLS extremals are generated by

$R(f)=|f|^{-\frac{4s}{n+2s}}f-(-\Delta)^{-s}f$6

and the Sobolev extremals by

$R(f)=|f|^{-\frac{4s}{n+2s}}f-(-\Delta)^{-s}f$7

together with translations, dilations, and conformal symmetries (Chen et al., 19 May 2026).

The identity is built from the Nemytskii map

$R(f)=|f|^{-\frac{4s}{n+2s}}f-(-\Delta)^{-s}f$8

With

$R(f)=|f|^{-\frac{4s}{n+2s}}f-(-\Delta)^{-s}f$9

the fundamental residual relation is

$^*$0

This is the core operator identity behind the duality of critical-point stability (Chen et al., 19 May 2026).

Passing to norms yields the quantitative bridge

$^*$1

In this formulation the identity does not merely restate the duality of the inequalities themselves; it transfers residual information between the two Euler–Lagrange equations, which is what allows it to serve as a stability device rather than only a variational correspondence (Chen et al., 19 May 2026).

4. Extremals, stability inequalities, and Struwe-type decomposition

A central feature of the 2026 framework is that the dual identity is used to compare distances to the two extremal manifolds. On the HLS side the natural distance is

$^*$2

which is non-Hilbertian. On the Sobolev side the distance is

$^*$3

which is Hilbertian (Chen et al., 19 May 2026).

For HLS critical points, Chen–Lu–Tang prove a global stability inequality measured in the non-Hilbertian $^*$4 distance: $^*$5 For Sobolev critical points they derive

$^*$6

The second statement is obtained by mapping to the HLS side, applying HLS stability, and returning through the dual residual inequality (Chen et al., 19 May 2026).

The method is tailored to the non-Hilbertian geometry of the HLS problem. The paper decomposes in $^*$7 rather than in $^*$8, proves that the $^*$9 and $\int |\nabla u|^2\,dx=\int f\,(-\Delta)^{-1}f\,dx$0 distances are equivalent at small scales, establishes local stability through spherical harmonics and the Funk–Hecke formula, and then promotes local control to global control by a Struwe-type decomposition and energy decoupling. On the sphere, the remainder is made orthogonal to the first spherical harmonics after conformal transport, reproducing the role of translation and scaling orthogonality in Hilbert settings (Chen et al., 19 May 2026).

The same framework yields one-bubble decomposition results. If a sequence $\int |\nabla u|^2\,dx=\int f\,(-\Delta)^{-1}f\,dx$1 lies in the prescribed energy window and its HLS residual tends to zero, then there exist $\int |\nabla u|^2\,dx=\int f\,(-\Delta)^{-1}f\,dx$2 and $\int |\nabla u|^2\,dx=\int f\,(-\Delta)^{-1}f\,dx$3 such that

$\int |\nabla u|^2\,dx=\int f\,(-\Delta)^{-1}f\,dx$4

Likewise, if $\int |\nabla u|^2\,dx=\int f\,(-\Delta)^{-1}f\,dx$5 lies in the corresponding Sobolev window and its Sobolev residual tends to zero, then

$\int |\nabla u|^2\,dx=\int f\,(-\Delta)^{-1}f\,dx$6

The argument shows that only one bubble survives in the window and uses testing with positive and negative parts to remove nonnegativity assumptions on the HLS side. The same removal then transfers to the Sobolev side for all $\int |\nabla u|^2\,dx=\int f\,(-\Delta)^{-1}f\,dx$7 (Chen et al., 19 May 2026).

The paper also gives an explicit lower bound for Palais–Smale sequences of the HLS integral equation, involving gamma functions and $\int |\nabla u|^2\,dx=\int f\,(-\Delta)^{-1}f\,dx$8. According to the authors, this is the first explicit quantitative lower bound for Palais–Smale sequences in a non-Hilbertian distance (Chen et al., 19 May 2026).

5. Dual energy identities beyond the classical Euclidean setting

In metric measure spaces, the phrase denotes a dual representation of the Sobolev energy in terms of tangent-module vector fields and divergence measures. For $\int |\nabla u|^2\,dx=\int f\,(-\Delta)^{-1}f\,dx$9 and conjugate exponent $-\Delta u=f$0, one has

$-\Delta u=f$1

and for measures $-\Delta u=f$2,

$-\Delta u=f$3

This identity is obtained without doubling, Poincaré inequality, or curvature assumptions, and it ties the primal Cheeger $-\Delta u=f$4-energy to divergence-measure fields in the tangent module (Ambrosio et al., 14 Oct 2025).

For regular symmetric Dirichlet forms, the dual Sobolev viewpoint is encoded by the Sobolev–Bregman form

$-\Delta u=f$5

whenever $-\Delta u=f$6 and $-\Delta u=f$7. The corresponding Hardy–Stein identity reads

$-\Delta u=f$8

so $-\Delta u=f$9 is exactly the dissipation functional of the $R(f)=(-\Delta)^{-s}F(g)$0-norm along the semigroup. The same paper interprets the jump part of $R(f)=(-\Delta)^{-s}F(g)$1 as a Bregman divergence for $R(f)=(-\Delta)^{-s}F(g)$2, thereby placing the duality simultaneously in generator-pairing form and in convexity form (Gutowski et al., 2023).

A further nonlocal realization appears in extension theory for unimodal Lévy generators. If $R(f)=(-\Delta)^{-s}F(g)$3 is exterior data on $R(f)=(-\Delta)^{-s}F(g)$4 and $R(f)=(-\Delta)^{-s}F(g)$5 is its Poisson extension, then the nonlinear nonlocal Douglas identity gives

$R(f)=(-\Delta)^{-s}F(g)$6

This is an exact identity, not merely an equivalence of norms, and it identifies the bulk Sobolev–Bregman energy with a boundary trace form through the Dirichlet-to-Neumann operator (Bogdan et al., 2020).

6. Related specialized usages and analytical significance

A different but structurally analogous usage occurs for Sobolev maps with values in a dual Banach space. For $R(f)=(-\Delta)^{-s}F(g)$7, $R(f)=(-\Delta)^{-s}F(g)$8 belongs to the weak$R(f)=(-\Delta)^{-s}F(g)$9 Sobolev space precisely when there exist weak${\sf F}_p(\mu)={\sf D}_q(\mu)$0 partial derivatives ${\sf F}_p(\mu)={\sf D}_q(\mu)$1 such that

${\sf F}_p(\mu)={\sf D}_q(\mu)$2

for all ${\sf F}_p(\mu)={\sf D}_q(\mu)$3 and ${\sf F}_p(\mu)={\sf D}_q(\mu)$4. Equivalently,

${\sf F}_p(\mu)={\sf D}_q(\mu)$5

in the sense of distributions. The paper proves that ${\sf F}_p(\mu)={\sf D}_q(\mu)$6 as sets, so target duality becomes a differentiation principle for Banach-valued and metric-valued Sobolev maps (Creutz et al., 2023).

In several complex variables, the expression is used in yet another sense. A modified Morrey–Kohn–Hörmander identity on the pseudoconcave side of an annulus yields a coercive mixed ${\sf F}_p(\mu)={\sf D}_q(\mu)$7 estimate, and this is dualized to obtain ${\sf F}_p(\mu)={\sf D}_q(\mu)$8 solvability for ${\sf F}_p(\mu)={\sf D}_q(\mu)$9 on the hole. In that literature, the “dual Sobolev identity/estimate” refers to the passage from a pseudoconcave $$\partial_i\langle u(\cdot),y\rangle=\langle \partial_i^{w^*}u(\cdot),y\rangle$$0 identity to a first-order Sobolev estimate on the complementary domain (Chakrabarti et al., 2018).

A higher-order norm-representation version appears in the Brezis–Seeger–Van Schaftingen–Yung framework on ball Banach function spaces. For $$\partial_i\langle u(\cdot),y\rangle=\langle \partial_i^{w^*}u(\cdot),y\rangle$$1, under the sharp constraint $$\partial_i\langle u(\cdot),y\rangle=\langle \partial_i^{w^*}u(\cdot),y\rangle$$2, the paper proves

$$\partial_i\langle u(\cdot),y\rangle=\langle \partial_i^{w^*}u(\cdot),y\rangle$$3

with a corresponding limiting identity involving averages of directional $$\partial_i\langle u(\cdot),y\rangle=\langle \partial_i^{w^*}u(\cdot),y\rangle$$4-th derivatives on $$\partial_i\langle u(\cdot),y\rangle=\langle \partial_i^{w^*}u(\cdot),y\rangle$$5. Even when $$\partial_i\langle u(\cdot),y\rangle=\langle \partial_i^{w^*}u(\cdot),y\rangle$$6, the higher-order formulas for $$\partial_i\langle u(\cdot),y\rangle=\langle \partial_i^{w^*}u(\cdot),y\rangle$$7 are stated to be completely new (Hu et al., 22 May 2025).

Taken together, these formulations indicate a persistent analytic pattern. The term “Dual Sobolev Identity” is used whenever a Sobolev object can be recovered, controlled, or exactly represented by a dual quantity that preserves the natural scaling and extremal structure of the problem. In the classical Euclidean theory the dual object is the HLS energy; in the fractional critical-point theory it is the HLS residual; in metric and nonsmooth settings it is a divergence measure or tangent field; in semigroup theory it is the generator pairing; and in nonlocal trace theory it is a boundary Bregman form. The Chen–Lu–Tang formulation is distinctive because it operates directly at the level of critical points, stability inequalities, and Palais–Smale analysis, thereby turning Sobolev–HLS duality into a quantitative mechanism for bubble decomposition and stability transfer (Chen et al., 19 May 2026).