- The paper proves that every continuous local functional on convex functions is a valuation, extending Hadwiger- and Alesker-type classifications.
- It establishes explicit integral representations by expressing functionals via integrals of polynomials in the function, its gradient, and its Hessian with a measure decomposition.
- The work leverages Paley–Wiener–Schwartz theory and module methods to connect Fourier–Laplace transforms with density results in affine-invariant submodules.
Integral Representation of Polynomial Local Functionals on Convex Functions
Introduction and Motivation
The paper "Integral representation of polynomial local functionals on convex functions" (2604.24485) provides a systematic treatment of the representation and structure of polynomial local functionals on spaces of convex functions $\Conv(\mathbb{R}^n, \mathbb{R})$, with particular focus on continuous and smooth functionals. It advances the understanding of local functionals in convex geometry, with integral representation results linked to polynomial structures and modules of polynomials associated to minors of the Hessian. The theoretical context includes geometric valuation theory, functional analysis, and the calculus of variations, addressing extension and classification challenges which arise both in classical and modern geometric applications.
Polynomial Local Functionals and Valuation Structure
A central object considered in this work is the class of local functionals, maps $\Psi:\Conv(\mathbb{R}^n, \mathbb{R})\to M(\mathbb{R}^n)$, with local determination implying invariance with respect to local modifications of the function. The Dirichlet energy, Monge–Ampère-type operators, and classical variational functionals are prototypical examples. By focusing on polynomial local functionals, i.e., those for which the dependence on affine translations is polynomial (not merely linear or invariant), the paper extends the classical theory of translation-invariant valuations on convex bodies to a richer algebraic setting.
The main structural result, building on prior work [KnoerrPolynomiallocalfunctionals2025], establishes that every continuous local functional on convex functions is a valuation, leveraging Błocki’s result for the complex Monge–Ampère operator. This extends Hadwiger-type and Alesker-type classification tools from convex bodies to convex functions. The polynomial degree and homogeneity structure are exploited to decompose the space $P_d\LV(\mathbb{R}^n)$, the space of continuous degree-d polynomial local functionals, into homogeneous components by the scaling action.
Integral Representation and Module Structure
The main technical results are the explicit integral representations for $P_d\LV(\mathbb{R}^n)^{tr}$, the space of translation-invariant degree-d polynomial local functionals. The author establishes that each such functional, for sufficiently regular (twice differentiable) convex functions f, can be written in terms of integrals of polynomials in f, its gradient df, and its Hessian D2f:
$\Psi:\Conv(\mathbb{R}^n, \mathbb{R})\to M(\mathbb{R}^n)$0
where $\Psi:\Conv(\mathbb{R}^n, \mathbb{R})\to M(\mathbb{R}^n)$1 is a unique polynomial in $\Psi:\Conv(\mathbb{R}^n, \mathbb{R})\to M(\mathbb{R}^n)$2, the space generated by $\Psi:\Conv(\mathbb{R}^n, \mathbb{R})\to M(\mathbb{R}^n)$3-minors of symmetric matrices up to degree $\Psi:\Conv(\mathbb{R}^n, \mathbb{R})\to M(\mathbb{R}^n)$4. The translation-invariant part is thus a finite-dimensional module, whose dimension $\Psi:\Conv(\mathbb{R}^n, \mathbb{R})\to M(\mathbb{R}^n)$5 is computed in terms of minors and symmetry.
For general continuous polynomial local functionals (not necessarily translation-invariant), the paper goes further: any such $\Psi:\Conv(\mathbb{R}^n, \mathbb{R})\to M(\mathbb{R}^n)$6 admits a unique decomposition in terms of measures $\Psi:\Conv(\mathbb{R}^n, \mathbb{R})\to M(\mathbb{R}^n)$7 and a basis $\Psi:\Conv(\mathbb{R}^n, \mathbb{R})\to M(\mathbb{R}^n)$8 of $\Psi:\Conv(\mathbb{R}^n, \mathbb{R})\to M(\mathbb{R}^n)$9:
$P_d\LV(\mathbb{R}^n)$0
Here, the measures $P_d\LV(\mathbb{R}^n)$1 encode non-invariant aspects of $P_d\LV(\mathbb{R}^n)$2 and are not necessarily absolutely continuous with respect to Lebesgue measure; determining their properties is generally non-trivial.
Analogous results are given for smooth functionals, where a characterization as a free $P_d\LV(\mathbb{R}^n)$3-module of rank $P_d\LV(\mathbb{R}^n)$4 is proved. This module-theoretic perspective provides algebraic control over the structure and density of smooth local functionals within the continuous space.
Paley–Wiener–Schwartz-Type Classification
A pivotal aspect of the proof is the adaptation of Paley–Wiener–Schwartz theory to the Fourier–Laplace transforms of Goodey–Weil distributions (distributions associated to valuations). By characterizing these transforms as belonging to specifically generated modules of holomorphic functions (by minors), the author reduces the integral representation problem to the algebraic structure of modules with explicit generators. This method not only yields uniqueness results for integral representations but also provides explicit estimates and an understanding of support and regularity.
Density and Affine Invariant Submodules
The paper establishes that for degree-zero homogeneous components and, more generally, for affine-invariant $P_d\LV(\mathbb{R}^n)$5-submodules of $P_d\LV(\mathbb{R}^n)$6, any nontrivial module is dense in the total space with respect to suitable topologies. This has strong consequences for Monge–Ampère-type operators: e.g., the module generated by mixed Monge–Ampère operators (on affine-invariant families) is either trivial or dense. The irreducibility result for the module of translation-invariant functionals plays a crucial role, linking algebraic and geometric aspects.
Implications and Future Developments
The integral representation and density theorems have significant practical implications in geometric analysis, convex optimization, and variational calculus. They provide tools for the systematic extension, classification, and algebraic manipulation of functionals relevant in PDEs, geometric flows, and metric geometry. The explicit connection to polynomial minor modules and distribution theory suggests potential for implementation in computational convex geometry and for further investigation into non-smooth extensions, regularity of measures, and applications to plurisubharmonic and tropical convexity.
Theoretically, the results strengthen the bridge between classical valuation theory, functional analysis, and modern convex function theory. The module-theoretic and Paley–Wiener–Schwartz approaches are likely to be valuable in extending to other function spaces (e.g., Sobolev spaces, spaces of plurisubharmonic functions), investigating invariant functionals under broader symmetry groups, and exploring connections to operator theory and harmonic analysis.
Conclusion
This paper synthesizes and generalizes deep results on the structure of polynomial local functionals on convex functions, contributing integral representations, module-theoretic classifications, and density results. The combination of algebraic and analytic techniques offers a robust framework for geometric, algebraic, and analytic treatment of valuations and their extensions, laying the foundation for further theoretical exploration and practical application in convex and geometric analysis.
