[Paper Review] On the foundations of nonlinear generalized functions I

Éva Farkas, Michael Grosser+2 more|ArXiv.org|Dec 28, 1999

Mathematical and Theoretical Analysis1 references57 citations

TL;DR

This paper establishes a diffeomorphism-invariant differential algebra of generalized functions, extending Colombeau algebras to resolve foundational issues in nonlinear distribution theory. By employing calculus on convenient vector spaces and introducing a rigorous, invariant construction via smooth functions on $Ω_\varepsilon$, it enables consistent treatment of nonlinear PDEs with singular data, ensuring compatibility with smooth operations and diffeomorphism invariance.

ABSTRACT

We construct a diffeomorphism invariant (Colombeau-type) differential algebra canonically containing the space of distributions in the sense of L. Schwartz. Employing differential calculus in infinite dimensional (convenient) vector spaces, previous attempts in this direction are unified and completed. Several classification results are achieved and applications to nonlinear differential equations involving singularities are given.

Motivation & Objective

• To resolve the foundational issue of diffeomorphism invariance in Colombeau-type algebras of generalized functions.
• To unify and complete prior attempts at constructing nonlinear generalized function algebras using infinite-dimensional differential calculus.
• To provide a canonical embedding of distributions into a differential algebra that preserves pointwise products of smooth functions and supports nonlinear operations.
• To establish a framework compatible with nonlinear PDEs involving singular coefficients, data, or solutions.
• To overcome limitations of earlier constructions that failed to preserve invariance under diffeomorphisms, particularly those based on non-invariant test objects.

Proposed method

• Constructs a differential algebra of generalized functions using smooth functions on sets $U_\varepsilon(\Omega)$, which are not linear spaces but admit calculus via convenient vector space theory.
• Applies differential calculus in convenient vector spaces to define smoothness and differentiability of generalized functions, replacing earlier approaches based on Silva-differentiability.
• Introduces a translation formalism between C- and J-formalisms to unify the construction and ensure consistency across different representations.
• Employs a representative-based approach with nets $R \circ S^\varepsilon$ to define generalized functions, ensuring moderateness and null conditions via estimates derived from Gronwall’s inequality.
• Uses Volterra integral equations and Kirchhoff’s formula to derive existence and uniqueness results for nonlinear PDEs, with smooth dependence on parameters ensured by smooth curve composition.
• Applies Theorem 7.12 and Remark 6.7 to justify interchanging differentiation with respect to test functions and integration, enabling rigorous derivation of $\mathcal{E}_M$ and $\mathcal{N}$ estimates.

Experimental results

Research questions

• RQ1 Can a Colombeau-type algebra be constructed such that it is invariant under diffeomorphisms of the underlying domain?
• RQ2 Is it possible to achieve a canonical embedding of distributions into a differential algebra that preserves the pointwise product of smooth functions?
• RQ3 How can differential calculus be consistently applied on non-linear spaces like $U_\varepsilon(\Omega)$ to define smoothness of generalized functions?
• RQ4 Can the moderateness and null conditions in generalized function algebras be verified efficiently using infinite-dimensional calculus?
• RQ5 To what extent can nonlinear PDEs with singular data be solved within this framework, and how is uniqueness established?

Key findings

• The paper constructs a diffeomorphism-invariant differential algebra of generalized functions that canonically contains the space of distributions in the sense of Schwartz.
• The construction resolves the long-standing issue of non-invariance in earlier Colombeau algebras by using a rigorous framework based on convenient vector spaces and smooth curves.
• Moderateness of generalized functions is verified via iterated differentials and Gronwall’s inequality, avoiding direct and unwieldy estimation of nets.
• Uniqueness of solutions to nonlinear ODEs and wave equations is established without explicit differentiation, relying on the structure of the algebra and the properties of the representative maps.
• The approach enables smooth dependence of solutions on parameters and test functions, ensuring compatibility with nonlinear operations and PDE theory.
• The framework supports applications to semilinear wave equations with singular data, extending prior results from classical Colombeau theory to the invariant setting.