Cartesian convexity as the key notion in the variational existence theory for nonlocal supremal functionals

TL;DR

This paper characterizes the weakly* lower semicontinuity of nonlocal supremal functionals through a new convexity notion called Cartesian level convexity, extending the analysis in the calculus of variations for $L^{ obreak ext{infty}}$.

Contribution

It introduces Cartesian level convexity as the key convexity notion for nonlocal supremal functionals and provides criteria for structure-preservation and representation formulas.

Findings

Cartesian level convexity characterizes weakly* lower semicontinuity.

Relaxed functionals may not preserve the supremal form, with criteria for when structure is maintained.

Connection established between supremal functionals and double integrals via $L^p$-approximation and $ ext{Gamma}$-convergence.

Abstract

Motivated by the direct method in the calculus of variations in $L^{\infty}$, our main result identifies the notion of convexity characterizing the weakly$^*$ lower semicontinuity of nonlocal supremal functionals: Cartesian level convexity. This new concept coincides with separate level convexity in the one-dimensional setting and is strictly weaker for higher dimensions. We discuss relaxation in the vectorial case, showing that the relaxed functional will not generally maintain the supremal form. Apart from illustrating this fact with examples of multi-well type, we present precise criteria for structure-preservation. When the structure is preserved, a representation formula is given in terms of the Cartesian level convex envelope of the (diagonalized) original supremand. This work does not only complete the picture of the analysis initiated in [Kreisbeck \& Zappale, Calc.~Var.~PDE, 2020], but also establishes a connection with double integrals. We relate the two classes of functionals via an $L^p$-approximation in the sense of $\Gamma$-convergence for diverging integrability exponents. The proofs exploit recent results on nonlocal inclusions and their asymptotic behavior, and use tools from Young measure theory and convex analysis.