Paperlane

no. 16 · Tuesday 17 November 2026

The Optimal Interface Method: A Modelling Paradigm for Two-Body Contact Problems via Optimal Control and $Γ$-convergence

Zhuopeng Liu and Paolo Piersanti

math.OC

Abstract

This paper is concerned with the derivation of a model governing the equilibrium configuration of two linearly elastic bodies that may come into contact with each other. The contact zone, or coincidence set, is an \emph{a priori} unknown of the model, and the equilibrium configuration is governed by variational inequalities. We introduce a new modelling paradigm in which the contact interface between the two linearly elastic bodies under consideration is now regarded as the \emph{primary variable} of an optimal control problem whose states are the unique solutions of two \emph{independent} unilateral contact problems for which this very contact interface represents the obstacle itself. As a result, the free boundary emerges as the solution of a low-dimensional optimisation problem rather than as the active set of coupled variational inequalities. We establish the existence and uniqueness of optimal controls minimising the reduced cost functional arising from the modelling paradigm introduced beforehand, and we show that the family of marginal energy functionals associated with such model $Γ$-converges to the energy functional associated with the original variational inequalities as the positive \emph{cost parameter} $β$ in the reduced cost functional tends to zero. This $Γ$-convergence argument also yields a characterisation of the coincidence set of the limit problem in terms of the optimal interfaces. We further show that, after a suitable regularisation, the reduced cost functional is Gâteaux differentiable, and that its Gâteaux derivative can be explicitly computed \emph{without solving any adjoint system}. Finally, we corroborate the $Γ$-convergence result obtained beforehand with numerical experiments on a simplified and exactly solvable configuration.

fields
Optimization and Control, Analysis of PDEs
pdf ↗ page arXiv:2611.20369 ↗