A multimaterial topology optimisation approach to Dirichlet control with piecewise constant functions

TL;DR

This paper develops a novel multimaterial topology optimization method for Dirichlet control problems with piecewise constant functions, addressing non-convexity and irregularity issues, and demonstrates its effectiveness through numerical examples.

Contribution

It introduces a weak solution framework and topological derivatives for multimaterial Dirichlet control, enabling advanced optimization techniques.

Findings

Derived first-order optimality conditions using topological derivatives.

Proved existence of weak topological state derivatives for multimaterial controls.

Implemented a level-set algorithm with finite element software for 3D examples.

Abstract

In this paper we study a Dirichlet control problem for the Poisson equation, where the control is assumed to be piecewise constant function which is allowed to take M > 1 different values. The space of admissible Dirichlet controls is non-convex and therefore standard derivatives in Banach spaces are not applicable. Furthermore piecewise constant functions are too irregular that the standard extension techniques apply. Therefore we resort to the notion of very weak solutions of the state equation in Lp spaces. We then study the differentiability of the shape-to-state operator of this problem and derive the first order necessary optimality conditions using the topological state derivative. In fact we prove the existence of the weak topological state derivative introduced for a multimaterial shape functional which is then expressed via an adjoint variable. The topological derivative resembles formulas found for derivative in the more standard Dirichlet control problems. In the final part of the paper we show how to apply a multimaterial level-set algorithm with the finite element software NGSolve and present several numerical examples in dimension three.