Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Classical Optimal Designs for Stationary Diffusion with Multiple Phases

Matko Grbac, Ivan Ivec, Marko Vrdoljak

TL;DR

This work develops a homogenization-based relaxation framework for optimal designs in stationary diffusion with multiple anisotropic phases, aiming to maximize a weighted sum of state energies and to identify when the optimal design is bang-bang (composed of pure materials). A dual flux formulation yields a saddle-point characterization that guides the determination of optimal local material proportions, with a complete description in the radially symmetric ball, where explicit classical solutions can be computed. The analysis reveals constructive, layer-by-layer designs (sequential laminates) determined by level-set thresholds of the auxiliary function $\\psi$, and it provides explicit radial solutions and radii in several illustrative examples, including single- and multi-state problems and scenarios with nonuniqueness. The results offer rigorous benchmarks for numerical methods in optimal design and highlight how symmetry and material eigenvalue ordering influence the emergence of classical designs. The framework connects relaxation, duality, and explicit radial constructions to yield practical benchmarks for anisotropic composite design.

Abstract

We study optimal design problems for stationary diffusion involving one or more state equations and mixtures of an arbitrary number of anisotropic materials. Since such problems typically do not admit classical solutions, we adopt a homogenization-based relaxation framework. The objective considered is the maximization of a weighted sum of the energies associated with each state equation, with particular emphasis on identifying cases in which the optimal design is classical, that is, of bang-bang type, composed solely of the original pure materials. Such cases provide valuable benchmarks for numerical methods in optimal design. A simplified optimization problem expressed in terms of local material proportions is analyzed through a dual formulation in terms of fluxes. Using a saddle-point characterization, we establish a complete description of its optimal solutions. The proposed approach is applied in detail to spherically symmetric problems. In the case of a ball, the method yields explicit classical solutions of the homogenization-based relaxation problem.

Classical Optimal Designs for Stationary Diffusion with Multiple Phases

TL;DR

This work develops a homogenization-based relaxation framework for optimal designs in stationary diffusion with multiple anisotropic phases, aiming to maximize a weighted sum of state energies and to identify when the optimal design is bang-bang (composed of pure materials). A dual flux formulation yields a saddle-point characterization that guides the determination of optimal local material proportions, with a complete description in the radially symmetric ball, where explicit classical solutions can be computed. The analysis reveals constructive, layer-by-layer designs (sequential laminates) determined by level-set thresholds of the auxiliary function , and it provides explicit radial solutions and radii in several illustrative examples, including single- and multi-state problems and scenarios with nonuniqueness. The results offer rigorous benchmarks for numerical methods in optimal design and highlight how symmetry and material eigenvalue ordering influence the emergence of classical designs. The framework connects relaxation, duality, and explicit radial constructions to yield practical benchmarks for anisotropic composite design.

Abstract

We study optimal design problems for stationary diffusion involving one or more state equations and mixtures of an arbitrary number of anisotropic materials. Since such problems typically do not admit classical solutions, we adopt a homogenization-based relaxation framework. The objective considered is the maximization of a weighted sum of the energies associated with each state equation, with particular emphasis on identifying cases in which the optimal design is classical, that is, of bang-bang type, composed solely of the original pure materials. Such cases provide valuable benchmarks for numerical methods in optimal design. A simplified optimization problem expressed in terms of local material proportions is analyzed through a dual formulation in terms of fluxes. Using a saddle-point characterization, we establish a complete description of its optimal solutions. The proposed approach is applied in detail to spherically symmetric problems. In the case of a ball, the method yields explicit classical solutions of the homogenization-based relaxation problem.
Paper Structure (9 sections, 3 theorems, 65 equations, 4 figures, 1 table)

This paper contains 9 sections, 3 theorems, 65 equations, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Formulation and Relaxation of the Problem
  3. Interpretation through a minimax problem
  4. Maximum points of $I$ on ${\cal T}$
  5. Radially Symmetric Problems
  6. Examples
  7. Single State Problem with Unique and Classical Solution
  8. Appearance of Nonunique Classical Designs
  9. Multiple State Problem

Key Result

Lemma 3.1

For any ${\bm{\theta}}\in{\cal T}$ we have and the minimum is attained at unique ${\bm{\sigma}}\in {\cal S}$ given by ${\bm{\sigma}}_i=\lambda_-({\bm{\theta}})\nabla u_i$, where $u_i$ solves semt, for any $i=1,\ldots, m$.

Figures (4)

  • Figure 1: Example of a case where $\alpha_1 > \alpha_2 > \alpha_3$. This is a bang–bang solution in which the open strips are filled with the corresponding materials, while the level set $\psi = \alpha_2$ is non-negligible and is occupied by materials ${\bf{M}}_2$ and ${\bf{M}}_3$.
  • Figure 2: Example where $\alpha_1 > \alpha_2 = \alpha_3 > \alpha_4$. This is a bang–bang solution in which, on the horizontal level set $\psi = \alpha_2 = \alpha_3$, one uses the remaining amount of material ${\bf{M}}_2$, the fully prescribed material ${\bf{M}}_3$ (as implied by $\alpha_2=\alpha_3$), and a portion of material ${\bf{M}}_4$.
  • Figure 3: Graph of the function $\psi$ for the example in Subsection \ref{['pr1']}, with the radii indicating the interfaces between the materials
  • Figure 4: Graph of the function $\psi$ for the example in Subsection \ref{['pr3']}, with the radii indicating the interfaces between the materials

Theorems & Definitions (10)

  • Lemma 3.1
  • Remark 3.2
  • Lemma 3.3
  • proof
  • Remark 3.4
  • Remark 3.5
  • Theorem 4.1
  • Remark 4.2
  • proof
  • Remark 4.3