Abstract:We introduce a novel method for solving density-based topology optimization problems: \underline{Si}gmoidal \underline{M}irror descent with a \underline{P}rojected \underline{L}atent variable (SiMPL). The SiMPL method (pronounced as "the simple method") optimizes a design using only first-order derivative information of the objective function. The bound constraints on the density field are enforced with the help of the (negative) Fermi--Dirac entropy, which is also used to define a non-symmetric distance function called a Bregman divergence on the set of admissible designs. This Bregman divergence leads to a simple update rule that is further simplified with the help of a so-called latent variable. %Introducing a generalized Barzilai-Borwein step size rule accelerates the convergence of SiMPL. Because the SiMPL method involves discretizing the latent variable, it produces a sequence of pointwise-feasible iterates, even when high-order finite elements are used in the discretization. Numerical experiments demonstrate that the method outperforms other popular first-order optimization algorithms. To outline the general applicability of the technique, we include examples with (self-load) compliance minimization and compliant mechanism optimization problems.

What problem does this paper attempt to address?

The problems that this paper attempts to solve are some key challenges in density - based topology optimization (DBTO). Specifically, the author introduces a new method named SiMPL (Sigmoidal Mirror descent with a Projected Latent variable), aiming to optimize the design by using only the first - order derivative information of the objective function and ensuring that the boundary constraints of the density field \(0 \leq \rho \leq 1\) are strictly satisfied during the optimization process. The following are the specific problems that the paper attempts to solve: 1. **Optimization efficiency and stability**: Existing optimization algorithms such as Optimality Criteria (OC) and Method of Moving Asymptotes (MMA) have some limitations, for example, they require complex parameter adjustments, are highly mesh - dependent, and their update rules lack a theoretical basis. The SiMPL method aims to provide a more efficient and stable optimization strategy. 2. **Strict satisfaction of boundary constraints**: An important problem in DBTO is to ensure that the density field always satisfies the boundary constraints of \(0 \leq \rho \leq 1\) during the optimization process. The SiMPL method defines an asymmetric distance function (Bregman divergence) by introducing negative Fermi - Dirac entropy and combines it with projection operations to ensure that these constraints are strictly satisfied at each iteration step. 3. **Feasibility under high - order discretization**: Traditional optimization methods may generate infeasible design points when using high - order finite - element discretization. The SiMPL method introduces the so - called latent variable representation method, enabling the generation of point - by - point feasible iteration sequences even under high - order discretization. 4. **Step - size selection strategy**: Initial experiments show that the choice of step - size has a significant impact on the convergence and the quality of the optimization solution. The SiMPL method solves this problem through adaptive step - size strategies (such as those related to the Barzilai - Borwein method) and backtracking line - search algorithms, thereby reducing the number of iterations and improving the optimization efficiency. 5. **General applicability**: To verify the wide applicability of the SiMPL method, the paper demonstrates its application in different types of optimization problems, including minimum compliance problems and flexible mechanism optimization problems. In summary, by introducing the SiMPL method, this paper solves key problems in density - based topology optimization such as optimization efficiency, strict satisfaction of boundary constraints, feasibility under high - order discretization, and step - size selection, providing a more efficient and reliable optimization tool for engineering design.

A Simple Introduction to the SiMPL Method for Density-Based Topology Optimization

Analysis of the SiMPL method for density-based topology optimization

An asymptotically concentrated method for structural topology optimization based on the SIMLF interpolation

An ODE-driven level-set density method for topology optimization

Explicit Tunnels and Cavities Control Using SIMP and MMA in Structural Topology Optimization

A Novel 3-D Topology Optimization Methodology Based on the Min-Cut Theorem

A novel quasi-smooth tetrahedral numerical manifold method and its application in topology optimization based on parameterized level-set method

Topology Optimization on Two-Dimensional Manifolds

A smooth single-variable-based interpolation function for multi-material topology optimization

Level set topology and shape optimization by density methods using cut elements with length scale control

Multi-material topology optimization based on multiple simp of variable density method

Application of the New Interpolation Functions on Topology Optimization of Continuum Structure for Static and Mode Analysis

Multiscale topology optimization for the design of spatially-varying three-dimensional lattice structure

Some considerations on multi-material topology optimization using ordered SIMP

Bi-material Topology Optimization Using Analysis Mesh-Independent Point-Wise Density Interpolation

Accelerated topology optimization by means of deep learning

An unified material interpolation for topology optimization of multi-materials

Sequential conservative integer programming method for multi-constrained discrete variable structure topology optimization

Momentum-based Accelerated Mirror Descent Stochastic Approximation for Robust Topology Optimization under Stochastic Loads

Density-based isogeometric topology optimization of shell structures

Converting pixel‐type topology optimization results to MMC‐representation based on sparse optimization and its applications