Populating cellular metamaterials on the extrema of attainable elasticity through neuroevolution

• URL: http://arxiv.org/abs/2412.11112v3
• Date: Wed, 25 Dec 2024 06:08:22 GMT
• Title: Populating cellular metamaterials on the extrema of attainable elasticity through neuroevolution
• Authors: Maohua Yan, Ruicheng Wang, Ke Liu,
• Abstract summary: Trade-offs between different mechanical properties of materials pose challenges in engineering material design. We employ a neuroevolution algorithm to efficiently solve a multi-objective optimization (MOO) problem. Our method serves as a universal framework for the computational discovery of diverse metamaterials across a range of fields.
• Score: 4.172612182606883
• License:
• Abstract: The trade-offs between different mechanical properties of materials pose fundamental challenges in engineering material design, such as balancing stiffness versus toughness, weight versus energy-absorbing capacity, and among the various elastic coefficients. Although gradient-based topology optimization approaches have been effective in finding specific designs and properties, they are not efficient tools for surveying the vast design space of metamaterials, and thus unable to reveal the attainable bound of interdependent material properties. Other common methods, such as parametric design or data-driven approaches, are limited by either the lack of diversity in geometry or the difficulty to extrapolate from known data, respectively. In this work, we formulate the simultaneous exploration of multiple competing material properties as a multi-objective optimization (MOO) problem and employ a neuroevolution algorithm to efficiently solve it. The Compositional Pattern-Producing Networks (CPPNs) is used as the generative model for unit cell designs, which provide very compact yet lossless encoding of geometry. A modified Neuroevolution of Augmenting Topologies (NEAT) algorithm is employed to evolve the CPPNs such that they create metamaterial designs on the Pareto front of the MOO problem, revealing empirical bounds of different combinations of elastic properties. Looking ahead, our method serves as a universal framework for the computational discovery of diverse metamaterials across a range of fields, including robotics, biomedicine, thermal engineering, and photonics.

Related papers

• Generalized Factor Neural Network Model for High-dimensional Regression [50.554377879576066]
  We tackle the challenges of modeling high-dimensional data sets with latent low-dimensional structures hidden within complex, non-linear, and noisy relationships. Our approach enables a seamless integration of concepts from non-parametric regression, factor models, and neural networks for high-dimensional regression.
  arXiv  Detail & Related papers  (2025-02-16T23:13:55Z)
• Consistent machine learning for topology optimization with microstructure-dependent neural network material models [0.0]
  We present a framework for multiscale structures with spatially varying microstructural symmetry and differentiably different microstructural descriptors. Our findings highlight the potential of integrating consistency with density-based design optimization.
  arXiv  Detail & Related papers  (2024-08-25T14:17:43Z)
• Machine Learning-Guided Design of Non-Reciprocal and Asymmetric Elastic Chiral Metamaterials [0.0]
  This paper defines several design spaces for chiral metamaterials, including the ligament contact angles, the ligament shape, and circle radius. We then leverage machine learning approaches, and specifically Bayesian optimization, to determine optimally performing designs satisfying maximal non-reciprocity or stiffness asymmetry. Our analysis of the underlying mechanisms reveals that chiral metamaterials that can display multiple different contact states under loading in different directions are able to simultaneously exhibit both high non-reciprocity and stiffness asymmetry.
  arXiv  Detail & Related papers  (2024-04-19T23:39:56Z)
• Discovering Interpretable Physical Models using Symbolic Regression and Discrete Exterior Calculus [55.2480439325792]
  We propose a framework that combines Symbolic Regression (SR) and Discrete Exterior Calculus (DEC) for the automated discovery of physical models. DEC provides building blocks for the discrete analogue of field theories, which are beyond the state-of-the-art applications of SR to physical problems. We prove the effectiveness of our methodology by re-discovering three models of Continuum Physics from synthetic experimental data.
  arXiv  Detail & Related papers  (2023-10-10T13:23:05Z)
• Differentiable graph-structured models for inverse design of lattice materials [0.0]
  Architected materials possessing physico-chemical properties adaptable to disparate environmental conditions embody a disruptive new domain of materials science. We propose a new computational approach using graph-based representation for regular and irregular lattice materials.
  arXiv  Detail & Related papers  (2023-04-11T18:00:21Z)
• A Pareto-optimal compositional energy-based model for sampling and optimization of protein sequences [55.25331349436895]
  Deep generative models have emerged as a popular machine learning-based approach for inverse problems in the life sciences. These problems often require sampling new designs that satisfy multiple properties of interest in addition to learning the data distribution.
  arXiv  Detail & Related papers  (2022-10-19T19:04:45Z)
• Deep learning for the rare-event rational design of 3D printed multi-material mechanical metamaterials [0.0]
  Multi-material 3D printing techniques have paved the way for the rational design of metamaterials. We study the resulting anisotropic mechanical properties of the network in general and the rare designs in particular. Deep learning-based algorithms can accurately predict the mechanical properties of the different designs.
  arXiv  Detail & Related papers  (2022-04-04T18:04:23Z)
• Learning the nonlinear dynamics of soft mechanical metamaterials with graph networks [3.609538870261841]
  We propose a machine learning approach to study the dynamics of soft mechanical metamaterials. The proposed approach can significantly reduce the computational cost when compared to direct numerical simulation.
  arXiv  Detail & Related papers  (2022-02-24T00:20:28Z)
• How to See Hidden Patterns in Metamaterials with Interpretable Machine Learning [82.67551367327634]
  We develop a new interpretable, multi-resolution machine learning framework for finding patterns in the unit-cells of materials. Specifically, we propose two new interpretable representations of metamaterials, called shape-frequency features and unit-cell templates.
  arXiv  Detail & Related papers  (2021-11-10T21:19:02Z)
• A deep learning driven pseudospectral PCE based FFT homogenization algorithm for complex microstructures [68.8204255655161]
  It is shown that the proposed method is able to predict central moments of interest while being magnitudes faster to evaluate than traditional approaches. It is shown, that the proposed method is able to predict central moments of interest while being magnitudes faster to evaluate than traditional approaches.
  arXiv  Detail & Related papers  (2021-10-26T07:02:14Z)
• GeoMol: Torsional Geometric Generation of Molecular 3D Conformer Ensembles [60.12186997181117]
  Prediction of a molecule's 3D conformer ensemble from the molecular graph holds a key role in areas of cheminformatics and drug discovery. Existing generative models have several drawbacks including lack of modeling important molecular geometry elements. We propose GeoMol, an end-to-end, non-autoregressive and SE(3)-invariant machine learning approach to generate 3D conformers.
  arXiv  Detail & Related papers  (2021-06-08T14:17:59Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.