Diverse Topology Optimization using Modulated Neural Fields
Not on the map. Why?
Abstract (arXiv)
Topology optimization (TO) is a family of computational methods that derive near-optimal geometries from formal problem descriptions. Despite their success, established TO methods are limited to generating single solutions, restricting the exploration of alternative designs. To address this limitation, we introduce Topology Optimization using Modulated Neural Fields (TOM) - a data-free method that trains a neural network to generate structurally compliant shapes and explores diverse solutions through an explicit diversity constraint. The network is trained with a solver-in-the-loop, optimizing the material distribution in each iteration. The trained model produces diverse shapes that closely adhere to the design requirements. We validate TOM on 2D and 3D TO problems. Our results show that TOM generates more diverse solutions than any previous method, all while maintaining near-optimality and without relying on a dataset. These findings open new avenues for engineering and design, offering enhanced flexibility and innovation in structural optimization.
Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 31 of its 47 references that are in the map. First posted on arXiv 17 Feb 2025.
Builds on
31 of 47 references.
References · 31
- 2020Denoising Diffusion Probabilistic Modelscited by 21,076
- 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,058
- 2022Generative Adversarial Networkscited by 5,627
- 2019DeepSDF: Learning Continuous Signed Distance Functions for Shape Representationcited by 2,689
- 2020Implicit Neural Representations with Periodic Activation Functionscited by 2,238
- 2018Occupancy Networks: Learning 3D Reconstruction in Function Spacecited by 2,041
- 2018Learning Implicit Fields for Generative Shape Modelingcited by 1,060
- 2016HyperNetworkscited by 1,027
- 20233DShape2VecSet: A 3D Shape Representation for Neural Fields and Generative Diffusion Modelscited by 458
- 2019SAL: Sign Agnostic Learning of Shapes from Raw Datacited by 390
- 2023WIRE: Wavelet Implicit Neural Representationscited by 220
- 2020Solver-in-the-Loop: Learning from Differentiable Physics to Interact with Iterative PDE-Solverscited by 205
19 more
- 2021Modulated Periodic Activations for Generalizable Local Functional Representationscited by 140
- 2016Mode Regularized Generative Adversarial Networkscited by 123
- 2022Diffusion Models Beat GANs on Topology Optimizationcited by 79
- 2020TopologyGAN: Topology Optimization Using Generative Adversarial Networks Based on Physical Fields Over the Initial Domaincited by 65
- 2019Neural reparameterization improves structural optimizationcited by 44
- 2024Neural Redshift: Random Networks are not Random Functionscited by 43
- 2022A Level Set Theory for Neural Implicit Evolution under Explicit Flowscited by 42
- 2021NTopo: Mesh-free Topology Optimization using Implicit Neural Representationscited by 41
- 2017Neural networks for topology optimizationcited by 39
- 2022On the use of Artificial Neural Networks in Topology Optimisationcited by 36
- 2024Gauss-Newton Natural Gradient Descent for Physics-Informed Computational Fluid Dynamicscited by 28
- 2022Attention Beats Concatenation for Conditioning Neural Fieldscited by 20
- 2023An adaptive augmented Lagrangian method for training physics and equality constrained artificial neural networkscited by 16
- 2024Geometry-Informed Neural Networkscited by 12
- 2021A Parametric Level Set Method for Topology Optimization based on Deep Neural Network (DNN)cited by 11
- 2024Neural topology optimization: the good, the bad, and the uglycited by 10
- 2020Computing multiple solutions of topology optimization problemscited by 7
- 2021Topology Optimization Methods for 3D Structural Problems: A Comparative Studycited by 5
- 2023Neural Implicit Shape Editing using Boundary Sensitivitycited by 4
Built on by
Citing papers per year: 2026: 1.
1 paper.