Explainable quantum neural networks for multi-material topology optimization
2026-07-01 • Computational Engineering, Finance, and Science
Computational Engineering, Finance, and Science
AI summaryⓘ
The authors created a quantum neural network called XQNN to help design structures by deciding where to place materials and how to shape the structure based on forces and boundaries. They use data from the design process, like strain energy and boundaries, and encode this into a quantum system to predict material layout. Even though trained on simple 2D examples, XQNN works on more complex and bigger 3D designs without extra training. The authors also showed that certain quantum measurements relate clearly to physical design features, making the model's decisions easier to understand.
quantum neural networktopology optimizationstrain energyboundary conditionsqubitmaterial assignmentSobel boundary descriptorload paths3D voxel meshsensitivity analysis
Authors
Dahyun Joo, Naruethep Sukulthanasorn, Kenjiro Terada, Do-Nyun Kim
Abstract
We propose an explainable quantum neural network for multi-material topology optimization, XQNN, that determines both load-carrying structural layout and material type assignment for given boundary/loading conditions. Intermediate solution histories are first converted into element-wise strain energy, sensitivity, density, and Sobel boundary descriptors. Then, they are encoded in a ten-qubit circuit and qubit-wise $Z$ observables are mapped onto material type labels. Trained only on two-dimensional topology optimization histories obtained with a fixed mesh resolution, XQNN can be generalized to handle out-of-distribution boundary/loading conditions, progressively refined high-resolution meshes, and voxel-wise three-dimensional problems without additional training. We find that it is important to preserve qubit-wise observables and add boundary information for improving the optimization accuracy, and certain observables have consistent links to load paths, material type regions, and interfaces, demonstrating their usability as auditable mechanics-facing variables.