Answers are grounded in this paper and cite their sources.
Data-driven generative design of lattice metamaterials often demands excessively large training datasets to capture complex three-dimensional geometries. This work introduces an exploratory hierarchical workflow that circumvents this limitation through a compact, cubic symmetry-constrained topological representation, enabling efficient design space exploration with minimal data and a lightweight machine learning surrogate. Using 213 additively manufactured sub-millimetre AlSi10Mg specimens, the approach identifies two novel lattice topologies: fractal face cubic and cross face cubic. These metallic architectures achieve a significant increase of up to 30% higher yield strength compared to simple cubic lattices, a widely used high-strength lattice benchmark (stretch-dominated), across multiple relative densities and a broad range of strut slenderness ratios, covering distinct deformation regimes. Notably, the emergent fractal face cubic topology reveals an unprescribed transition toward thin-plate-like architectures, where strut fusion promotes uniform load distribution and maximizes structural efficiency. This small-but-efficient-data framework provides a scalable path for discovering non-intuitive, high-performance architected materials within symmetry-constrained design spaces, with direct relevance to thin-walled structural applications.
No takes yet. Share an insight, caveat, or question.
Haberl et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: