Sungkyunkwan University · 工学
Professor Jae-Hyeok Shim's research lab specializes in multiscale materials modeling and generative 3D shape synthesis, focusing on the atomic-scale mechanisms of dislocation-precipitate interactions in metallic systems and the development of advanced deep generative models for 3D shape creation. The lab investigates martensitic transformations and dislocation dynamics in complex alloys, particularly in bcc Fe-Cu systems, using molecular dynamics simulations to uncover deformation mechanisms. Concurrently, the lab pioneers diffusion-based generative models that leverage signed distance fields (SDF) for high-fidelity 3D shape generation, addressing memory and resolution challenges through a two-stage diffusion framework. These interdisciplinary efforts bridge materials science and machine learning to enable next-generation design of functional materials and digital twins of complex geometries.
Figures are computed from collected data and may differ slightly.
Molecular dynamics simulations of the interaction between a screw dislocation and a coherent bcc Cu precipitate in bcc Fe indicate that the screw dislocation stress field assists a martensitic transformation into a close-packed structure for precipitate diameters larger than 1.8nm, resulting in a stronger obstacle to dislocation glide. The observed martensitic transformation mechanism agrees with the Nishiyama-Kajiwara [Jpn. J. Appl. Phys. 2, 478 (1963)] model. For coherent bcc Cu precipitates w
We propose a 3D shape generation framework (SDF-Diffusion in short) that uses denoising diffusion models with continuous 3D representation via signed distance fields (SDF). Unlike most existing methods that depend on discontinuous forms, such as point clouds, SDF-Diffusion generates high-resolution 3D shapes while alleviating memory issues by separating the generative process into two-stage: generation and super-resolution. In the first stage, a diffusion-based generative model generates a low-r
Open papers in the app to read, cite, and organize with AI.
