Download file (485 KB)
Download as zip (485 KB)

Collection

Citation

Description:

This tutorial aims to give an introduction of how to use a deep generative model, 3D conditional generative adversarial network (3D-CGAN). The 3D-CGAN can be used for the inverse design of 3D voxelized microstructures with target properties. The 3D-CGAN is trained with supervised learning using a labeled dataset. The dataset consists of a large number of geometries (3D arrays) and their corresponding properties (e.g., elastic moduli). After training, the 3D-CGAN can generate a batch of geometries using target properties at inputs. In our previous tutorial, we have demonstrated how to use CGAN for the inverse design of 2D microstructures. This work is based on our previous publication for the inverse design of 3D architected materials. We hope this tutorial can be useful for those who are interested in the inverse design problems of microstructures.

Rights:

• Creative Commons BY-NC Attribution-NonCommercial 4.0 International
  

Keyword: Deep learning; Generative adversarial network; Inverse design; 3D shape; Microstructure; Mechanical metamaterial

Date published: 2023-12-31

Publisher: Taylor & Francis

Journal:

• Science and Technology of Advanced Materials (ISSN: 14686996)

Funding:

• Japan Society for the Promotion of Science 22J11202 (Grant-in-Aid for JSPS Fellows DC2)

Manuscript type: Not a journal article

MDR DOI: https://doi.org/10.48505/nims.4230

First published URL: https://doi.org/10.1080/14686996.2022.2157682

Related item:

Other identifier(s):

Contact agent: Xiaoyang Zheng (Center for Basic Research on Materials, National Institute for Materials Science) ZHENG.Xiaoyang@nims.go.jp

Updated at: 2024-01-05 22:11:09 +0900

Published on MDR: 2023-09-14 13:30:06 +0900