Categorization of congruent TPMS by geometric features using manifold learning and clustering

In this study, the geometric features of 14 congruent TPMS structures were evaluated at 9 different SVF. This dataset, further curated through manifold learning and clustering, was used to create four separate categories of TPMS structures, each of which represented by certain defining features. The main goal of this study is to decrease the computational cost associated with selecting TPMS structures for novel applications, as well as providing a good basis for the rational selection of TPMS types based on defining geometric features.

The work is published in open-access format, and can be viewed here. The code is also published in open-source format, and is available on my GitHub.

Recommended citation: Stallard, S., Adam, A., Yang, G., Bergman, T. L., & Li, X. (2026). Categorization of congruent TPMS by geometric features using manifold learning and clustering. Computational Materials Science, 268, 114667. https://doi.org/10.1016/j.commatsci.2026.114667
Download Paper