Abstract
Combinatorial problems arising in puzzles, origami, and (meta)material design have rare sets of solutions, which define complex and sharply delineated boundaries in configuration space. These boundaries are difficult to capture with conventional statistical and numerical methods. Here we show that convolutional neural networks can learn to recognize these boundaries for combinatorial mechanical metamaterials, down to finest detail, despite using heavily undersampled training sets, and can successfully generalize. This suggests that the network infers the underlying combinatorial rules from the sparse training set, opening up new possibilities for complex design of (meta)materials.
| Original language | English |
|---|---|
| Article number | 198003 |
| Pages (from-to) | 1-7 |
| Number of pages | 7 |
| Journal | Physical Review Letters |
| Volume | 129 |
| Issue number | 19 |
| DOIs | |
| Publication status | Published - 4 Nov 2022 |
Bibliographical note
Funding Information:We thank David Dykstra, Marc Serra-Garcia, Jan-Willem van de Meent, Edan Lerner, and Tristan Bereau for discussions. This work was carried out on the Dutch National e-infrastructure with the support of SURF Cooperative. C. C. acknowledges funding from the European Research Council under Grant Agreement No. 852587.
Publisher Copyright:
© 2022 American Physical Society.
Funding
We thank David Dykstra, Marc Serra-Garcia, Jan-Willem van de Meent, Edan Lerner, and Tristan Bereau for discussions. This work was carried out on the Dutch National e-infrastructure with the support of SURF Cooperative. C. C. acknowledges funding from the European Research Council under Grant Agreement No. 852587.