Skip to main content
. 2022 Feb 16;8(7):eabk0644. doi: 10.1126/sciadv.abk0644

Fig. 5. Inference of the deformed patterns compared with FEM ground truth for all cases shown in Fig. 2.

Fig. 5.

We displayed the visual outlines of deformed configurations of FEM/Abaqus (blue) and PINN results (red for matrix; green for inclusion in case 5). Three snapshots are shown for each case after different numbers of training iterations, which (from left to right) correspond to the completion of pretraining (beginning of parameter estimation), amid the training, and the completion of training, respectively (k = 103 and M = 106).