Fig. 9: Evaluation in unrolled prediction scenarios for a/c tip-induced domain switching. | npj Computational Materials

Fig. 9: Evaluation in unrolled prediction scenarios for a/c tip-induced domain switching.

From: Machine learning surrogate for 3D phase-field modeling of ferroelectric tip-induced electrical switching

Fig. 9

a Evolution of the \({{\mathcal{L}}}_{2}\) error and b MARE during the unrolled prediction for the \({{\mathcal{P}}}_{x}\), \({{\mathcal{P}}}_{y}\) and \({{\mathcal{P}}}_{z}\) polarization components. For each starting frame, complete domain switching trajectories are performed until the final frame. Results are averaged over 200 test dataset simulations, with the solid line indicating the mean and the shaded region representing the interquartile range.

Back to article page