Figure 1
From: Data-driven discovery of Green’s functions with human-understandable deep learning
Schematic of our DL method for learning Green’s functions from input-output pairs. (A) The covariance kernel of the Gaussian process (GP), which is used to generate excitations. (B) The random excitations and the system’s response are recorded (C). (D) A loss function is minimized to train rational NNs (E). (F) The learned Green’s function and homogeneous solution are visualized by sampling the NNs.
