Figure 3From: Operator compression with deep neural networksResults of Experiment 1: Unstructured multiscale coefficient sampled from \(\mathfrak{A}_{\text{ms}}\) (top left), \(\vert u_{h}(x) - \widehat{u}_{h}(x) \vert \) (top right), and comparison of \(u_{h}\) vs \(\widehat{u}_{h}\) along the cross sections \(x_{1} = 0.5\) (bottom left) and \(x_{2} = 0.5\) (bottom right)Back to article page