Figure 7
From: ML for fast assimilation of wall-pressure measurements from hypersonic flow over a cone
Left: training data obtained via DNS, with 68 inflows chosen with Latin Hypercube Sampling on the prior distribution of \(\varvec{c}\). Center: DeepONet architecture: branch and trunk sub-networks with 10 layers each and 60 neurons per layer. Right: data loss \(\mathcal {L}\) for the training data (blue) and the validation data (red).
