Physics-informed neural networks for inverse problems in supersonic flows

•We solved the inverse problems in supersonic flows using PINNs, and extended PINNs (XPINNs) methods.•Such problems are notoriously difficult or even impossible to solve using traditional numerical methods.•The only data we used is on the inflow and wall boundary, as well as the density gradients from Schileren methodology.•We compared the outcomes of PINNs and XPINNs on test cases involving shock as well as expansion waves.