Some questions about deepwave

[guoketing]

Dear Alan,

I'm very glad to get the so wonderful codes. I try to implement other PML methods some days ago, but the PML method in deepwave have good enough result. So I give up this idea. I have read most codes in the deepwave. I have same questions about it.
question 1,
Is that possible to extend to wave case from acoustic case? I mean using velocity-stress form of the elastodynamic equations to generate seismic data and then we can inverse vp,vs and ρ using deepwave.
question 2,
Is the deepwave using RNN? I have read the paper that you wrote at 2018 "Seismic Full-Waveform Inversion Using Deep Learning Tools and Techniques". In this paper, you believe conventional seismic FWI can be constructed as a RNN, but I can't found any codes in deepwave about RNN. So, is that true that the forward function of finite difference is the process of RNN?