koopman-laboratory / koopmanlab Goto Github PK

Great work! Congratulation!

I just curious why the self.time_marching() method is called twice for both the high-frequency and low-frequency Fourier features, in the "kno.py" line 148 and line 149, did I miss something?

out_ft[:, :, :self.modes_x, :self.modes_y] = self.time_marching(x_ft[:, :, :self.modes_x, :self.modes_y], self.koopman_matrix)
out_ft[:, :, -self.modes_x:, :self.modes_y] = self.time_marching(x_ft[:, :, -self.modes_x:, :self.modes_y], self.koopman_matrix)

Thank you!

Congratulations on your impressive work! I am eager to learn more about testing the Koopman-vit operator like KNO. Would it be possible for you to provide a demo similar to demo_ns.py, or suggest some other ideas for how to test it effectively? Many thanks.

Awesome work! And I just wonder if KNO can be used to identify controlled ODE that represents the kinematical equation of mobile robot, like unicycle model? Whether you will try to do this?

It seems that this network only works for the 2D or 1D images. If I want to solve a chemical ODE, does it work?

awesome working！I am trying to use it.But i wonder weather it can solve ODE questions and can you give the data of your demo.thanks a lot.