You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Thanks for sharing the codes of the PDE-based PINN. However, I cannot understand why the derivatives need to multiply and divide the model size, like in the code for first order derivatives: du_y, dv_x = du_ya/b, dv_xb/a, and for second order derivatives: du_xx, du_yy, du_xy = du_xx/a, du_yy*a/b**2, du_xy/b. We're guessing that these are modifications for implicit derivatives. We sincerely invite your guidance, thank you!
The text was updated successfully, but these errors were encountered:
Thanks for your interest in this work. Those modifications are for the purpose of rescaling the input coordinates within [0, 1] or [-1, 1].
Suppose that the true coordinates x', y' are in the range [0, a] and [0, b], the actual input coordinates will be x=x'/a and y=y'/b within range [0, 1]. The same was done for the outputs u=u'/a and v=v'/b (this is actually not necessary).
Therefore, we will have the true derivative du'/dy' = du/dy*a/b.
Hope that helps!
Thanks for sharing the codes of the PDE-based PINN. However, I cannot understand why the derivatives need to multiply and divide the model size, like in the code for first order derivatives: du_y, dv_x = du_ya/b, dv_xb/a, and for second order derivatives: du_xx, du_yy, du_xy = du_xx/a, du_yy*a/b**2, du_xy/b. We're guessing that these are modifications for implicit derivatives. We sincerely invite your guidance, thank you!
The text was updated successfully, but these errors were encountered: