search

ehsankharazmi / hp-VPINNs

hp-VPINNs: variational physics-informed neural network with domain decomposition is a general framework to solve differential equations
MIT License
69 stars 33 forks source link

About the One-dimensional Poisson’s equation with boundary layer solution #7

Open HyyydeXD opened 4 months ago

HyyydeXD commented 4 months ago

Hello Ehsan, I changed the u_exit() and the f_exit() in the hp-vpinn-poission-1d.py with the same equation in the paper as:

    def u_ext(x):
        utemp = 0.1 * np.sin(5 * np.pi * x) + np.e**(-99 - 100 * x)
        return amp * utemp
    def f_ext(x):
        gtemp = -(0.5 * np.cos(5 * np.pi * x) - 100 * np.e**(-99 - 100 * x))
        return -amp * gtemp

However, I cant get the same solution, instead I get a completely wrong solution. BTW I didn't change anywhere else but this two functions. And I noticed that the output of the lossv is weird. It seems too small. Maybe the wrong loss is to be blamed. Could you please tell me where I went wrong?

Best Hyde

HyyydeXD commented 4 months ago

And would you please upload the codes about solving the Bugers' Equation? That would be of great help.