Shaikha Alkhadhr

An update on this issue: By running the previous changes I got these results. Any suggestions on how to improve the accuracy of this problem? I also tried running the...

``` Layer (type) Output Shape Param # ================================================================= dense (Dense) (None, 64) 256 dense_1 (Dense) (None, 64) 4160 dense_2 (Dense) (None, 64) 4160 dense_3 (Dense) (None, 64) 4160 dense_4 (Dense)...

> The 2D case is tougher - that one may require some creativity. Try adding in self-adaptive weighting and see if that coaxes your residual errors to decrease a little...

I just want to make sure Am I changing the IC weight using the correct lines: ``` def func_ic(x,y): return tdq.utils.constant(0) init = IC(Domain, [func_ic], var=[['x','y']]) upper_x = dirichletBC(Domain, val=0.0,...

> This looks correct to me. Modification of the learning rate can be done by overwriting the default optimizer with another Adam optimizer with a different lr. See here for...

I was trying to follow the example of self-adaptive training in https://tensordiffeq.io/model/sa-compiling-example/index.html, but I got this error: ``` TypeError Traceback (most recent call last) c:\Users\User\OneDrive - The Pennsylvania State University\Research-MacBook...

Hello @levimcclenny , Thanks for your suggestions. I tried increasing Adam epochs to 100000, but the results are still not good enough. Please have a look at my setup and...

I'll be interested to apply this in my code as well. Any suggestions? Thanks!

Yes, I increased Adam epochs to 40,000. The results got better only on the first slice of time. I was also trying to apply SA here, but when turning isAdaptive...

Also, it'll be helpful to know how can I change the source function? sin (2 pi x).....(1) to sin (2 pi t).....(2) Simply modifying the initial condition that contains the...