rezaakb / pinns-torch Goto Github PK

Why are the boundary conditions such as non-slip and inflow conditions not included in the loss in the example (aneurysm3D). I tried doing 3-D test calculation and cannot get a solution (loss doesn't decrease enough), probably because it does not consider the loss of the boundary conditions.

And I couldn't understand why epoch is set only 1 step in aneurysm3D example. Is it enough to predict three dimensional flow field ?

I installed your packages as described and tried to run

python examples/navier_stokes/train.py

but i get an error;

`
Traceback (most recent call last):
File "[..]pinns-torch/examples/navier_stokes/train.py", line 8, in
import pinnstorch
File [..]local/lib/python3.10/site-packages/pinnstorch/init.py", line 1, in
from pinnstorch.train import train
File "[..].local/lib/python3.10/site-packages/pinnstorch/train.py", line 15, in
rootutils.setup_root(file, indicator=".project-root", pythonpath=True)
File "[..].local/lib/python3.10/site-packages/rootutils/rootutils.py", line 148, in setup_root
path = find_root(search_from, indicator)
File "[..].local/lib/python3.10/site-packages/rootutils/rootutils.py", line 70, in find_root
raise FileNotFoundError(f"Project root directory not found. Indicators: {indicator}")
FileNotFoundError: Project root directory not found. Indicators: ['.project-root']

`
Hope you can help...

I just directly cloned this repository, had virtual env setup and ran tutorials/0-Schrodinger.ipynb, but the resulted prediction did not reach convergence. Is there any additional setting should be done or any hyperparameters?

I ran the Navier-Stokes example for solving the NS equation using PINN, but I'm unsure about the meaning of 'idx_t' in the 'pred_dataset' section. If it specifies time, can I obtain a time series of prediction results?