maziarraissi / hfm Goto Github PK

Dear author, I am very interested in your article. Do you have the original simulation data? Not processed by MATLAB.
Thank you very much!

Hello,

I have some questions about the implementations. Please see the questions below:

1. For the Aneurysm 3D example I see that the structure is fixed wall, the walls might move in reality. I was wondering if you've tried for moving structure. Or should I ask does the proposed method work for moving structure (number of spatial points are different at each time instant)?
2. I was thinking that the maximum concentration field is 1 and does it work for lower concentration field data? Such as c_max ~=0.1
3. Can we also apply initial condition just like applying boundary conditions in the example?

Hello，

I ran the code of the Aneurysm3D.py using the Aneurysm3D.mat data downloaded from the google drive provided in README.md, and I didn't modified anything. The version of tensorflow i used is 1.14. The loss was descending at first, but after a while it became 'nan'. I don't know what is wrong.

Thanks.

1. What is the version of tensorflow you use?
2. When iteration reached to 10000, the loss value just converge to 2e-3 and relative L2 errors just converge to the value between 1~2, which is great different with the performance in your paper.
3. What is the specific parameters set in order to get your results in cylinder2D example, such like learning rate, epochs, batchsize etc?

hello professor 。I run the code
PYcharm prompt ：[Errno ] No such file or directory: '../Data/Aneurysm3D.mat'
where i can find this file