PDEBench icon indicating copy to clipboard operation
PDEBench copied to clipboard

Config files for incompressible Navier-Stokes equations

Open ZhangHuixin1103 opened this issue 2 years ago • 4 comments

Howdy!

Thanks for the great work you've done.

I have trouble in model training now, and I'm seeking for your help.

I wanna use FNO model to train and evaluate on the dataset ns_sim_2d-i.h5.

However, I thought the command to run the forward model training can be found in models/run_forward_1D.sh, but there's not.

Neither did I find any config_ns_incomp.yaml file for the model training input in models/config/args/. I saw config_diff-react.yaml, config_diff-sorp.yaml and config_rdb.yaml, though.

Your reply will be appreciated.

ZhangHuixin1103 avatar Nov 14 '22 13:11 ZhangHuixin1103

Thank you very much for your comment. In our paper, 2D incompressible NS was used only for inverse problems, so for the moment no config file for forward problem. However, 2D incompressible NS is very popular for AI4Sci community, so that should be on our ToDo list. What do you think @timothypraditia @leiterrl @danmackinlay ?

mtakamoto-D avatar Nov 17 '22 11:11 mtakamoto-D

Howdy!

Thanks for your comment.

I double checked but still didn't find any config file for inverse problem. Training process of 2D compressible NS went on well, though, with config_2DCFD.yaml.

ZhangHuixin1103 avatar Nov 21 '22 15:11 ZhangHuixin1103

@falesiani @danmackinlay Can I ask if we did not use incompressible NS equations for inverse problem??

mtakamoto-D avatar Nov 25 '22 17:11 mtakamoto-D

Howdy!

Thanks for your comment.

I double checked but still didn't find any config file for inverse problem. Training process of 2D compressible NS went on well, though, with config_2DCFD.yaml.

how can i tran 2D compressible NS with config_2DCFD.yaml ?

WH0226 avatar Dec 13 '23 07:12 WH0226