Replies: 1 comment 12 replies
-
Hi @jacktang, you can start by playing with the svd_rank of the POD and compute the projection error, that is the error on the training dataset. Maybe you have to many modes and the RBF interpolation creates a lot of oscillations that add up to a huge error. If you want to use RBF you can also check out the documentation: there is a smoothing parameter you can pass. Otherwise try with GPR. |
Beta Was this translation helpful? Give feedback.
12 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hello There!
I've read tutorial 1 and 2, and try to apply the same ROM workflow on my own tests of cold and hot flow in mixing tee.
The space lists below:
and there are four paramters:
The outputs are velocity of x, velocity of y, and pressure. I generated 300 snapshots by uniform sampling and splitted 200 snapshots for trainning, 50 for validation, and 50 for testing.
Finally, the average error is 0.8
I don't figure out what's wrong with my experience. Could you please give me some hits on how to make it work? Thanks a lot!
Beta Was this translation helpful? Give feedback.
All reactions