Issues about Testing phase in the RANS case

[zzzhengtong]

Hi Romit,

I have a minor question about the Testing phase in the RANS case

Your previous paper (https://doi.org/10.1016/j.compfluid.2020.104777) can realize "Deploy the previously trained framework to predict an approximation for the steady-state turbulent eddy viscosity and fix this quantity permanently."

I have read your code row by row but I am not sure how the quantity could be fixed permanently.

I guess this row in ML_SA_CG
"nut_ml_ = MyFilter_(nut_ml_)"

Could you kindly give me some clues?

[zzzhengtong]

I think the quantity could be fixed permanently by commenting out "turbulence->correct()" in simpleFoam_ML.c.

However, I still have a confuse about "nut_ml_ = MyFilter_(nut_ml_)".

[Romit-Maulik]

Thank you for your interest in this work!
The MyFilter operation performs a low-pass spatial filtering operation on the ML predictions - to enhance stability.