The results of ldc_2d_zeroEq.py is not the same as those in SimNet_v21.06_User_Guide

I tried to run the example ldc_2d_zeroEq.py without any modification, but the results I got were not the same as shown in SimNet_v21.06_User_Guide.

The overall loss goes down, but the momentum_imbalance takes a huge value compared to the user guide. As a result, the validation score takes a terrible value.



I made a graph of the prediction result of u, but it is not the prediction that appears in the user guide.
image

By the way, when I trained ldc_2d.py, it did not match the user guide, but it scored better than the user guide.


So, I don’t think the computational environment is the cause of the problem. But just to be sure, here’s how my environment is.

Tesla V100
NVRM version: NVIDIA UNIX x86_64 Kernel Module 470.82.01 Wed Oct 27 21:21:55 UTC 2021
GCC version: gcc version 7.5.0 (Ubuntu 7.5.0-3ubuntu1~18.04)
Modulus 21.06 container for linux