@team For stage2, we require Lx and Ly to be specified as the solution is different for different Ly (and maybe a small variation due to Lx - periodical boundary conditions).
Your supplied test voltage boundary condition does not match your scoring.
Also, just an observation on how this competition is run.
Multiple errors throughout the problem description and train/test dataset. This is the first time I am participating in a xeek competition and it is by far the worst experience compared to other platforms kaggle/drivendata/topcoder/aicrowd. It looks like you really do not care and do not value participants’ time and effort.
We apologize for the mistakes that you’ve encountered while trying to participate. This challenge is one of the most expansive we’ve done, and missteps were made while getting it out to users. We hope you can make it up to you with 2023’s challenges list, which will be more varied and robust.
@team PLEASE could you tell us what modified RSME you are using? In stage-3, the answer depends on the loss function. I am using softL1, L1, L2, and combinations. I am getting better out-of-fold but not better LB. Best result LB for me is with [ |true_phi - pred_phi|/(0.01 + abs(true_phi)) ]^2, but I am just guessing.