Handle the numerical instability of MPM when generate data

[hoaitrungchar]

Hi, thanks for the great work.
I completely agree with your response about the numerical instability of MPM in the rebuttal phase. However, how did you solve this instability when generate synthetic data. Do you need to filter out failed simulations manually?
Thanks

[cwchenwang]

Thanks for your question. We did filter failed simulations by checking NaN values and out of boundary samples.