adamantine
adamantine copied to clipboard
Semi-deterministic failure, likely involving radiation BCs
@AshGannon has seen semi-deterministic errors in adamantine simulations of the IMTS gear. My understanding (correct me if I have it wrong) is that the failures occur around the same point in the simulation, but not at exactly the same point. (@AshGannon, is this true with identical meshes, or is it just that the failure time shifts slightly with different meshes.) The error manifests as NAN in the solution. It seems that this error doesn't happen if the clang compiler is used and the clang sanitizer doesn't show any issues. It might also only appear in debug mode.
We believe that the source of the error is in the radiation BCs, but haven't localized it further. Anecdotally, I thought I had issues with radiation BCs a while ago, although several bugs have been fixed since then.
@AshGannon, is there additional context that's worth listing?