You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I have a bug with the function predict_variances. It seems like variances are rounded to 0 when they are too small.
A code producing the bug is attached. A simple function with mixed variables is used, a DoE is performed and surrogate models are built. 10 points' variances are predicted and displayed. The bug appears when the DoE size is big enough (20 points are sufficient) and disappears when the size is small (5 points for instance). The bug appears from the version 2.5.1 (it wasn't here in 2.5.0).
I have solved it by commenting the line 1859 from the file smt/smt/surrogate_models/krg_based.py
Hello,
I have a bug with the function predict_variances. It seems like variances are rounded to 0 when they are too small.
A code producing the bug is attached. A simple function with mixed variables is used, a DoE is performed and surrogate models are built. 10 points' variances are predicted and displayed. The bug appears when the DoE size is big enough (20 points are sufficient) and disappears when the size is small (5 points for instance). The bug appears from the version 2.5.1 (it wasn't here in 2.5.0).
I have solved it by commenting the line 1859 from the file smt/smt/surrogate_models/krg_based.py
Thanks !
bug_version_SMT.zip
The text was updated successfully, but these errors were encountered: