-
Is there a reason that Optimization.LevenbergMarquardtMinimizer.Minimum is not re-entrant? It stores things like bounds and scaling in static variables so that if you try to run the routine from multiple threads simultaneously you get spurious errors (e.g., I've got two different fits running in two different threads and the fit model for one has less parameters than the fit model for the other which leads to "argument out of bounds" errors for the fit with more parameters if the other fits bounds end up in the static variables). I might try and fix it with judicious use of ThreadLocal, but wanted to check see if there's some deeper reason for things to be the way they are. |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
Nevermind, I was using version 4.15.0. This appears to have been fixed in 5.0.0. |
Beta Was this translation helpful? Give feedback.
Nevermind, I was using version 4.15.0. This appears to have been fixed in 5.0.0.