During a refinement, I encountered a failed refinement result where some variables (but not all) refined to zero with an error of -2147483648 (yes, a negative number!) which is the smallest you can get for a 32 bit signed integer. I think, there is somewhere an unchecked overflow or a NAN-like error. Or do the error calculations come from the LevMar library?
Update:
This is apparently happening when RefinableVariables are defined but not used (I forgot to add the correlation using these variables to that specific refinement).
During a refinement, I encountered a failed refinement result where some variables (but not all) refined to zero with an error of -2147483648 (yes, a negative number!) which is the smallest you can get for a 32 bit signed integer. I think, there is somewhere an unchecked overflow or a NAN-like error. Or do the error calculations come from the LevMar library?
Update: This is apparently happening when RefinableVariables are defined but not used (I forgot to add the correlation using these variables to that specific refinement).
Full output: