Error Messages
'Variable not found', or 'Load dataset/model first' while you have already loaded a datafile/model.This message is most likely caused by the user having run another program with the Globe_Toolbox gui open, and the other program has cleared some variables, or overwritten some existing ones. Reselecting your datafile and model should solve these problems. If not, try restarting the global analysis program.
'Warning: A is rank deficient to within machine precision'
This message is a sign that the fit algorithm has trouble converging. Still, it could converge normally. The starting guess may be too far off the solution, too many parameters are being asked to be fitted, or two time constants are very close which causes the algorithm to oscillate around the solution (when the convergence criteria are met). Try with less components, or make sure that two time constants are not virtually identical (on your time base), and see if the message persists. Alternatively, try fixing time zero, OD offset and/or any other values that not really need to be fitted, i.e. can be estimated by eye. After having found a satisfactory fit, release the fixed parameters to see if they can be optimised. Again, the fit might end normally anyway.
'Warning: Cannot extrapolate'
Apparently, the fit does not converge if you see this apparently infinite list of error messages in the Command Window. The most common reason is that there is a problem with time zero. Fixing time zero while activating the constraints box usually solves the problem. Matlab tries to look for a solution outside the available time points, which the used algorithm (the built-in 'timeseries' command is used that can only interpolate) can't do, hence the error message. Try using a different algorithm (under 'Advanced' fitting options) that is capabable of extrapolating. If this still does not work, try to use the time shift parameter to subtract a fixed time from your time axis. See also Limitations
'The time vector must have monotonically increasing values such that no two times are the same'Basically the same problem as with the above extrapolate error. Solve this in similar fashion.
'Attempted to access U(:,#); index out of bounds because size(U)=[#-1,#-1]'
Too many SVD components are asked to be fitted. The dataset has fewer time points than SVD components. This is not an error of the program, but a mathematical impossibility. Therefore, lower the requested nr (#) of SVD components to be fitted to be not more than the number of time points the data set contains.See Also