Potential memory leak when running fitter

[joe smith]

It appears that if I run the database fitter on a Touchstone file with a large number of points (16001) that the METAS software will use up about 7G of memory.  The software slows down a fair amount as well.  The problem is that once it completes, the software does not appear to release the memory.   Running a small data set (200 points), the software continues to run very slow.   The only way to recover is to restart the METAS software. 

I also notice that the fitter can have problems converging.  I am finding that it seems to want data with minimal gain and phase discontinuities.   I am curious if you have written an application note to help aid users in solving convergence problems.  

I have been posting my results here if interested:

https://www.eevblog.com/forum/rf-microwave/calculating-correction-data-for-an-unknown-vna-calibration-kit/

[Michael Wollensack METAS]

Can you provide me those two data files (16001 data points and 200 data points).

The convergence problem with can be overcome when you unchecked to fit the offset delay value. The problem here is that if you have data only up to a few GHz and the standard has only small length then this can be modeled with the offset delay value or the polynom for the open or the short plane.