Brief description for the two examples of error simulations on this page: http://www.phy.anl.gov/atlas/TRACK/Trackv39/Examples/Windows/Errors/ The first is "Random-Errors" with randomly generated errors inside TRACK, generating a file of random numbers "vrand_nnn.dat" with the actual random numbers used for each element. The "align" line in "sclinac.dat" sets the errors amplitudes (max values if Uniform, rms if Gaussian). In "track.dat" I set iflag_corr=-1, so errors are included but no corrections. If you want to generate more seeds, NRZ is the number of seeds to run and NRZ_min is the starting seed so you can skip the seeds that you have already used. every seed will have its own output files with their number "_nnn.out" for seed nnn (001 for 1). The second is "Assigned-Errors" with random numbers assigned by the user in the "vrand_nnn.dat" files. In this example I set all numbers to 1 so I simulate the case with max errors. For this I had to switch the flag "iflag_vrand" in "track.dat" to -2 to tell the code to read the "vrand" files instead of generating new ones. I think this is useful only for the extreme cases because filling 100 "vrand" files manually is not practical. This is also useful if you want to repeat a given run for which you have the "vrand" files and you want to use the exact same random numbers to test reproducibility of the result for example.