Well, you can chose to try all models.
Then relax will do a model selection, due to AIC criterion.
Relax will make model.out file in the "final" directory, and perform monte carlo error analysis.
Then one can then consider doing a global fitting/clustering analysis for some residues, choosing one of the models.
Reading in the old results directory from the previous run. That will average the values for the global parameter kex.
This will probably modify the R2 values, since they are now fitted under restriction of the global parameters.
If have not yet completed a tutorial for setting up R1rho.
But you can see the system test in:
test_suite/system_tests/relax_disp.py under def test_r1rho_kjaergaard(self):
Best
Troels