I am still new to this alchemical transformation. I have repeated same calculation 3 times for state A becoming state B. The results were DG=26 (4.2) kJ/mol, -2 (2.2) kJ/mol, and -4 (2.3) kJ/mol for the 1st run, 2nd run and 3rd run, respectively. All of these are generated from same files when I add ions. Is there some sign (i.e. energy fluctuation in NPT for either state A or stateB ) that I should justify which of these are not a good results? Or these errors looks normal and I should take the average value of all results? These are resulted from non equilibrium methods (80 frames x 50ps).
it could be that you have poor overlap between forward and reverse work distributions? If that’s the case, you may try to increase the length of the transitions (e.g. double the number of steps, half the delta-lambda).
Alternatively, it may also be that a drift is happening in one of the equilibrium runs, resulting in earlier snapshots pruducing significantly different work values than later snapshots from that equilibrium simulation. If that’s the case, it would be worth investigating the cause of that drift and potentially repeat that run (equilibrium simulation and transitions spawned from it).
I will look into those possibilities. Thank you very much for the insightful advice.