Dear all,
I am simulating a protein in solution, GROMACS 5.1, Verlet cut-off scheme.
I realised I put wrong charge groups in the topology file. I have read that with Verlet they are not considered but as a safety check I recompiled a tpr with the correct topology and run two simulation, from the same cpt file of an old one: one simulation continuing with the wrong set up, one with the correct.
The output is not identical (I run them on the same computer), but I am wondering if it is because of the charge groups or because with the new tpr I am introducing new sources of randomness. Can you help me? Or suggest a check to understand whether the charge groups are influencing the dynamics or not?
[in the tpr, original and new, gen_vel=no, continuation=yes, ld_seed=10 (same value)]
Many thanks
The Verlet scheme ignores charge groups so the use of the "wrong" topology has never mattered. Your use of ld_seed implies the use of Langevin dynamics, which inherently introduces randomness. No two trajectories are likely to be identical, but their ensemble properties should agree.
Many thanks.
Just to be clear (but I think we meant the same), with charge groups I meant the number I give to charge groups in the itp files (6th column), the charges themselves are arranged correctly.