[gmx-users] Time step in umbrella sampling

[rose rahmani]

Hi;

i'm doing umbrella sampling.i did pulling of AA from surface  and then
different conformation with trjconv and then umbrella sampling.....i
thought that 0.9 nm was the closest distance from surface because when i
changed the pullratejust for 0.001 nm/ns the run crashed, so ithought it
couldn't be closer more.so i run md for umbrella then gmx wham.
but when i check and draw the profile.xvg file, energy had  min value in
0.6 nm! and was repulsive in closer distances. i didn't sample any closer
distance from 0.9 nm and didn't md_umbrella run for it because i didn't see
closer distances in summary distances.datfile(token fro gmx distance).How
could be possible?i didn't know it before.
Then i modified the dt to half of its former value and increased nsteps
2times so there is not any change in simulation time,BUT at this time i got
the closer distances(in comparison with 0.9 nm), 0.5 nm!
HOW can i be sure that dt=X is the best value and my AA will not be closer
more even i change the simulation time or pullrate or anything else?
Before these happenings i thought that when the run doesn't crush
everything is OK, but now i don't have any idea!
could you please help me?

Thank you so much