[gmx-users] Umbrella sampling - Pull distance is larger than 0.49 times box size error

[Shrinath Kumar]

Hi,

I'm doing umbrella sampling and running into the above error. I can't
really increase the box size as that would make the system too big nor can
I use pull-geometry = direction-periodic as I need to use NPT.

But I'm confused on why I get this error in the first place because my
periodic box is a cuboid and the pulling distance is less than 0.5*box size
in each dimension. So the pulling vector should be uniquely determined and
there should be no discontinuities.

To solve this I tried having 3 pull coordinates each acting on a separate
dimension. This solves the grompp error. But would this produce the same
results as having 1 pull coordinate acting on all 3 dimensions? Or is there
any better workaround for this?

Thanks,
Shrinath