[gmx-users] seeming paradox with gmx wham
nedomacho at gmail.com
Mon Mar 12 20:42:36 CET 2018
I actually understood your tutorial perfectly well. What I didn't expect
is such a significant dependence on direction (assuming spherical
symmetry clears the situation).
I also had to use direction-periodic instead of distance in your
tutorial, because the box is 4 nm wide and grompp is screaming about
distance exceeding half-box size. I used 5.0.3 before and it was fine
with the 'distance' directive. I think 'direction-periodic' is also
wrong for spherical symmetry. So, stick with distance and increase box size?
On 3/12/2018 1:35 PM, Justin Lemkul wrote:
> On 3/12/18 3:28 PM, Alex wrote:
>> Just to make sure I'm doing this right this time. By merely changing
>> the dim directive with everything else following your tutorial, I
>> would be probing the spherical domain around the pore with a bunch of
>> different radius values? Because this is exactly what I want.
> Yes, you're applying a 1-D bias when you actually want a 3-D bias.
> People often misuse the tutorial .mdp files, though I will have a new
> version of the tutorial out this summer (along with a paper) that
> should clear all this up.
>> On 3/12/2018 8:43 AM, Justin Lemkul wrote:
>>> On 3/11/18 8:05 PM, Alex wrote:
>>>> Just to add to my question... The pull code for the umbrella
>>>> sampling from each of the N configs, as used in Justin's tutorial, is
>>>> pull_coord1_type = umbrella
>>>> pull_coord1_geometry = distance
>>>> pull_coord1_dim = N N Y
>>>> So, in each of the generated pullf and pullx files we sample a flat
>>>> slice at a given Z, and the reported forces and displacements are
>>>> along Z. What appears to be the reaction coordinate in our case is
>>>> the radius from the pore mouth, and one has to sample a set of
>>>> hemispheres from a series of radius values. Is this at all a
>>> Yes, but not by following the tutorial. You'd need to use
>>> pull_coord1_dim = Y Y Y to restrict the ions to a given radius
>>> around the pore itself. As it is, you're probably underestimating
>>> entropic contributions to the free energy.
More information about the gromacs.org_gmx-users