[gmx-users] Re: distance/direction PMF

[Gavin Melaugh]

Hi Justin

Many thanks for the reply I know technically I am doing two different
things. However due to my starting conifgurations in both cases I would
expect similar results.
Could you confirm the following points

1) In pull_geometry=distance, and pull_dim =YYY; is the absolute
distance between the COM of both groups in all directions considered ?

2) In pull_geometry = direction, with a specified vector; I assume that
it is the distance between the two groups along this vector that is
considered?

3) Is the specified vector taken to be the vector connecting the two
points? Therefore are the desired distances in the mdp files essentially
the magnitudes of this vector at each window.

Cheers

Gavin


Justin A. Lemkul wrote:
>
>
> Gavin Melaugh wrote:
>> Hi all
>>
>> Could someone please get back to me on this. I have ran two sets of
>> umbrella sampling simulations
>> 1- Using pull_geometry = distance, and pull_dim = Y Y Y
>> 2- Using pull_geometry = direction, with pull_vec = 0.462808 0.494125
>> 0.735968
>>  
>> In both cases I wish to calculate the PMF for taking a host out of a
>> guest. I specify the distance between the COM of the two groups in the
>> respective mdp files. The two curves I get are completely different
>> which has leads me to ask a couple of questions:
>>
>
> You're doing two different things here, so I'd suspect that you're
> going to get a different answer with different settings.
>
>> 1) Does the vector change as the dynamics evolves (i.e is it just
>> relative between the two points) or does it remain fixed in space?
>>
>
> It is relative.
>
>> 2) Using umbrella sampling in the second case does the relative position
>> of the pull group (guest) fluctuate about the distance along the
>> specified vector?
>>
>
> Yes.
>
>> Note- I couldn't run g_wham with pull_geometry = direction in my current
>> version of gromacs, so I ran it in 4.5.3 but I had to use the -if
>> pullf-files.xvg option.
>>
>
> We'll have to assume that your "current version" is something in the
> 4.0.x series, so this outcome is expected.
>
> http://www.gromacs.org/Documentation/How-tos/Potential_of_Mean_Force
>
>> Some help would be appreciated as I have to be sure of what the program
>> is actually doing.
>>
>
> There's always the code :)
>
> -Justin
>