[gmx-users] pbc atom

[Gavin Melaugh]

O.K thanks Justine I'll look at these files and see if there is any
weirdness.
Just to make sure that there are no cut-off artefacts. If I  set the
cut-offs to zero as in the following mdp file; that does mean that I am
considering all electrostatic and LJ interactions between the molecules?
and not including them at all. I assume the former given that there are
both electrostatic and LJ energy values in the log and energy files.

Gavin

title       = Pull test
cpp         =
include     =
define      =
integrator  = md
nsteps      = 50000000
dt          = 0.002
nstxout     = 250000
nstvout     = 250000
nstlog      = 250000
nstenergy   = 5000
nstfout     = 250000
pbc         = no
nstlist     = 10
ns_type     = simple
vdwtype     = cut-off
rlist       = 0
rvdw_switch = 0
rvdw        = 0
coulombtype = cut-off
rcoulomb    = 0
tcoupl      = nose-hoover
tc_grps     = system
tau_t       = 0.1
ref_t       = 600
gen_vel     = no
gen_temp    =
constraints = none
comm_mode   = angular
pull        = umbrella
pull_geometry = distance
pull_dim = Y Y Y
pull_start = no
pull_ngroups = 1
pull_group0 = cage_1
pull_group1 = cage_2
pull_init1 = 2.59
pull_rate1 = 0.0
pull_k1 = 1000
pull_nstxout = 1000
pull_nstfout = 1000


Justin A. Lemkul wrote:
>
>
> Gavin Melaugh wrote:
>> For the umbrella sampling I am not pulling in any direction I am just
>> applying the umbrella potential so that two molecules can sample space
>> about a give value of r0 (COM distance). Maybe I am confused but should
>> I not plot the absolute value of the displacement (modulus) as supposed
>> to the displacement along a given vector component?
>>
>
> For diagnosing weird behavior, I would always start with the primary
> data, not any post-processed interpretation of it.  Use the pullx.xvg
> file.  Any weird changes in sign or sudden changes in magnitude would
> indicate problems.  If you start manipulating the data, you may be
> hiding the real reason for the problem.
>
> -Justin
>
>> Gavin
>>
>> Justin A. Lemkul wrote:
>>>
>>> Gavin Melaugh wrote:
>>>> I always used g_dist to plot the COM distances because the pullx.xvg
>>>> files doesn't give this value directly. Can you access the COM
>>>> distance
>>>> from the pullx.xvg file
>>>>
>>> That's what is in pullx.xvg - coordinate (X,Y,Z) of the reference
>>> group, then displacement (dX,dY,dZ) of the pull group relative to the
>>> reference.  Not all terms may be present, depending on the axis along
>>> which you're pulling, i.e. if you only pull along Z, there will be two
>>> terms in pullx.xvg - Z and dZ.
>>>
>>> -Justin
>>>
>>>> Gavin
>>>>
>>>> Justin A. Lemkul wrote:
>>>>> Gavin Melaugh wrote:
>>>>>> Justin
>>>>>>
>>>>>> I have looked at the movies but it's very hard to tell what's
>>>>>> going on
>>>>>> as I save teh trajectories every 250000 steps for every 50000000
>>>>>> step
>>>>>> simulation (100 ns). Ill look at them  again in more detail and post
>>>>>> back.
>>>>>>
>>>>> Plotting the pullx.xvg file(s) may be useful, too, to indicate where
>>>>> any weird jumps or sudden changes in position might occur.  Mapping
>>>>> that information back onto the trajectory could help focus your
>>>>> attention.
>>>>>
>>>>> -Justin
>>>>>
>>>>>> Gavin
>>>>>>
>>>>>> Justin A. Lemkul wrote:
>>>>>>> Gavin Melaugh wrote:
>>>>>>>> O.K thanks anyway
>>>>>>>>
>>>>>>> I saw the plots you posted during a conversation with Chris, but
>>>>>>> I'll
>>>>>>> ask the obvious anyway: have you watched the trajectories for
>>>>>>> any of
>>>>>>> the problematic windows?  It didn't seem like you had two
>>>>>>> metastable
>>>>>>> states, but maybe having a look at the movie would shed some
>>>>>>> light on
>>>>>>> what's going on.  I know that's the first thing I'd do.
>>>>>>>
>>>>>>> -Justin
>>>>>>>
>>>>>>>> Justin A. Lemkul wrote:
>>>>>>>>> Gavin Melaugh wrote:
>>>>>>>>>> Thanks Justin
>>>>>>>>>>
>>>>>>>>>> Have you any idea why when generating umbrella histograms for
>>>>>>>>>> the
>>>>>>>>>> pmf I
>>>>>>>>>> would get two peaks in the histograms above a distance of 2 nm,
>>>>>>>>>> but
>>>>>>>>>> below 2 nm I get well behaved histograms that lead to a very
>>>>>>>>>> good profile in the pmf. To the best of my knowledge the
>>>>>>>>>> configurations
>>>>>>>>>> are all very well equilibrated at their respective COM
>>>>>>>>>> distances.
>>>>>>>>>> Umbrella sampling is performed on all windows using a force
>>>>>>>>>> constant of
>>>>>>>>>> 1000 kj/mol at 600 K.
>>>>>>>>>>
>>>>>>>>> Sorry, no clue.
>>>>>>>>>
>>>>>>>>> -Justin
>>>>>>>>>
>>>>>>>>>> Cheers
>>>>>>>>>>
>>>>>>>>>> Gavin
>>>>>>>>>>
>>>>>>>>>> Justin A. Lemkul wrote:
>>>>>>>>>>> Justin A. Lemkul wrote:
>>>>>>>>>>>> Gavin Melaugh wrote:
>>>>>>>>>>>>> Hi all
>>>>>>>>>>>>>
>>>>>>>>>>>>> I am generating a series of configurations using the pull
>>>>>>>>>>>>> code to
>>>>>>>>>>>>> calculate the pmf. I am using no pbc i.e. pbc =no, howver the
>>>>>>>>>>>>> output
>>>>>>>>>>>>> from grommp gives me info on pbc atom.
>>>>>>>>>>>>>
>>>>>>>>>>>>> Pull group  natoms  pbc atom  distance at start   
>>>>>>>>>>>>> reference at
>>>>>>>>>>>>> t=0
>>>>>>>>>>>>>        0        72        36
>>>>>>>>>>>>>        1        72       360   4.290                 4.290
>>>>>>>>>>>>>
>>>>>>>>>>>>> why is this so?
>>>>>>>>>>>>>
>>>>>>>>>>>> grompp always assigns the numerical middle atom of a group as
>>>>>>>>>>>> the
>>>>>>>>>>>> PBC 
>>>>>>>>>>> ...unless over-ridden by providing a different value for
>>>>>>>>>>> pbc_pullatom1.
>>>>>>>>>>>
>>>>>>>>>>> -Justin
>>>>>>>>>>>
>>>>>>>>>>>> reference point.  In the case of pbc=no, it shouldn't matter.
>>>>>>>>>>>>
>>>>>>>>>>>> http://manual.gromacs.org/current/online/mdp_opt.html#pull
>>>>>>>>>>>>
>>>>>>>>>>>> -Justin
>>>>>>>>>>>>
>>>>>>>>>>>>> Cheers
>>>>>>>>>>>>>
>>>>>>>>>>>>> Gavin
>>>>
>>
>>
>