[gmx-users] Lost particles while sorting

[Alex]

So, I have a bit of an update -- hopefully this will somewhat help others
encountering this error, because it may not be entirely a bug...

1. The issue can be overcome by first and foremost not applying walls to a
system previously run without any -- that was my mistake and it makes
physical sense. Dynamic equilibration has to be performed with walls and
corresponding pbc setting. This, however, does not guarantee anything if
the system is e.g. solvated -- that is, if there are atoms expected very
close to walls, or beyond them. In addition, if there atoms looking for
neighbors across the box (through previously set pbc), that can be
catastrophic. See below.

2. After solvating, but before EM, reinsert the system in a box slightly
larger in the z-direction. For a box initially 4 nm long in Z in my case,
reinserting in a 4.6 nm long box worked here.

3. Particles seem to be lost when wall-r-linpot is set improperly, so try
various values (0.25 nm worked for me).

4. The entire concept here is naturally very messy. I could think of
optional functionality, which allows for removing entire molecules whose
atoms are anticipated to be beyond walls, or close enough to result in
local explosions. However, that is also messy and can be done manually in
previous steps, so no elegant suggestion from my end, unfortunately, other
than being very mindful when setting this up.

Alex


On Wed, Nov 1, 2017 at 6:26 PM, Alex <nedomacho at gmail.com> wrote:

> Mark,
>
> Could you please share your email address, so I can send you the tpr
> privately? I don't want this work to be public just yet, if that's okay.
>
> Thanks,
>
> Alex
>
>
> On 11/1/2017 4:42 PM, Mark Abraham wrote:
>
>> Hi,
>>
>> We have had several reports of "lost particles while sorting" but we would
>> like a .tpr on an issue on redmine.gromacs.org to be able to investigate,
>> please!
>>
>> Mark
>>
>> On Wed, Nov 1, 2017 at 10:19 PM Alex <nedomacho at gmail.com> wrote:
>>
>> Hi all,
>>>
>>> I am using walls in the cowboy mode, GMX v. 2016.3, system is a porous
>>> membrane (periodic in XY) with water and ions -- everything works fine
>>> with
>>> pbc=xyz and no walls. With walls, mdrun throws "Software inconsistency
>>> error: Lost particles while sorting."
>>>
>>> mdp exerpt below:
>>>
>>> pbc                 = xy
>>> nwall               = 2
>>> wall-type           = 12-6
>>> wall-r-linpot       = 0.0
>>> wall_atomtype       = opls_996 opls_996
>>> wall-ewald-zfac     = 3
>>> periodic_molecules  = yes
>>> ns_type             =  grid
>>> rlist               =  1.0
>>> coulombtype         =  pme
>>> ewald-geometry      =  3dc
>>> fourierspacing      =  0.135
>>> rcoulomb            =  1.0
>>> rvdw                =  1.0
>>> vdwtype             =  cut-off
>>> cutoff-scheme   = Verlet
>>>
>>> This uses a custom type "opls_996" properly defined in ffnonbonded of the
>>> FF (custom OPLSAA). Any suggestions? I can provide more info, if needed.
>>>
>>> Thank you,
>>>
>>> Alex
>>> --
>>> Gromacs Users mailing list
>>>
>>> * Please search the archive at
>>> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
>>> posting!
>>>
>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>>>
>>> * For (un)subscribe requests visit
>>> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
>>> send a mail to gmx-users-request at gromacs.org.
>>>
>>>
>