[gmx-users] Error at FEP lambda = 1; seg. fault. with "larger than the table limit" message

[Justin Lemkul]

On 4/13/17 9:12 PM, Eric Smoll wrote:
> Hello Mark,
>
> I did not answer one of your questions. You asked "what is the alchemical
> state at 1.0?" The transformation involves lengthening the hydrocarbon
> substituent of an organic molecule. At lambda=0.0, we have the short-chain
> state. At lambda=1.0, we have the long-chain state. Since lambda=1.0
> corresponds to a larger molecule, the overlap issue you describe is
> possible. However, I can visualize the initial state of the total system
> (atoms and dummies) to ensure that there is no problematic overlap and (as
> far as I can tell) there is no overlap).
>
> Another interesting note is that the problem atom-pair indices reported in
> the error are different if small changes in the topology and mdp-settings
> are made. This suggests that there is something fundamentally wrong with my
> topology. However, the simulations run fine for all other lambda values
> from lambda=0.0 to lambda=0.9 with the same topology/input-structure/mdp
> settings.
>

I wouldn't suggest running each lambda state from the same input structure; you 
would do better to run each sequentially so that the transformation is gradual 
and the solvent can relax around the transforming solvent as a function of 
lambda.  That will almost certainly alleviate any clashes that arise from an 
otherwise sudden change in the topology.

-Justin

> Best,
> Eric
>
> On Thu, Apr 13, 2017 at 6:04 PM, Eric Smoll <ericsmoll at gmail.com> wrote:
>
>> Good point. I only mutate one solute to a different solute. I do not have
>> separate coulomb and vdw stages. If I examine the initial frame all
>> atoms/dummies that undergo transformation are far from solvent molecule
>> throwing the error. since the error is thrown on the first step I just
>> don't see how it could be an overlap problem.
>>
>> I have soft-core coul and VDW settings on. I assume they are applied to
>> all atoms with B state parameters in the topology. how would I know if
>> something is not being soft-cored properly?
>>
>> Is there something else I can check?
>>
>>
>>
>>> On Apr 13, 2017, at 5:54 PM, Mark Abraham <mark.j.abraham at gmail.com>
>> wrote:
>>>
>>> Hi,
>>>
>>>> On Fri, Apr 14, 2017 at 1:49 AM Eric Smoll <ericsmoll at gmail.com> wrote:
>>>>
>>>> Hi Mark,
>>>>
>>>> Thanks for the rapid reply! I am using version 5.1.2. I can reproduce
>> the
>>>> error with 2016.1.
>>>
>>> OK, that is useful to know.
>>>
>>>
>>>> Note that "particles" (I assume atom indices counting from 1) 541 and
>> 554
>>>> do not correspond to the alchemical system, but to a dihedral or pair
>> in a
>>>> solvent molecule. So there should be no particle overlap issue.
>>>
>>>
>>> Why couldn't one of the solvent atoms have overlapped with an eliminated
>>> particle in the solute, and then gone flying? What is the alchemical
>> state
>>> at 1.0? Are there particles needing soft coring that perhaps aren't?
>>>
>>> Mark
>>>
>>> That being
>>>> said, I am using VDW and coulomb soft-core settings.
>>>>
>>>> Best,
>>>> Eric
>>>>
>>>>
>>>> On Thu, Apr 13, 2017 at 5:40 PM, Mark Abraham <mark.j.abraham at gmail.com
>>>
>>>> wrote:
>>>>
>>>>> Hi,
>>>>>
>>>>>> On Fri, Apr 14, 2017 at 1:30 AM Eric Smoll <ericsmoll at gmail.com>
>> wrote:
>>>>>>
>>>>>> Hello GROMACS users,
>>>>>>
>>>>>> I am running NVT alchemical free energy simulations (change in atom
>>>>> number,
>>>>>> bonding, etc.). These simulations appear to go well at all lambda
>>>> values
>>>>>> below 1.0. At lambda=1.0, the system returns an error before or during
>>>>> the
>>>>>> first step:
>>>>>>
>>>>>> WARNING: Listed nonbonded interaction between particles 541 and 554
>>>>>> at distance 183739120.000 which is larger than the table limit 5.500
>>>> nm.
>>>>>
>>>>> On face value, some force went enormous last step to produce a
>> ridiculous
>>>>> distance here. To what alchemical state does lambda = 1.0 correspond?
>> Can
>>>>> this be some kind of overlap of two particles that e.g. should have
>> been
>>>>> soft-cored?
>>>>>
>>>>> However the same system (input structure + topology + mdp-settings)
>> will
>>>>>> minimize without any issue at same lambda=1.0 setting.
>>>>>
>>>>>
>>>>> Doesn't really mean anything.
>>>>>
>>>>> What GROMACS version is this? Some more recent versions have some more
>>>>> checks for things having gone crazy before this error would trigger.
>> Can
>>>>> you reproduce this with 2016.3?
>>>>>
>>>>> Mark
>>>>>
>>>>>
>>>>>> I am not sure what
>>>>>> is going wrong.
>>>>>>
>>>>>> Best,
>>>>>> Eric
>>>>>> --
>>>>>> 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.
>>>>> --
>>>>> 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.
>>>> --
>>>> 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.
>>> --
>>> 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.
>>

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================