[gmx-users] Freeze + NPT + constraints

[Justin Lemkul]

On 9/10/13 5:17 PM, HANNIBAL LECTER wrote:
> I would think that too. However, that is not what happens. If you freeze a
> group it gets scaled according to the box vectors during a npt simulation.
>

Are the positions of the atoms actually changing?  If they are, that's a bug. 
It shouldn't happen, and in previous version of Gromacs, I know it didn't.  Can 
you point to the code that is making this happen?

> Hence in order to keep the group (in my case a carbon nano tube) of fixed
> diameter i use position restraints. This protocol works fine with with npt
> simulations.
>
> Now, i wanted to check if i include the bonded interactions in my topology
> and constrain all the bonds things should not matter at all, since the
> bonded interactions are redundant in this case. However, that is not the
> case and the system crashes. If however the bonds are not constrained,
> things work fine.
>
> I know this is a very round about way of getting things done, still i am
> curious.

Constraints and barostats both try to modify coordinates.  The freezegrps 
mechanism prevents modifications of coordinates, so what you're trying to 
satisfy is a barostat that wants to scale coordinates (you say that is 
accomplished somehow; I would like to see the evidence so the issue can be 
addressed) and a constraint algorithm that wants to satisfy bond lengths.  One 
of those is bound to fail.  I can imagine that if the initial configuration 
perfectly satisfied the constraints, then maybe the run could proceed (because 
freezegrps and constraints wouldn't be fighting one another), but I still can't 
get my head around any way that you could satisfy freezing and a barostat.

In reality, what we should do is make it a fatal error, because truly, frozen 
groups should be used very sparingly and certainly never with NPT.  It just 
doesn't make sense.  I understand what you're trying to accomplish, but the 
underlying theory becomes flawed.

-Justin

> On Sep 10, 2013 4:12 PM, "Justin Lemkul" <jalemkul at vt.edu> wrote:
>
>>
>>
>> On 9/10/13 2:59 PM, HANNIBAL LECTER wrote:
>>
>>> Hi,
>>>
>>> Sorry to bother you again regarding this. But I am not sure, as to why
>>> freeze + constraints should not work during a NPT simulation?
>>>
>>> The update algorithm as shown in the manual does not say anything that
>>> would fundamentally prevent this. In fact, one can use freezegroups +
>>> constraints in NVT. What prevents it from running in NPT?
>>>
>>>
>> The use of a barostat requires that the positions of the atoms can be
>> scaled according to the pressure.  Freezing atoms, by definition, makes
>> them immobile and thus they cannot be scaled.  Thus, the two algorithms are
>> fundamentally incompatible.
>>
>> -Justin
>>
>>   I get the foll. error.
>>>
>>> step 0
>>> Step 1  Warning: pressure scaling more than 1%, mu: -2.7758e+23
>>> -2.7758e+23
>>> -2.7758e+23
>>>
>>> Step 1  Warning: pressure scaling more than 1%, mu: -2.7758e+23
>>> -2.7758e+23
>>> -2.7758e+23
>>>
>>> Step 1  Warning: pressure scaling more than 1%, mu: -2.7758e+23
>>> -2.7758e+23
>>> -2.7758e+23
>>>
>>> Step 1  Warning: pressure scaling more than 1%, mu: -2.7758e+23
>>> -2.7758e+23
>>> -2.7758e+23
>>>
>>> Step 1  Warning: pressure scaling more than 1%, mu: -2.7758e+23
>>> -2.7758e+23
>>> -2.7758e+23
>>>
>>> Step 1  Warning: pressure scaling more than 1%, mu: -2.7758e+23
>>> -2.7758e+23
>>> -2.7758e+23
>>>
>>> Step 1  Warning: pressure scaling more than 1%, mu: -2.7758e+23
>>> -2.7758e+23
>>> -2.7758e+23
>>>
>>> Step 1  Warning: pressure scaling more than 1%, mu: -2.7758e+23
>>> -2.7758e+23
>>> -2.7758e+23
>>> Warning: Triclinic box is too skewed.
>>> Warning: Triclinic box is too skewed.
>>> Warning: Triclinic box is too skewed.
>>> ...
>>> ...
>>> ...
>>> Segmentation fault
>>>
>>>
>> --
>> ==============================**====================
>>
>> Justin A. Lemkul, Ph.D.
>> Postdoctoral Fellow
>>
>> Department of Pharmaceutical Sciences
>> School of Pharmacy
>> Health Sciences Facility II, Room 601
>> University of Maryland, Baltimore
>> 20 Penn St.
>> Baltimore, MD 21201
>>
>> jalemkul at outerbanks.umaryland.**edu <jalemkul at outerbanks.umaryland.edu> | (410)
>> 706-7441
>>
>> ==============================**====================
>> --
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://lists.gromacs.org/**mailman/listinfo/gmx-users<http://lists.gromacs.org/mailman/listinfo/gmx-users>
>> * Please search the archive at http://www.gromacs.org/**
>> Support/Mailing_Lists/Search<http://www.gromacs.org/Support/Mailing_Lists/Search>before posting!
>> * Please don't post (un)subscribe requests to the list. Use the www
>> interface or send it to gmx-users-request at gromacs.org.
>> * Can't post? Read http://www.gromacs.org/**Support/Mailing_Lists<http://www.gromacs.org/Support/Mailing_Lists>
>>

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

Justin A. Lemkul, Ph.D.
Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441

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