[gmx-users] umbrella sampling

Justin Lemkul jalemkul at vt.edu
Tue Jul 11 22:37:08 CEST 2017



On 7/11/17 1:38 PM, Ben Tam wrote:
> Hi Justin,
> 
> 
> I am trying to get an energy profile when water molecules go through the porous cage. I am restraint it to z direction because I would like to see the energy that require to jump from one cage to another. I set up the simulation so that the structure is repeated at z direction which this will give me more sample slice for starting configuration. Sorry I am still new to the umbrella sampling.
> 

OK, I see what you're doing now.  That should be fine, but without seeing the 
histograms it's impossible to know how well things are working.  If you're 
getting NaN somewhere, that means a gap in the sampling.

-Justin

> 
> Best regards,
> 
> 
> Ben
> 
> 
> 
> 
> ________________________________
> From: gromacs.org_gmx-users-bounces at maillist.sys.kth.se <gromacs.org_gmx-users-bounces at maillist.sys.kth.se> on behalf of Justin Lemkul <jalemkul at vt.edu>
> Sent: Tuesday, July 11, 2017 18:07
> To: gmx-users at gromacs.org
> Subject: Re: [gmx-users] umbrella sampling
> 
> 
> 
> On 7/11/17 12:53 PM, Ben Tam wrote:
>> Hi Andre,
>>
>>
>> Thanks for your answer. Actually I should clarify it is metal-organic-framework that I am working on which is a porous material. With the k value larger than 100, the histogram become multiple sharp single peak.
> Furthermore when I look at individual 0.1nm slides, it runs completely fine
> without obvious anomaly. Therefore I am really confused why the profile.xvg
> become infinite or "nan".
>>
> 
> I'm really confused about your setup.  From what it sounds like, your windows
> are too narrow to get good sampling, perhaps the force constant is too strong,
> and I'm really not sure why you're only applying a restraint along z.  What is
> the geometry of your system?  Don't follow my tutorial too closely, it's not
> appropriate for most cases...
> 
> -Justin
> 
>>
>> Best regards,
>>
>>
>> Ben
>>
>>
>> ________________________________
>> From: gromacs.org_gmx-users-bounces at maillist.sys.kth.se <gromacs.org_gmx-users-bounces at maillist.sys.kth.se> on behalf of André Farias de Moura <moura at ufscar.br>
>> Sent: Tuesday, July 11, 2017 17:44
>> To: Discussion list for GROMACS users
>> Subject: Re: [gmx-users] umbrella sampling
>>
>> I never did it myself, but generally speaking you don' t expect that there'
>> s a lot of room inside a crystal for any molecule to diffuse there,
>> especially when it comes to cross something like a crystallographic plane,
>> meaning that huge energy barriers should be there, leading to NaN and other
>> weird numerical issues.
>>
>> in principle, larger force constants might help you to sample high energy
>> barrier, but it will be most likely a trial and error procedure until you
>> fine-tune the forces along the profile.
>>
>> best
>>
>> Andre
>>
>> On Tue, Jul 11, 2017 at 1:09 PM, Ben Tam <btam125 at hotmail.co.uk> wrote:
>>
>>> Dear All,
>>>
>>> I am doing umbrella sampling for a water molecules moving inside a crystal
>>> structure, however I am running into a problem on the output file with
>>> profile.xvg all value showing “nan”. This error has occurred when I reduce
>>> the slide width from 0.2 nm to 0.1 nm. I have used the exact pull code for
>>> 0.2 and 0.1, however only 0.2 has given me meaningful profile. The pull
>>> code I used is the following:
>>>
>>> ;Pull code
>>> pull = yes
>>> pull_ngroups = 2
>>> pull_ncoords = 1
>>> pull_group1_name = MOL
>>> pull_group2_name = SOL
>>> pull_coord1_type = umbrella ; harmonic biasing force
>>> pull_coord1_geometry = distance ; simple distance increase
>>> ;pull_coord1_vec = 0 0 1
>>> pull_coord1_groups = 1 2
>>> pull_coord1_dim = N N Y
>>> pull_coord1_rate = 0.01 ; 0.01 nm per ps = 10 nm per ns
>>> pull_coord1_k = 10 ; kJ mol^-1 nm^-2
>>> pull_coord1_start = yes ; define initial COM distance > 0
>>>
>>> and the time step I used is 500 ps, but I have also tried 100 and 50 ps. I
>>> have tried increasing the k value to 50 and 100 with the same results.
>>>
>>> The histogram shows many overlap, however when I use k value 100, it is
>>> only giving me individual single peak. Therefore can someone enlighten on
>>> what is going on?
>>>
>>> Best regards,
>>>
>>> Ben
>>>
>>>
>>> --
>>> Gromacs Users mailing list
>>>
>>> * Please search the archive at http://www.gromacs.org/Support
> 
>> Support - Gromacs<http://www.gromacs.org/Support>
> 
>> www.gromacs.org<http://www.gromacs.org>
>> When posting a request for assistance to the Mailing Lists, be sure to. cite your GROMACS version number (and if it's not the most recent, try installing that first!),
>>
>>
>>
>>> /Mailing_Lists/GMX-Users_List before posting!
>>>
>>> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>> Mailing Lists - Gromacs<http://www.gromacs.org/Support/Mailing_Lists>
>> www.gromacs.org<http://www.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

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


More information about the gromacs.org_gmx-users mailing list