[gmx-users] How to overcome Fatal error of bonded interactions Gromacs 4.0.3

Claus Valka lastexile7gr at yahoo.de
Wed Mar 4 16:12:47 CET 2009


Hello,

the pair interactions I have are 1-4 and 1-5, meaning that they are 
at most 4 bonds away. I have declared them one by one under 
the pairs section.
I do not use any other pair interactions. 

Thank you for your interest,
Nikos

--- Berk Hess <gmx3 at hotmail.com> schrieb am Mi, 4.3.2009:

Hi,

So it is not a charge group problem then.

You also don't have pair interactions that act on much longer distances than atoms separated by 4 bonds?

Berk

Date: Wed, 4 Mar 2009 05:33:15 -0800
From: lastexile7gr at yahoo.de
Subject: RE: [gmx-users] How to overcome Fatal error of bonded interactions	Gromacs 4.0.3
To: gmx-users at gromacs.org

--- Claus Valka <lastexile7gr at yahoo.de> schrieb am Mi, 4.3.2009:

Hello,

by charge groups you are referring to charge group number (cgnr) in the topology file.
Every single atom in my chains belongs to a different charge group. My chains are polymer ones (identical) with 455 atoms each. A sample of my topology file is the following:

[ atoms ]
 ; nr   type  resnr  residu  atom  cgnr  charge
    1    CH3      1   a1_pp     C1    1   0.000
    2    CH3      1   a1_pp     C2    2   0.000
    3        H      1   a1_pp     H3   
 3   0.000
    4        C      1   a1_pp     C4    4   0.000
    5        H      1   a1_pp     H5    5   0.000
    6        H      1   a1_pp     H6    6   0.000
    7        C      1   a1_pp     C7    7   0.000
....

I think that that way I do not have big charge groups. My simulations seem reasonable using this notation. 

Thank you,
Nikos

--- Berk Hess <gmx3 at hotmail.com> schrieb am Mi,
 4.3.2009:
Von: Berk Hess <gmx3 at hotmail.com>
Betreff: RE: WG: RE: [gmx-users] How to overcome Fatal error of bonded interactions Gromacs 4.0.3
An: "Discussion list for GROMACS users" <gmx-users at gromacs.org>
Datum: Mittwoch, 4. März 2009, 14:00


--- Berk Hess <gmx3 at hotmail.com> schrieb am Mi, 4.3.2009:

Von: Berk Hess <gmx3 at hotmail.com>
Betreff: RE: WG: RE: [gmx-users] How to overcome Fatal error of bonded interactions Gromacs 4.0.3
An: "Discussion list for GROMACS users" <gmx-users at gromacs.org>
Datum: Mittwoch, 4. März 2009, 14:00




#yiv340993236 .hmmessage P
{
margin:0px;padding:0px;}
#yiv340993236 {
font-size:10pt;font-family:Verdana;}

Hi,

The question is not if you are using electrostatics or not, but how large your charge groups are
(a better name for charge groups in Gromacs would be neighbor search groups).

The warning you get is for bonded interactions, but it will usually mean that your system
is exploding.
Because changing table-extension solved your problem, one guess would be that you have
too large charge groups, causing non-bonded interactions to be at a too long distance,
which makes your system explode, which in turn gives the bonded interaction error.

Berk

Date: Wed, 4 Mar 2009 03:48:33 -0800
From: lastexile7gr at yahoo.de
Subject: WG: RE: [gmx-users] How to overcome Fatal error of bonded	interactions Gromacs 4.0.3
To: gmx-users at gromacs.org



--- Claus Valka <lastexile7gr at yahoo.de> schrieb am Mi, 4.3.2009:
Von: Claus Valka <lastexile7gr at yahoo.de>
Betreff: RE: [gmx-users] How to overcome Fatal error of bonded interactions Gromacs 4.0.3
An: "Berk Hess" <gmx3 at hotmail.com>
Datum: Mittwoch, 4. März 2009, 12:47

Dear Sir Berk,

Thank you for your quick response.. I simulate a system which has 80 chains in total each having 455 atoms. I do not use any
 electrostatics. The box size vary in each direction with the biggest dimension to be less than 30nm. 

The function I'm using is switch. The cut-off I'm using is 0.895nm. I also have frozen the 64 chains in freezegrps option in the mdp file in all directions. Judging from what you are telling me, I deduce that I correctly assign the table-extension parameter big enough. I'm not interested in taking into account LJ interactions (either pairs or not) that are more than my cut-off distance. So not even to mention in a distance bigger than: table-ext+cut off.

Also reading again the error it is referring to bonded interactions...

Yours SIncerely,
Nikos

--- Berk Hess
 <gmx3 at hotmail.com> schrieb am Mi, 4.3.2009:
Von: Berk Hess <gmx3 at hotmail.com>
Betreff: RE: [gmx-users] How to overcome Fatal error of bonded interactions Gromacs 4.0.3
An: lastexile7gr at yahoo.de
Datum: Mittwoch, 4. März 2009, 12:21




#yiv340993236 .ExternalClass #EC_yiv1174951385 #EC_yiv1959079355 .EC_hmmessage P
{padding:0px;}
#yiv340993236 .ExternalClass #EC_yiv1174951385 #EC_yiv1959079355
{font-size:10pt;font-family:Verdana;}

Hi,

table-extension affects the tables for both pair and non-bonded ionteractions.
non-bonded interactions use tables when you use, shift, switch, PME or user potentials.

You are not by chance simulating large molecules which consist of a single charge group
or of very large charge groups?

Berk

Date: Wed, 4 Mar 2009 03:12:19 -0800
From: lastexile7gr at yahoo.de
Subject: RE: [gmx-users] How to overcome Fatal error of bonded interactions	Gromacs 4.0.3
To: gmx-users at gromacs.org

Hello,

the system I'm referring at is an already equilibrated one. Before, I had already made steepest
 descents. 

You are correct that I have non-bonded interactions at distances that are more than cut-off+table ext distance. I used to put table ext very bin in contrast to my box. My system box was something less than 5nm before and table-ext was 5nm at the time. Now that I simulate a much bigger system than before (in one dimension 6 times bigger), I realized that I had to increase also this parameter.

Since this parameter affects only the pairs interactions I do not see the reason why my simulations are affected by it, even if it is already big enough (5nm in a 30nm box). My pairs interactions are (should) not (be) more that 1nm far away..

I had extensively tested this parameter in gromacs 3, and I saw that it
 affects the pairs interactions only. So I decided to increase it very much and not bother with it.

Is there any difference in newest versions? Does it affect anything more?

Thank you,
Nikos


--- Berk Hess <gmx3 at hotmail.com> schrieb am Mi, 4.3.2009:
Von: Berk Hess <gmx3 at hotmail.com>
Betreff: RE: [gmx-users] How to overcome Fatal error of bonded interactions Gromacs 4.0.3
An: "Discussion list for GROMACS users" <gmx-users at gromacs..org>
Datum: Mittwoch, 4. März 2009, 9:23




#yiv340993236 .ExternalClass #EC_yiv1174951385 #EC_yiv1959079355 .EC_ExternalClass #EC_EC_yiv174728022 #EC_EC_yiv779874946 .EC_EC_hmmessage P
{padding:0px;}
#yiv340993236 .ExternalClass #EC_yiv1174951385 #EC_yiv1959079355 .EC_ExternalClass #EC_EC_yiv174728022 #EC_EC_yiv779874946
{font-size:10pt;font-family:Verdana;}


 



Hi,

Your solution seems to indicate that your system is unstable.
My guess is that your system was not well equilibrated and at some point
some non-bonded interactions were at a distance of more than cut-off+tab_ext.
That would make your system explode. In that case increasing tab_ext to 2 nm
should also suffice.

Berk


Date: Tue, 3 Mar 2009 08:33:52 -0800
From: lastexile7gr at yahoo.de
To: gmx-users at gromacs.org
Subject: [gmx-users] How to overcome Fatal error of bonded interactions	Gromacs 4.0.3

Hello,



this has to do with my personal experience of using Gromacs version
 4.0.3. So this is not an
 official response.

The story:


While I was trying to run an NVT simulation in a monoclinic box I encountered the following problem:


Fatal error:
25 of the 336080 bonded interactions could not be calculated because some atoms involved moved further apart than the multi-body cut-off distance (0.895 nm) or the two-body cut-off distance (0.895 nm), see option -rdd, for pairs and tabulated bonds also see option -ddcheck




and the:
Not all bonded interactions have been properly assigned to the domain decomposition cells

Trying to put aside the fact that my system had any bug I found in the net the
 following:
http://www.gromacs.org/pipermail/gmx-developers/2008-January/002373.html

In order to overcome this error I increased the value of
table-extension          = 30
in my mdp file in such a way, as to be bigger from my biggest dimension I had in my system. The ran proceeded (until now that I'm writing this e-mail) without any problem.

This value has to do with the cut-off of pairs1-4 LJ interactions and is calculated in nm.

JFYI
Regards,
Nikos





See all the ways you can stay connected to friends and family 
_______________________________________________
gmx-users mailing list    gmx-users at gromacs.org
http://www.gromacs.org/mailman/listinfo/gmx-users
Please search the archive at http://www.gromacs.org/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/mailing_lists/users.php

What can you do with the new Windows Live? Find out 




      

Express yourself instantly with MSN Messenger! MSN Messenger 
_______________________________________________
gmx-users mailing list    gmx-users at gromacs.org
http://www.gromacs.org/mailman/listinfo/gmx-users
Please search the archive at http://www.gromacs.org/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/mailing_lists/users.php


      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20090304/d16ee424/attachment.html>


More information about the gromacs.org_gmx-users mailing list