[gmx-users] BUG!!! AFM pulling/Constraint

[Berk Hess]

>From: Jiang ran <ran_jiong_lang2 at hotmail.com>
>Reply-To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>Subject: [gmx-users] BUG!!! AFM pulling/Constraint
>Date: Thu, 24 May 2007 21:56:18 +0800
>
>
>Date: Wed, 23 May 2007 02:29:36 -0700From: yesint4 at yahoo.comTo: 
>gmx-users at gromacs.orgSubject: [gmx-users] Restarting AFM pulling
>
>Hello everyone,I'me trying to run a very long simulation with the AFM 
>pulling, which should be partitioned into many smaller pieces. However, 
>this seems that the pull code does not preserve the force even if the 
>position of the spring is set to what was observed at the end of the 
>previous run.Here is an example. I have one pull group pulled in Z 
>direction, so in the .pdo file, according to the manual, I have the time, 
>the position of the reference, the position of the pull group and the pos. 
>of the spring.The end of the .pdo from the first run looks like 
>this:...12.000001
>  4.537002        1.228887        1.279260I start new run with the 
>afm_init1 = 0 0 -3.257742   (the last number is 1.279260-4.537002 i.e. 
>relative pos. of the spring as it should be).And in the beginning of the 
>new run I get:Hello Everybody,I just red the email of mister Semen and I 
>indeed found a bug in the pull code. I allready checked it together with my 
>professor and could duplicate both error and prove solution.In the code 
>md.c in the funtion :pull(&pulldata,state->x,f,force_vir,state->box,        
>    top,inputrec->delta_t,step,t,           
>mdatoms,START(nsb),HOMENR(nsb),cr);The integer parameter is step incorrect 
>incase of continuation of the run, step has to be equal to step_rel instead 
>of step. Otherwise the following equation in pull.c is not correct:In 
>do_afm();for(m=0;m<DIM;m++)      pull->grp[i].spring[m] = 
>pull->grp[i].AfmInit[m] + pull->ref.x_unc[m] +        
>pull->grp[i].AfmVec[m]*pull->grp[i].AfmRate*step*dt;As 
>pull->grp[i].AfmInit[m] is reset, step isn't. In case of resubmitting the 
>occurring error scales with rate and the length of the run. The same error 
>in the constraint part in pull.c.I reported the bug 
>at:http://www.pku.edu.cn/ehomepage.htmgreetings,Jiang,

You are right.
But in the current pull code there is no real solution to this problem,
since the initial position of the spring is set to the pull group position,
which is not equal to the position of the spring at the end of the previous 
run.

There are many more problem in the pull code in 3.3.1 (and before).
For 4.0 I have completely rewritten the pull code, so (nearly) all problems
should be solved then.

Berk.

_________________________________________________________________
Play online games with your friends with Messenger 
http://www.join.msn.com/messenger/overview