[gmx-users] Very High Forces During Minimisation

Justin A. Lemkul jalemkul at vt.edu
Fri Aug 7 01:15:30 CEST 2009



Nancy wrote:
> I have tried 15,000 steps; I noticed that Fmax starts out at ~3.0e+03 
> and increases to ~1.2e+06 before coming down very slowly to ~5.0e+04.  
> Additionally, letting the minimisation run for a long time gives me a 
> "converged to machine precision, but not to Fmax < 10" message.
> 

And is that Fmax always on the order of 5e+4?  An Fmax of 10 is very low, most 
systems run with single-precision will not reach this value.

> I am wondering why the Lysozyme tutorial at:
> 
> http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/index.html
> 
> does not have the same problems?
> 

Because when I wrote the tutorial, I chose a system that would reliably 
energy-minimize: a well-defined protein structure from a high-quality model, 
surrounded by only water.  Such systems routinely converge to Fmax < 1000, much 
less rigorous than what it seems you are trying to do.

Note that just because one tutorial works with a protein in water, that does not 
mean all systems will inherently be problem-free.

> Is there a way to stop minimisation and continue later from that same point?
> 

You can try to extend it like any other simulation:

http://oldwiki.gromacs.org/index.php/Extending_Simulations

-Justin

> Thanks,
> 
> Nancy
> 
> On Thu, Aug 6, 2009 at 6:45 PM, Justin A. Lemkul <jalemkul at vt.edu 
> <mailto:jalemkul at vt.edu>> wrote:
> 
> 
> 
>     Nancy wrote:
> 
>     <snip>
> 
> 
>         Note: the ethanediol.log file contains a section that has
>         several lines with asterisks:
> 
>         ============================
>            Angles Force Field Results
>         Angle        Atoms        force    angle     method measured
>           1   H12-   C1-  H11   ******   ******   ******    109.403
> 
> 
>     Probably because these angles don't exist in a UA force field, as
>     they involve non-polar hydrogens.
> 
> 
> 
>         $ editconf -f ethanediol.gro -o ethanediol_box.gro -c -d 1.0 -bt
>         cubic
> 
>         $ genbox -cp ethanediol_box.gro -cs spc216.gro -o
>         ethanediol_solv.gro -p ethanediol.top
> 
>         $ grompp  -f minim.mdp -c ethanediol_solv.gro -p ethanediol.top
>         -o em.tpr
> 
>         the "minim.mdp" file is:
> 
>         ============================
>         define        = -DFLEXIBLE
>         integrator    = steep
>         emtol        = 10.0
>         emstep          = 0.01
>         nsteps        = 2000
> 
> 
>     Increase the value of nsteps dramatically.  If you are aiming for
>     Fmax < 10, 2000 steps likely is not sufficient.  Use a ridiculously
>     high value, and if mdrun completes all those steps without
>     converging, keep increasing the value. I typically set nsteps =
>     500000 to make sure the EM doesn't stop before reaching emtol.  If
>     it does, then something may be wrong.
> 
> 
>         nstlist        = 1
>         ns_type        = grid
>         rlist        = 1.0
>         coulombtype    = PME
>         rcoulomb_switch    = 1.0
> 
> 
>     With PME, rcoulomb_switch is meaningless.
> 
>         rvdw_switch    = 1.3
> 
> 
>     With the default vdwtype (cutoff), rvdw_switch is also meaningless.
> 
>     <snip>
> 
> 
>         Steepest Descents did not converge to Fmax < 10 in 2001 steps.
>         Potential Energy  = -3.3810168e+04
>         Maximum force     =  5.4957539e+04 on atom 559
>         Norm of force     =  2.7119446e+03
>         ============================
> 
>         I have already tried increasing the number of steps, but I found
>         no way
> 
> 
>     How many steps did you try?
> 
> 
>         of lowering the forces.  I have found that by limiting the
>         number of solvent molecules to ~5, I can manage to acheive lower
>         energies, however, the amount of time it takes for a larger
>         number of water molecules increases exponentially.
> 
>         So, I am trying to proceed to equilibration:
> 
> 
>     Stop right here.  It is a waste of time.  If you EM does not
>     converge, you *cannot* force the system to run.  It will explode,
>     sooner or later.  Depending on how soon it does will determine how
>     much time you waste.
> 
>     At this point, you are really getting the same advice over and over,
>     because the question is always the same, but the values of Fmax are
>     different.  Can you post the topology you're using for this system,
>     in particular, the parameters for ethanediol?
> 
>     -Justin
> 
>     -- 
>     ========================================
> 
>     Justin A. Lemkul
>     Ph.D. Candidate
>     ICTAS Doctoral Scholar
>     Department of Biochemistry
>     Virginia Tech
>     Blacksburg, VA
>     jalemkul[at]vt.edu <http://vt.edu> | (540) 231-9080
>     http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin
> 
>     ========================================
>     _______________________________________________
>     gmx-users mailing list    gmx-users at gromacs.org
>     <mailto:gmx-users at gromacs.org>
>     http://lists.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
>     <mailto:gmx-users-request at gromacs.org>.
>     Can't post? Read http://www.gromacs.org/mailing_lists/users.php
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://lists.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

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

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

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



More information about the gromacs.org_gmx-users mailing list