[gmx-users] NM analysis

Oleg Katchanov oleg at drug.org.ru
Fri Mar 29 15:00:34 CET 2002


  Guten Tag, Herr Bert! :-)

Thank you for your help!
But. I've tried to do the EM procedure (first - steep, second - CG), 
I've done it repeatedly, but I can't to get the max. force smaller than 
N.NNNe+03. I've written the nsteps 10 000, emtol 0.1, emstep 0.01 in the 
em.mdp. MDRUN reports:

Conjugate gradients:
   Tolerance         =  1.00000e-01
   Starting Energy   = -1.806502929687500e+04
   F-Norm            =  2.86931e+03

Negative w:  -6.317044878279e+15
z=  -7.175720000000e+07
gpa=   2.911606067096e+07, gpb=   3.938080970864e+08
a=   0.000000000000e+00, b=   4.470657231392e-06
EpotA=  -1.817659960938e+04, EpotB=  -1.743941601562e+04
Negative number for sqrt encountered (-6317044878278656.000000)
Terminating minimization

writing lowest energy coordinates.
Maximum force:  6.26841e+03

Conjugate Gradients did not converge in 10000 steps
{*usually mdrun make here about 30-100 steps*}
  Function value at minimum =  -1.8177e+04

 After several repeat of the EM the value of  max force and E pot stops 
to change with the next repeat.
What can you advise in this case? Maybe, the temperature must be decreased?
Or another way: the following advise in Manual is given. To run MD at 
the high temperature for a while, and then quench it slowly down to 
required t. When do this repeatedly, the global energy minimum /may be/ 
discovered...
I'll be glad for any advises!
Thank you in advance!

Oleg.

Bert de Groot wrote:

>Oleg Katchanov wrote:
>
>>I have some problems with a Normal Mode analysis. 
>>
>>
>>editconf -f MOL.gro -o MOL.gro
>>grompp -f em.mdp -c MOL.gro -p MOL.top -o MOL_em.tpr
>>#integrator steep in em.mdp;
>>
>
>it's better to take cg for the later stages of minimization. Remember that
>for NM, you need to have a fully minimised system (ie a maximal force
>smaller than approx 0.01), steep is not efficient after the first 100 steps
>or so. 
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20020329/fe90268a/attachment.html>


More information about the gromacs.org_gmx-users mailing list