[gmx-users] Re: inconsistent energies

[Mark Abraham]

On 8/01/2011 1:38 PM, Vlad Isaev wrote:
> Thanks, Mark.
>
> No, it is not picking any chk files. But the coordinates in the gro 
> file after mdrun change slightly for some of the atoms on using cg and 
> l-bfgs, whereas with steep they remain the same.

OK, well this must be a consequence of the algorithms, despite being 
zero-step.

>
> e.g., in b4min.gro submitted to cg or l-bfgs or steep:
>
> 2GLY      O   13   0.736   0.922  -0.350
>
> in min.gro after cg and l-bfgs:
>
>  2GLY      O   13   0.735   0.920  -0.349
>
> in min.gro after steep:
>
> 2GLY      O   13   0.736   0.922  -0.350
>
> So, I guess I have to trust steep on this if I just need the energy of 
> the molecule as the coordinates of atoms do not change.
>
> By the look of it, AMBER10 also changes coordinates slightly on using:
>
>  imin=0, maxcyc = 0, ncyc = 0, ntb = 0, cut = 99999
>
> Is there any other way of calculating a single point energy by GROMACS?

There's zero-step MD, so long as you make sure the start is 
unconstrained. The mdrun -rerun functionality does single point energies 
over a trajectory, which can be one configuration. By definition, the 
latter doesn't change configurations.

Mark

>
> -Vlad
>
> On 8/01/2011 12:10 AM, Vlad Isaev wrote:
> >/  Hi,
> />/
> />/  I want to calculate the energy of a peptide. No minimization, or md,
> />/  just the energy at 0 K. I use this input:
>
> />/
> />/  integrator               = steep
> />/  nsteps                   = 0
> />/  nstenergy                = 1
> />/  nstlist                  = 0
> />/  ns_type                  = simple
>
> />/  pbc                      = no
> />/  rlist                    = 0
> />/  rcoulomb                 = 0
> />/  rvdw                     = 0
> />/
> />/
> />/  I get:
>
> />/
> />/  Potential Energy  = -1.1744482e+02
> />/
> />/  Changing steep to cg changes the potential energy even though nsteps=0:
> />/
> />/  Potential Energy  = -1.2410449e+02
>
> />/
> />/  The corresponding energy from AMBER10 is : -1.20172848e+02 (conversion
> />/  factor 4.184).
> />/
> />/  Any ideas why these are so different?
> /
> <  That's pretty weird. Please double check that you were really using the
>
> <  same input coordinates to grompp (and that mdrun wasn't picking up a
> <  checkpoint file).
>
> <  Mark
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20110108/dfe2ff98/attachment.html>