[gmx-users] infinite molecule problems - Follow up info

Sherwin J. Singer singer at chemistry.ohio-state.edu
Tue Sep 30 21:52:01 CEST 2003


While I did not get "inconsistent shifts" errors from grompp, as I
reported in my previous post, I did find

    "There were 180 inconsistent shifts. Check your topology"

in the log file from mdrun.

Of course, I've checked and re-checked the topology.  There could be a
mistake, but....

Thanks,
  Sherwin


On Tue, 2003-09-30 at 15:08, Sherwin J. Singer wrote:
> I am trying to build a slab of silicate as an infinite molecule,
> eventually run run with water and solutes.  A picture of what I'm trying
> to build is located at  www.chemistry.ohio-state.edu/~singer/md
> There are many posts to gmx-users with uses having similar problems with
> infinite, periodically replicated molecules.  Unfortunately, I can't
> seem to figure this one out from those previous posts.
> 
> Also unfortunate is that this message is so long, that I feel compelled
> to give a table of contents:
>  1) Description of symptoms
>  2) Is David van der Spoel's pbc = full the only way?  How to implement?
>  3) Is my procedure for making the topology file correct?
> 
> Thanks in advance for helping this newbie.
> 
> Sherwin Singer
> 
> ---------------------------------------------------------------------------
> 
> 1) Symptoms
> 
> Running bash shell, I have "declare -x  GMXFULLPBC=1", and I set "pbc =
> xyz" in the mdp file.
> 
> I do not get any "inconsistent shift" errors from grompp under these
> conditions.
> 
> I tried to run md with very small time steps (0.0002 ps) and low
> temperature (20K).  I first get the error,
>  "Warning: 1-4 interaction at distance larger than 2.25
>   These are ignored for the rest of the simulation"
> then a few of the following
>  "Warning: Only triclinic boxes with the first vector parallel to the
>   x-axis and the second vector in the xy-plane are supported.
>          Box (3x3):
>             Box[    0]={         nan,          nan,          nan}
>             Box[    1]={         nan,          nan,          nan}
>             Box[    2]={         nan,          nan,          nan}
>          Can not fix pbc."
> and finally
>  "Fatal error: ci = -1 should be in 0 .. -1 [FILE nsgrid.c, LINE 210]"
> I also failed to get energy minimization to work.
> 
> 2) David van der Spoel's pbc = full
> 
> Searching the gmx-users archives, I found a very interesting post,
> http://www.gromacs.org/pipermail/gmx-users/2003-September/033572.html ,
> where David van der Spoel writes that pbc = full may be necessary.  Is
> it the only way?  Presently I am using Gromacs 3.1.4 installed with
> rpm's.  (I had trouble compiling on RH 9.0.)  If necessary, I will build
> a later version.
> 
> 3) Perhaps my procedure for making the topology file is wrong.
> 
> The [ bonds ] and [ angles ] list seemed straightforward.  To generate
> the [ pairs ] and [ dihedrals ] lists I listed all unique 1-2-3-4 walks
> through the crystal and listed each one under [ dihedrals ].  Then I
> took the 1-4 members of each of those walks and listed them under [
> pairs ], and I'm using the parameters for those 1-4  interactions from
> the ffgmx files.  I was a little unsure about this last step, because in
> the discussion of the [ dihedrals ] section of the urea example (page
> 103 of the  3.1.1 Gromacs manual), it says to "do not forget to erase
> the 1-4 interactions in [ pairs ]!!"  What does this mean?
> 
> Also, I assigned charges to all the SI, OS, OA and HO atoms of my chunk
> of silicate.  I could not find default charge parameters in the force
> field.
-- 
Sherwin J. Singer
Department of Chemistry
Ohio State University
Columbus OH, 43214
singer at chemistry.ohio-state.edu
614-292-8909
614-292-1685 (fax)




More information about the gromacs.org_gmx-users mailing list