[gmx-users] mdrun -rerun + runedr: combo problem?

[Jones de Andrade]

Hi all.

I think I'm having a really strange problem:

Well, I have some simulations being performed, and wanted to remove from the
bigger molecules any vdw or coulombic intramolecular interaction "longer"
than 1-4. So, from 1-5 included forward, I would like to remove from the
energetic calculations, but not from the interaction dynamics.

To do so, I've performed the simulation with nexcl = 3 (standard, and which
would include them in the dynamics), and after I increased this number and
performed a mdrun -rerun "ressimulation". Then, I runned the runedr.

The problem is: I decided to make a test run. I expect that a "nexcl" number
bigger than the total should not yeld problems. So I choosen to increase the
nexcl in C2F6 from the standard 3 to 4. Then I runned runedr in both
results, and got the following:

nexcl = 3:

Energy                      Average       RMSD     Fluct.      Drift
Tot-Drift
-------------------------------------------------------------------------------
Bond                        3040.69    70.0697    66.5161   0.381591
76.3189
Angle                       4220.72    79.1057    78.0402  -0.224121   -
44.8247
Ryckaert-Bell.              22594.3   0.369332   0.369284 0.000103437
0.0206876
LJ-14                       4345.47    44.7911    44.5899 -0.0734496   -
14.6901
Coulomb-14                  26656.3    12.5831    12.5831 0.000785301
0.157062
LJ (SR)                    -7476.61    69.8678    69.7072 -0.0820061   -
16.4014
Disper. corr.              -233.013    1.80332    1.80306 -0.000532475  -
0.106496
Coulomb (SR)                202.088    3.63903    3.62648 -0.00522946    -
1.0459
Coul. recip.               -228.367     1.4159    1.40872 0.00246637
0.49328

nexcl = 4:

Energy                      Average       RMSD     Fluct.      Drift
Tot-Drift
-------------------------------------------------------------------------------
Bond                        312.923    891.374    891.099   0.383169
76.6345
Angle                       425.523    1210.88    1210.81  -0.224334   -
44.8673
Ryckaert-Bell.              2259.64    6430.13    6430.13 8.98693e-05
0.017974
LJ-14                       434.682    1236.91     1236.9  -0.073917   -
14.7835
Coulomb-14                  2665.87    7586.12    7586.12 0.000608334
0.121668
LJ (SR)                    -747.691    2127.89    2127.89 -0.0822151   -
16.4432
Disper. corr.              -23.3034    66.3157    66.3157 -0.000532475  -
0.106496
Coulomb (SR)                20.2105    57.5108    57.5101 -0.00521119   -
1.04225
Coul. recip.               -22.8387    64.9888    64.9886 0.00246332
0.492669

And the results, for me, doen't make sense. First, all the contributions,
*including* bonds, angles and dihedralss that are *not* to be affected by
the nexcl, are reduced by almost 10. Second, since we are talking about
C2F6, which doesn't have any atom farthen than 3 bonds from any other,
neither the vdw and coulombic contibutions should have been affected.

Could please someone help me with this matter? What have I missed in the
logics here?

Thanks a lot for everybody in advance!

Sincerally yours,

Jones
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20070806/6648640f/attachment.html>