[gmx-users] Positive Total energy

[Mark Abraham]

----- Original Message -----
From: Thanasis Koukoulas <koukoulas_th at yahoo.gr>
Date: Tuesday, June 15, 2010 4:55
Subject: [gmx-users] Positive Total energy
To: gmx-users at gromacs.org

-----------------------------------------------------------
| > Hello 
> I have sent several emails and i am sorry. It is my last attempt to finish this simulation.> The molecule i have is C-C-C-C-C-C-O-C-C-O-C-C-O-H or C10H1O3. (ciej molecule)> 
> I dont use gromacs libraries> the files i use are attached to the email in which i have descriped step by step the forcefield i use. > as you can see in [pairs] i have an extra non Lj 1-4 pair. i did it because the model i use has a potential between atom 10 and 14 ether oxygen and alchohol hydrogen interaction. It is described in the paper as a repulsive energy between these two atoms. and it has a formula:> 
> Erepulsive = 3.325789e-7/r^12> 
> the only way to import in gromacs i think is this:> 
> [ pairs ]> ; ai aj  funct > 10 14    2 0 0 0 0 3.325789e-7> 
> it is an intra-molecular only interaction. so i couldnt think another way to do it except this. as you can see i have several zeros for the charges, C6=o and C12 = 3.325789e-7 fidgeQQ=0 for this case> 
> the problem is i get positive total energy. i did this sim for only one molecule just to see what is wrong and i think that coulomb (SR) are wrong for sure.> 
> so i want > 
> coulomb 1-4 /2   !thats why i put fudgeQQ = 0.5> 
> the values i get at the step 0 are> 
> Energies (kJ/mol)>           Angle    Proper Dih.          LJ-14     Coulomb-14        LJ (SR)>     2.30340e+03    1.08417e+00     3.16570e+00    8.79204e+04   -2.54164e+03>   Disper. corr.   Coulomb (SR)   Coul. recip.      Potential    Kinetic En.>    -2.51250e+00   -4.81042e+04   -2.84931e+04    1.10867e+04    1.84852e+04>    Total Energy    Temperature Pressure (bar)  Cons. rmsd ()>     2.95719e+04    2.99528e+02   -3.03583e+02    3.13477e-06

 |
-----------------------------------------------------------
OK, now that you've simplified, it seems reasonably clear that Coulomb-14 is badly broken. My leading theory is that grompp is somehow reading (part of) the text of your C12 value into a charge field.

I would test with this [pairs] line commented out, with zero in place of your intended C12 value, and with your C12 value expressed as 0.00000325789. If that doesn't shed light on the nature of the problem, please report the Step 0 energies (as above) in each case.

Then, use

gmxdump -s topol.tpr | less

and search for 3.257 and maybe attach 50 lines either side of the line that looks like it matches that value in a parameter entry. Don't attach the whole file, it'll be huge.

As a side point, the contents of your #included files seems to achieve a correct result, although your organization of the content with respect to the filenames is badly at odds with the correct approach, and will give you problems later. Compare with a simple tutorial example.

Mark

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20100615/af1b03f3/attachment.html>