[gmx-users] box of methanol

[Luca Monticelli]

Hi David,

I followed your advices, and run a 5ns 
MD on my methanol boxes with constraints 
on all bonds and dispersion correction. 
The good news is that the difference in 
the density between the cut-offs 0.8/1.4 
and 0.9/1.4 reduces dramatically, as you 
predicted; the bad news is that both 
results get worse: density is now 856 
and 850, respectively (the error 
increased from ~1% to 7~8%). No 
dependence on the box size, in any case, 
as opposed to what reported by Yiannis 
using OPLS (but he was using very 
different cut-offs in his 2 boxes).
Actually, I couldn't find in van 
Gunsteren's papers if they used 
dispersion correction or not, nor any 
informations about the density they got. 
So I'm not sure what conditions to use 
in my simulations, advices are welcome

luca


David van der Spoel wrote:

> On Thu, 25 Jul 2002, Luca Monticelli wrote:
> 
> 
>>I used the G43a1 force field, with
>>cut-offs 0.8/1.4 (Gromos-96 was
>>parameterized with these twin-range
>>cut-offs, right?) and NPT conditions;
>>box size is approximately 4.3*4.3*4.3,
>>timestep=2fs, nstlist=5, no constraints,
>>compressibility= 4.575e-4; total time is
>>5 ns.
>>
>>These are the results.
>>It takes ~500 ps for the pressure to 
>>become relatively stable.
>>Excluding the initial 500 ps, the 
>>average d=749, with
>>an RMSD of 3.9.
>>
>>If I change the cut-offs to 0.9/1.4, 
>>leaving all other
>>parameters identical, I find d=782, with
>>an RMSD of 3.5. Not bad, the error is 
>>only ~1.1%.
>>
> 
> This is a serious difference. I guess you are not running with dispersion
> correction, use
> dispcorr = enerpres, that will make the difference between the two
> simulations smaller, and also contract your box slightly more.
> Furthermore a timestep of 2 fs is too long if you are not using
> constraints. If you really object to using all constraints, you should go
> down to 0.5 fs, or 1 fs when you constrain only the H.
> 
> 
>>If anybody is interested in using this 
>>box, just let me know.
>>
> It would also be interesting to compare the results to the all-atom OPLS
> model that was posted on the list recently.
> 
> Groeten, David.
> ________________________________________________________________________
> Dr. David van der Spoel, 	Biomedical center, Dept. of Biochemistry
> Husargatan 3, Box 576,  	75123 Uppsala, Sweden
> phone:	46 18 471 4205		fax: 46 18 511 755
> spoel at xray.bmc.uu.se	spoel at gromacs.org   http://zorn.bmc.uu.se/~spoel
> ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
> 
> 
> 


-- 
____________________________________________________

Luca Monticelli
Centre for bio-molecular 
Interdisciplinary Studies
and Industrial applications
University of Milan, Italy
Istituto di Chimica del Riconoscimento 
Molecolare
C.N.R., Milano, Italia
e-mail: luca.monticelli at unimi.it
phone: +39 02 2850 0031
fax:   +39 02 2850 0036
____________________________________________________