octadecanol+density+ problem

hello everyone

I’m modeling octadecanol alcohol with lammps and facing a problem for weeks. i tried so many suggestions but none worked. problem is that although code runs without problem my density value is extremely different from experimental data. my forcefield is opls-aa and i double checked all charges and parameters. if anyone can help i will be so much glad. data files are attached

octadecanol.txt (3.49 KB)

octadecanol.data (295 KB)

hello everyone
I'm modeling octadecanol alcohol with lammps and facing a problem for weeks.
i tried so many suggestions but none worked. problem is that although code
runs without problem my density value is extremely different from
experimental data. my forcefield is opls-aa and i double checked all charge
and parameters. if anyone can help i will be so much glad. data files are
attached

there are several *serious* and fundamental problems with your input
deck. don't you have somebody around, that has some experience in MD?
these are all issues an experienced person will spot immediately. in
decreasing order of severity:

1) your special_bonds parameters are _wrong_. OPLS-AA doesn't mix in
non-bonded contributions from bonds or angles. not sure about 1-4
interactions. check the paper describing the specific variant of the
force field you are using.

2) you are thermostatting all atoms twice, a *very* bad idea which is
known to cause unphysical behavior

3) you are using a timestep of 1 femtosecond on a system with
unconstrained hydrogens which will lead to lack of energy conservation
or worse. you need to reduce it or use fix shake to constrain all
bonds involving hydrogens

4) your nose-hoover time constants are a bit on the short side.

in addition, you may want to check out the following:

- your cutoff of 9 \AA seems on the short side. more typical for
systems like this is 12 \AA. again, check the force field paper and
try to validate your setting by reproducing data published results.

- similarly, your pressure may be significantly affected by the lack
of using a correction for the missing tail of the LJ interactions.
they can be added via pair_modify tail yes. again, running tests
against results from the force field paper should help your find the
best setting. including tail correction is the more important, the
shorter your lj cutoff is.

- you probably want to use pppm instead of ewald, as it is much faster.

- if you know the target density, it is going to be *much* easier to
initially scale your system to have the desired target density and
then equilibrate at the expected volume with fixed volume, but
elevated temperature. then cool down to further equilibrate and
finally run with npt to relax to the final desired result.

- keep in mind, that the density you will be observing is the density
due to the force field, which may be close to experimental or not as
close.

there is a lot of work ahead of you.

axel.