Hello lammps users.
A quick clarification is needed on dihedral_style fourier. It seems that the dihedral coefficients are not written on the data file when one uses the dihedral_style fourier.
Is this a bug or not mentioned in the manual.
Thanks
Arun
Hello lammps users.
A quick clarification is needed on dihedral_style fourier. It seems that the dihedral coefficients are not written on the data file when one uses the dihedral_style fourier.
Is this a bug or not mentioned in the manual.
Most, but not all styles have been adapted to support write_data. Specifically some styles in user packages are missing. You can contact the contributor or submit the changes yourself.
Axel
Thanks for reporting the issue. This issue is not specific to
dihedral_style fourier, and it is not easy to fix. But perhaps we
should add a warning to the documentation. The current documentation
for write_data warns users to use "pair_coeff" commands, but it does
not mention potential problems with bond_coeff, angle_coeff,
dihedral_coeff, improper_coeff commands.
http://lammps.sandia.gov/doc/write_data.html
Generally, my personal opinion is that I would not rely on
write_data to store any "coeff" force-field parameter information if
you use non-standard or user-contributed force-fields, especially if
they were written before the "write_data" command was added to LAMMPS.
The safest way to run LAMMPS is to store all force-field information
using "coeff" commands in an input script. In your case, I would add
"dihedral_style fourier" to the beginning of your LAMMPS input script,
and leave it out of the data file. (This works fine. All of the
moltemplate and topotools examples use coeff input script commands
instead of putting the coeffs in the data file.)
---- details ----
The "write_data" command is quite new. The problem is that many of
these force-fields were written by volunteers before the "write_data"
command existed, and so they don't support it. Furthermore, some
force-field styles require so many parameters, they store them in
auxiliary files (which would also have to be written). This is true
for the many-body pair styles and all of the "table" force fields,
such as dihedral_style table. It would be pain to get all of the
force-fields to support write_data, because the original authors are
often no longer actively involved.
On that note, I'm grateful the write_data command exists.
I hope this information helps.
Andrew
Every thing Andrew says is true.
However, if the force field has auxiliary files (e.g. pair style tersoff, eam,
reax, etc), then we don’t intend to write that data into the data file
via write_data. If the bond, angle, etc style just has a simple bond_coeff,
angle_coeff command with some args, then it is typically about a 5-line
method that needs to be added to write it to the data file.
So you can look at a couple of examples for other dihedral styles,
add your own method, debug it,
then send it to us and we will add it to dihedral style fourier.
Steve