Problem with OPLS-UA and shake

Hi all,
I have to use the united atoms OPLS-UA force field with LAMMPS to simulate n-butane molecules.
Since only dihedrals are allowed to change on this force field, I tried to fix bonds and angles with shake. LAMMPS, however, complains that shake clusters are connected and stops. I read on the manual that for efficiency issues shake does not work for long chain polymeric backbones. Is n-butane already too long?

At this point the only option seems to put a very high harmonic constants for bonds and angles in order to keep them as fix as possible, am I right?

Thanks,
Davide

Hi all,
I have to use the united atoms OPLS-UA force field with LAMMPS to simulate
n-butane molecules.
Since only dihedrals are allowed to change on this force field, I tried to
fix bonds and angles with shake. LAMMPS, however, complains that shake
clusters are connected and stops. I read on the manual that for efficiency
issues shake does not work for long chain polymeric backbones. Is n-butane
already too long?

yes. anything where a constrained bond connects
two atoms with other constrained bonds.

At this point the only option seems to put a very high harmonic constants
for bonds and angles in order to keep them as fix as possible, am I right?

yes. possibly in combination with r-RESPA and newton off for bonds.
so you don't have to do the (costly) non-bonded timesteps as frequent
as required for the stiff bonds.

the alternative would be to implement a better scaling constraining algorithm,
e.g. the revised p-LINCS from gromacs.

cheers,
     axel.

Thank you, Axel.

DV