respa

I am simulating a system consisting of 128 phosphonium ions made of one central charged phosphorous atom with four alkyle ligand and 128 negative borate ions. When I simulate this system at constant pressure using run_style respa with the recommended parameters in the manual

run_style respa 4 2 2 2 inner 2 4.5 6.0 middle 3 8.0 10.0 outer 4

the system contracts to an unphysically high density. If I use this alternative:

run_style respa 2 2 bond 1 angle 1 dihedral 1 improper 1 pair 1 kspace 2

I get the correct density or at least a density that agrees with the one obtained from other programs but only if the timestep is very short - 0.4 fs. Therefore I wonder whether there is a better choice of paramters for run_style respa. I use the following paramteres:

pair_style lj/charmm/coul/long 8.0 10.0

bond_style harmonic

angle_style harmonic

dihedral_style charmm

improper_style harmonic

kspace_style ewald 0.001

and the following thermostat/barostat:

fix 1 all npt temp 330.0 330.0 100.0 aniso 0.0 0.0 1000.0

Thank you in advance for an answer

Sten Sarman

you already posted this question. if you don't like the answer you
get, don't just repost the same question, but provide a reason why you
consider the answer you got is not satisfactory.

I am simulating a system consisting of 128 phosphonium ions made of one
central charged phosphorous atom with four alkyle ligand and 128 negative
borate ions. When I simulate this system at constant pressure using
run_style respa with the recommended parameters in the manual

run_style respa 4 2 2 2 inner 2 4.5 6.0 middle 3 8.0 10.0 outer 4

but why should that work on your system, when you have already
confirmed that you cannot run at a typical time step for such systems
with verlet run style.

the system contracts to an unphysically high density. If I use this
alternative:

run_style respa 2 2 bond 1 angle 1 dihedral 1 improper 1 pair 1 kspace 2

I get the correct density or at least a density that agrees with the one
obtained from other programs but only if the timestep is very short - 0.4
fs. Therefore I wonder whether there is a better choice of paramters for
run_style respa. I use the following paramteres:

pair_style lj/charmm/coul/long 8.0 10.0

bond_style harmonic

angle_style harmonic

dihedral_style charmm

improper_style harmonic

kspace_style ewald 0.001

and the following thermostat/barostat:

fix 1 all npt temp 330.0 330.0 100.0 aniso 0.0 0.0 1000.0

Thank you in advance for an answer

i already told you, that the problem is likely not with respa, but
your parameters.
you probably just have a metastable state in one case and not in the
other. that doesn't make it correct.

in general, you *first* have to properly get the verlet run style to
work before even considering using respa. also, on modern hardware, i
don't see much of a benefit using the inner/middle/outer scheme ...and
finally a 10.0 angstrom cutoff is very short for the current state of
the art.

axel.