Martini force field in Lammps

LAMMPS LAMMPS Mailing List Mirror
1

Hi all,

I want to simulate the formation of surfactant/nanoparticle micelles in water box and its effect on shear viscosity of the liquid. The force field I use is Martini force field which is commonly used in Gromacs. Lammps provide five different methods to compute the viscosity, and this is why I chose it.
(1) For Martini FF in Gromacs, LJ interactions are shifted to zero in the range 0.9-1.2 nm, and electrostatic interactions in the range 0.0-1.2 nm. I noticed there is also a pairstyle in Lammps can describe this, i.e. lj/gromacs/coul/gromacs. But in this case, PPPM cannot be applied to compute the long-ranged electrostatic interactions. My question here is which pairstyle is good for Martini FF coupled with PPPM.

(2) Besides, I also try to use “lj/cut/coul/long” to set the cut-off, and use “pair-modify shift yes” to make both shit to zero at cutoff. But this shift from 0 to cutoff is a bit different from what Martini FF requires in Gromacs. If there is no better pairstyle in lammps to describe Martini FF, can I just use this instead and then I can also use PPPM?

(3) The third question may be not firmly related to the Lammps application. For martini FF, it is parameterized with short range shited exlectrostatic interaction, and it is not recommended to use PME in Gromacs to compute electrostatic interactions, especially for the standard water model. But why so many people using Lammps simulate the viscous flow of liquid try to use PPPM to calculated the long range electrostatic force? After all, electrostatic interactions in the Martini model are not considered to be very accurate.

Best wishes,

2

Hi all,

I want to simulate the formation of surfactant/nanoparticle micelles in water box and its effect on shear viscosity of the liquid. The force field I use is Martini force field which is commonly used in Gromacs. Lammps provide five different methods to compute the viscosity, and this is why I chose it.
(1) For Martini FF in Gromacs, LJ interactions are shifted to zero in the range 0.9-1.2 nm, and electrostatic interactions in the range 0.0-1.2 nm. I noticed there is also a pairstyle in Lammps can describe this, i.e. lj/gromacs/coul/gromacs. But in this case, PPPM cannot be applied to compute the long-ranged electrostatic interactions. My question here is which pairstyle is good for Martini FF coupled with PPPM.

as you mention below, it is not intended to use a long-range solver. that said, you could either use lj/cut/coul/long and live with the fact, that you are not using a smooth truncation of the lennard-jones term, or use a different pair style with a smooth truncation, but a different one, or create tables for the LJ part of lj/gromacs/coul/gromacs and combine it with pair style coul/long through pair style hybrid/overlay.

(2) Besides, I also try to use “lj/cut/coul/long” to set the cut-off, and use “pair-modify shift yes” to make both shit to zero at cutoff. But this shift from 0 to cutoff is a bit different from what Martini FF requires in Gromacs. If there is no better pairstyle in lammps to describe Martini FF, can I just use this instead and then I can also use PPPM?

it is up to you to decide what is an acceptable error. technically this is possible. LAMMPS doesn’t care whether what you model is doing the correct/best science. it will compute what you ask it to do.

(3) The third question may be not firmly related to the Lammps application. For martini FF, it is parameterized with short range shited exlectrostatic interaction, and it is not recommended to use PME in Gromacs to compute electrostatic interactions, especially for the standard water model. But why so many people using Lammps simulate the viscous flow of liquid try to use PPPM to calculated the long range electrostatic force? After all, electrostatic interactions in the Martini model are not considered to be very accurate.

you are comparing apples and oranges, i.e. a coarse grain model and all-atom models. by their very nature the all-atom models will be more sensitive to errors on handling coulomb interactions. on the other hand, with a coarse grain model, you make rather drastic approximations, so that it is questionable, whether the error from (smoothly) truncating coulomb is significant. even more so, that with a coarse grain model, you have a different time axis. with a coarse grained solvent, solvent mediated processes are faster, and thus the nominal time is not the real time axis, your effective time step must be considered larger, but it is not simple to determine how much.

thus, when worrying about accurate long-range electrostatics for the properties you are interested in, you are neglecting how limited your overall accuracy is because of the uncertainty of your time axis.

axel.