Coming Attractions in LAMMPS

The current version of LAMMPS incorporates nearly all the features from previous parallel MD codes developed at Sandia. These include earlier versions of LAMMPS itself, Warp and ParaDyn for metals, GranFlow for granular materials, and GRASP for reactive potentials.

These are new features we'd like to eventually add to LAMMPS. Some are being worked on; some haven't been implemented because of lack of time, since some of them require a lot of development work. Some we're not even sure how to do.

Coupling to finite elements for streess-strain

The group at Sandia who wrote and released the atom-to-continuum USER-ATC package for coupling LAMMPS to a finite-element solver for thermal problems is also working on extending the finite-element portion of the package so LAMMPS could be coupled to a continuum stress/strain deformation solver.

New ReaxFF implementation

With collaborators from Purdue University, we're working on a new implementation of ReaxFF in LAMMPS that is written in C and should allow the charge-equilibration portion be be invoked independently, so it can be used by other potentials. The new implementation should be easier to compile (no Fortran) and hopefully be faster that the current Fortran library version.

Nudged elastic band

We're working on a nudged elastic band (NEB) solver for LAMMPS. We hope to have a preliminary version out by the end of 2009.

Temperature accelerated dynamics

We're working on an implementation of Art Voter's temperature accelerated dynnamics (TAD). It will probabaly by 2010 before we have a version ready to release.

Triangulated particles

We're starting to work on a new particle type: flat triangles. And the associated interaction potentials for them. The motivation is two-fold. To allow them to be coupled together to form general aspherical rigid bodies. And to allow them to be used as "wall" particles, to build interesting containers.

Stochastic rotation dynamics

We've implemented a fix that does stochastic rotation dynamics (SRD) which is a cheap solvent model. We plan to release it in sometime in 2010.

Stokesian dynamics via fast lubrication dynamics

We've worked with Prof Jon Higdon's group at U Illinois to implement a MD-like variant of Stokesian Dynamics called Fast Lubrication Dynamics. We're still experiementing with it, but plan to release it sometime in 2010.

NPT with changing box shape (Parinello-Rahman)

LAMMPS supports triclinic (non-orthongal) simulation domains and NEMD simulations (dynamic box shape change), but the NPT integrator has not been enhanced to allow for shape changes due to off-diagonal pressure tensor components. Aidan is working on this and we hope to have it out by then end of 2009.

Long-range point-dipole solver

A point-dipole force field with a cutoff is in LAMMPS, and we have code for a version that includes the long-range damping terms, but the long-range dipole-dipole solver (Ewald or PPPM) still remains to be done, as well as NVT/NPT integrators.

Per-atom energy and stress for long-range Coulombics

LAMMPS can calculate per-atom energy and stress (virial) for all its potential models, including many-body models. But not for the contribution from long-range Coulombics. We don't know how to formulate the equations for this, or if it is even meaningful.

Long-range Coulombics via Ewald and PPPM for triclinic boxes

LAMMPS currently does not allow either Ewald or PPPM for non-orthogonal boxes (triclinic). We're not sure how to do these computations in that geometry, but it would be useful for NEMD models.

Metadynamics

A user spoke with us about integrating his metadynamics work into LAMMPS, but nothing official has happened yet.

Direct Simulation Monte Carlo - DSMC

DSMC is a discrete low-density fluid model, which can be computed in a MD-like style. LAMMPS has an initial pair_dsmc pair style, which computes the basic rules for DSMC collisions. But many more DSMC options, such as boundary conditions and surface models, could be added to the code.