LAMMPS

What LAMMPS is

LAMMPS is a classical molecular dynamics code for solids, liquids, polymers and soft matter. It ships a large library of interatomic potentials and runs in parallel on CPUs and GPUs. For thermal transport it provides the heat flux, Green-Kubo and nonequilibrium methods. On the map it serves trajectories, correlation functions and the transport quantities built from them.

WebsiteDocumentationOn the mapAll codesLicense: GPL-2.0

In more detail

Algorithms

Classical molecular dynamics integrates Newton's equations for N point masses. For short-range models the forces come from pairs and triplets within a cutoff rc, so the work per step scales as N (Plimpton 1995). The code finds neighbors with Verlet lists built inside an extended radius rs = rc + delta and reused for several steps, or by binning atoms into cells of side rc and checking 27 bins. The fastest serial codes combine both and count each pair once through Newton's third law. Plimpton's three parallel algorithms assign each processor a fixed set of atoms, a fixed block of the force matrix, or a fixed spatial region, and the spatial algorithm scales optimally for large systems.

Numerics and scale

The benchmark is a Lennard-Jones liquid at reduced density 0.8442 and temperature 0.72 with cutoff 2.5 sigma, 55 neighbors per atom, neighbor lists rebuilt every 20 steps and a leapfrog step of 0.00462 reduced units (Plimpton 1995). Systems from 500 to 100,000,000 atoms ran on the nCUBE 2, Intel iPSC/860 and Paragon and Cray T3D. The 100 million atom case nearly filled the 30 GB of a 1904-processor Paragon and ran at 6.97 Gflops. On 1840 dual-processor Paragon nodes the assembler-optimized spatial code was 415 times faster than one Cray Y-MP processor and 165 times faster than one C90 processor. The largest runs on the nCUBE 2, Paragon and T3D were about 90 percent parallel efficient.

Runs in the application papers

Henry 2008 runs single polyethylene chains with AIREBO under periodic boundaries, with a 0.25 fs step for 10 ns in the microcanonical ensemble after 100 ps of equilibration. The conductivity is the Green-Kubo integral of the heat flux autocorrelation along the backbone, with the volume set by an 18 square Angstrom cross section. Wei 2011 holds ribbons at 300 K with a Nose-Hoover thermostat for 200 ps, then exchanges the velocities of the hottest and coldest atoms every 25 fs under NVE with a 0.5 fs step. Mortazavi 2021 relaxes strained 80 by 80 Angstrom sheets under NVT with moment tensor potentials.

Validation

The application papers carry their own checks. Wei 2011 reproduces the conductivity of a 49.26 nm (10,10) carbon nanotube at 302.29 W/m/K against 301.47 W/m/K from an earlier study under the same conditions. In Henry 2008 the fitted modal limit of a polyethylene chain approaches 350 W/m/K and agrees with the convergent Green-Kubo runs. It exceeds the 55 W/m/K inferred from an earlier nonequilibrium study whose boundary scattering would have hidden nonattenuating modes. Plimpton 1995 compares its parallel timings to the fastest reported vectorized Cray Y-MP and C90 code on the same benchmark.

What it computes on the map

12 quantities, drawn by tier with the data flow between them. Dashed boxes are intermediate quantities. Select one for its formula and the other codes that compute it.

Each quantity with the interface the representation records.

• ElasticConstantsexamples/ELASTIC log 'Elastic Constant CXYall = ... GPa' block (in.elastic)GPa
• HeatCurrentcompute <id> all heat/flux ke pe stress
• HeatCurrentACFfix <id> all ave/correlate <Nevery> <Nrepeat> <Nfreq> c_heatflux[1] c_heatflux[2] c_heatflux[3]
• MassDensitythermo 'density' column (thermo_style custom ... density)gram_per_cm3
• MeanSquaredDisplacementcompute <id> all msd com yes + fix <id> all ave/time <Nevery> <Nrepeat> <Nfreq> c_msd[4]
• PotentialLAMMPS input script: pair_style <name> + pair_coeff <args>
• Pressurethermo 'Press' / compute pressure scalar (auto-created thermo_press)GPa
• Temperaturecompute temp / thermo_style temp; profiles via compute chunk/atom + fix ave/chunkkelvin
• ThermalConductivity[transport_model=green_kubo]in-script trap() on a fix ave/correlate vector, scaled by a variable (no dedicated kappa.out file)
• ThermalConductivity[transport_model=nemd]post-processed: flux / (linear fit of binned T(z))
• Trajectorydump custom <id> all custom <every> <file> id x y z
• VelocityAutocorrelationcompute <id> all vacf + fix <id> all ave/time <Nevery> <Nrepeat> <Nfreq> c_vacf[*]

Releases on record

Versions on record for this code, with the license read from the release itself.

References

The code

The papers that describe the code; cite them when you use it.

1. S. Plimpton, Fast Parallel Algorithms for Short-Range Molecular Dynamics, Journal of Computational Physics, 1995. doi.org/10.1006/jcph.1995.1039
2. A. P. Thompson, H. M. Aktulga, R. Berger et al., LAMMPS - a flexible simulation tool for particle-based materials modeling at the atomic, meso, and continuum scales, Computer Physics Communications, 2022. doi.org/10.1016/j.cpc.2021.108171

Applications

Highly cited studies that cite the code, with a few picked by hand.

1. A. Henry, G. Chen, High Thermal Conductivity of Single Polyethylene Chains Using Molecular Dynamics Simulations, Physical Review Letters, 2008. doi.org/10.1103/PhysRevLett.101.235502
2. N. Wei, L. Xu, H. Q. Wang et al., Strain engineering of thermal conductivity in graphene sheets and nanoribbons: a demonstration of magic flexibility, Nanotechnology, 2011. doi.org/10.1088/0957-4484/22/10/105705
3. V. Singh, T. L. Bougher, A. Weathers et al., High thermal conductivity of chain-oriented amorphous polythiophene, Nature Nanotechnology, 2014. doi.org/10.1038/nnano.2014.44
4. B. Mortazavi, B. Javvaji, F. Shojaei et al., Exceptional piezoelectricity, high thermal conductivity and stiffness and promising photocatalysis in two-dimensional MoSi2N4 family confirmed by first-principles, Nano Energy, 2021. doi.org/10.1016/j.nanoen.2020.105716

How to cite

A. P. Thompson, H. M. Aktulga, R. Berger, et al., LAMMPS - a flexible simulation tool for particle-based materials modeling at the atomic, meso, and continuum scales, Comput. Phys. Commun. 271, 108171 (2022)

doi.org/10.1016/j.cpc.2021.108171