Four canonical molecular dynamics simulations wrapped as process-bigraph
Processes using the LAMMPS engine. Each simulation is driven by a standard
LAMMPS .in input file (shown inline below).
Binary LJ mixture demixing into condensate-like domains
A 50:50 binary Lennard-Jones mixture is quenched below its critical solution temperature. Same-species attractions (epsilon_AA = epsilon_BB = 1.0) are stronger than cross-species (epsilon_AB = 0.5), driving spontaneous phase separation via spinodal decomposition. Composition fluctuations amplify and coarsen into macroscopic A-rich and B-rich domains — the same mechanism underlying biomolecular condensate formation in cells. The domain growth follows the Lifshitz-Slyozov t^(1/3) scaling law.
spinodal.in1 # Spinodal decomposition: 50:50 binary LJ mixture quenched below T_c 2 # Same-species attractions are stronger than cross-species, driving 3 # spontaneous phase separation. 4 5 units lj 6 atom_style atomic 7 dimension 3 8 boundary p p p 9 10 lattice fcc 0.85 11 region box block 0 8 0 8 0 8 12 create_box 2 box 13 create_atoms 1 box 14 mass 1 1.0 15 mass 2 1.0 16 17 # Randomly relabel half the atoms as type 2 (50:50 mixture) 18 set type 1 type/fraction 2 0.5 48392 19 20 pair_style lj/cut 2.5 21 pair_coeff 1 1 1.0 1.0 2.5 # A-A attraction 22 pair_coeff 2 2 1.0 1.0 2.5 # B-B attraction 23 pair_coeff 1 2 0.5 1.0 2.5 # A-B weaker -> demixing 24 pair_modify shift yes 25 26 velocity all create 2.0 87287 dist gaussian 27 timestep 0.005 28 29 # NVT well below the consolute temperature 30 fix integ all nvt temp 0.7 0.7 0.5
Bead-spring chains with FENE bonds — the canonical coarse-grained model
The Kremer-Grest model is the foundation of computational polymer physics. 36 chains of 20 beads each interact via a purely repulsive WCA potential (shifted LJ with cutoff at 2^(1/6) sigma) and are connected by finitely extensible nonlinear elastic (FENE) bonds. Starting from straight-rod configurations, the chains rapidly randomize into a disordered melt. This model captures universal polymer dynamics — Rouse relaxation, reptation, and entanglement — and is widely used to study polymer blends, gels, and biomolecular assemblies.
polymer.in1 # Kremer-Grest polymer melt: bead-spring chains with FENE bonds. 2 # 36 chains of 20 beads each; WCA repulsion + FENE bonds. 3 4 units lj 5 atom_style bond 6 dimension 3 7 boundary p p p 8 9 read_data polymer_melt.data 10 11 # Purely repulsive Weeks-Chandler-Andersen potential 12 pair_style lj/cut 1.122462 13 pair_coeff 1 1 1.0 1.0 1.122462 14 pair_modify shift yes 15 16 # Finitely extensible nonlinear elastic bonds 17 bond_style fene 18 bond_coeff 1 30.0 1.5 1.0 1.0 19 special_bonds fene 20 21 velocity all create 1.0 87287 dist gaussian 22 timestep 0.005 23 24 fix integ all nvt temp 1.0 1.0 0.5
Two-phase coexistence with surface tension and capillary fluctuations
A dense Lennard-Jones liquid slab is placed in the center of an elongated simulation box with vacuum above and below. At T=0.85 (well below the critical temperature T_c ~ 1.3), the system maintains stable liquid-vapor coexistence. The pair cutoff is extended to 3.5 sigma to capture long-range interactions that determine surface tension. The interfaces exhibit thermal capillary fluctuations, and occasional atoms evaporate into the vapor phase. This geometry is the standard method for computing surface tension via the pressure tensor anisotropy: gamma = L_z/2 * (P_N - P_T).
slab.in1 # Liquid-vapor slab: dense LJ liquid sandwiched between vacuum. 2 # Long cutoff captures surface tension via pressure tensor anisotropy. 3 4 units lj 5 atom_style atomic 6 dimension 3 7 boundary p p p 8 9 lattice fcc 0.84 10 region box block 0 8 0 8 0 30 11 region slab block 0 8 0 8 10 20 12 create_box 1 box 13 create_atoms 1 region slab 14 mass 1 1.0 15 16 pair_style lj/cut 3.5 17 pair_coeff 1 1 1.0 1.0 18 19 velocity all create 0.85 87287 dist gaussian 20 timestep 0.005 21 22 # Stable liquid-vapor coexistence at T=0.85 (well below T_c ~ 1.3) 23 fix integ all nvt temp 0.85 0.85 0.5
Two crystalline clusters merge under surface energy minimization
Two spherical Lennard-Jones nanoparticles (radius ~ 5 sigma, ~500 atoms each) are placed with a small gap between them. At moderate temperature (T=0.4), surface atoms diffuse across the gap, forming a neck that grows over time as the system minimizes its total surface energy. This sintering process is fundamental to powder metallurgy, nanoparticle synthesis, and additive manufacturing. The shrinking boundary box visible in the viewer is an artifact of the non-periodic boundaries (shrink-wrapped).
sinter.in1 # Nanoparticle sintering: two crystalline LJ clusters merging. 2 # Shrink-wrapped boundary; surface diffusion forms a neck over time. 3 4 units lj 5 atom_style atomic 6 dimension 3 7 boundary s s s 8 9 lattice fcc 1.0 10 region box block -2 24 -2 24 -2 24 11 create_box 1 box 12 13 # Two spherical clusters separated by a small gap 14 region sphere1 sphere 7 11 11 5 units box 15 region sphere2 sphere 17 11 11 5 units box 16 create_atoms 1 region sphere1 17 create_atoms 1 region sphere2 18 mass 1 1.0 19 20 pair_style lj/cut 2.5 21 pair_coeff 1 1 1.0 1.0 22 pair_modify shift yes 23 24 velocity all create 0.4 87287 dist gaussian 25 timestep 0.005 26 27 fix integ all nvt temp 0.4 0.4 0.5