Skip to content

Simulation Workflows

All simulation workflows use LAMMPS as the MD engine via lammpsparser. Each workflow handles the complete lifecycle: setting up the simulation, running LAMMPS in a temporary directory, parsing outputs, and returning structured results.

Overview

Workflow Function Purpose
Melt-Quench melt_quench_simulation() Generate realistic glass structures from random initial configs
Molecular Dynamics md_simulation() Run NVT/NPT equilibration or production simulations
Elastic Moduli elastic_simulation() Calculate \(C_{11}\), \(C_{12}\), \(C_{44}\) via stress-strain
Viscosity viscosity_simulation() Compute viscosity via Green-Kubo (SACF)
CTE cte_simulation() Coefficient of thermal expansion from NPT

Common Architecture

All workflows share a common execution pattern:

graph LR
    A[Structure + Potential] --> B[Build LAMMPS input]
    B --> C[Execute LAMMPS]
    C --> D[Parse output]
    D --> E[Return results]
  1. Input: ASE Atoms object + potential DataFrame
  2. Build: lammpsparser generates LAMMPS input scripts and data files
  3. Execute: LAMMPS runs in a temporary directory via subprocess
  4. Parse: Output files (dump, log, custom) are read back
  5. Return: Structured dictionary with results + final structure

LAMMPS Command

The LAMMPS executable path is determined by get_lammps_command():

from amorphouspy.workflows.shared import get_lammps_command

cmd = get_lammps_command()
# Returns: "lmp_mpi" (or custom path from environment)

By default, it looks for lmp_mpi on the system PATH. Set the LAMMPS_COMMAND environment variable to override.

Detailed Guides

Each workflow has a dedicated page with full parameter documentation, implementation details, and examples: