skills.rest
learningmatter-mit
learningmatter-mitOFFICIAL
Explore all skills by this author →
WebsiteGitHub

mat-lammps-md

Builds and runs LAMMPS MD with MLIP-specific isolated binaries and environments.

144|21|Updated Jan 8, 2026
Search Tags:#mace#molecular dynamics#lammps#mlip#matgl#kokkos#gpu compute capability
One-click install
npx skills add https://github.com/learningmatter-mit/AtomisticSkills --skill mat-lammps-md
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mat-lammps-md
Source: https://github.com/learningmatter-mit/AtomisticSkills/tree/main/.agents/skills/mat-lammps-md
Command: npx skills add https://github.com/learningmatter-mit/AtomisticSkills --skill mat-lammps-md

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and assets (resource) components.

What problem does it solve?

This Skill prevents brittle, conflict-prone setup when running GPU-accelerated LAMMPS molecular dynamics using MLIP backends (MACE, MatGL/CHGNet, FairChem), by enforcing isolated build/runtime environments per backend.

Core Features & Use Cases

  • MLIP-isolated LAMMPS binaries: builds a backend-matched lmp executable using an isolated conda environment to avoid Python/Torch/stack conflicts.
  • GPU/Kokkos compatibility: detects GPU compute capability and selects the correct KOKKOS_ARCH_* flag for stable GPU performance.
  • Backend-correct execution: ensures strict binary–environment pairing so you never cross-run MACE/MatGL/FairChem binaries in the wrong runtime env.
  • Use case: run a repeatable melt-quench MD protocol (via the provided MACE example) or perform an MLIP-based adsorption/scan study using the backend-specific workflow and outputs.

Quick Start

Run a backend-matched LAMMPS MLIP molecular dynamics build by selecting the MLIP backend from the foundation-potential guide, building the corresponding isolated LAMMPS binary with the matching Kokkos arch flag, and then running that same binary inside its matching conda environment.

Frequently Asked Questions about mat-lammps-md

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run LAMMPS molecular dynamics with MACE without Python or Torch stack conflicts?▼

To run LAMMPS molecular dynamics with MACE conflict-free, this Skill builds an isolated MLIP-specific conda environment and a backend-matched lmp executable. This strict binary-environment pairing prevents the Python, Torch, and stack conflicts that typically break MLIP MD workflows.

How do I set the correct Kokkos architecture flag for GPU-accelerated LAMMPS MLIP molecular dynamics?▼

For GPU-accelerated LAMMPS MLIP molecular dynamics, the Skill detects your GPU compute capability and automatically selects the matching KOKKOS_ARCH_* flag. This ensures stable GPU performance and compatibility during the package compilation step.

Can I use MatGL or FairChem backends for atomistic simulations in LAMMPS, or is it limited to MACE?▼

You can use MatGL/CHGNet and FairChem backends for atomistic simulations alongside MACE. The Skill enforces backend-isolated binaries and environments, ensuring you never cross-run MACE, MatGL, or FairChem LAMMPS executables in the wrong runtime environment.

Why does my LAMMPS MLIP simulation fail when switching between different machine learning interatomic potentials?▼

LAMMPS MLIP simulations fail when switching potentials because standard setups lack environment isolation, causing runtime conflicts. This Skill solves the issue by building a separate, backend-matched conda environment for each MLIP backend to ensure correct execution.

What is the best way to run a repeatable melt-quench MD protocol using machine learning interatomic potentials?▼

The best way to run a repeatable melt-quench MD protocol using machine learning interatomic potentials is to use the provided MACE example within the Skill. It leverages an isolated LAMMPS binary and a paired conda environment to guarantee reproducibility.

Do I need a specific GPU compute capability to run Kokkos-accelerated LAMMPS with MLIP backends?▼

Yes, running Kokkos-accelerated LAMMPS with MLIP backends requires a compatible GPU compute capability. The Skill uses your specific GPU compute capability to determine and apply the correct Kokkos architecture flag during the LAMMPS package compilation.

Related Skills

Explore More Education & Research Skills→
yo-stevenyo-steven

exec-local-compile

Compile TensorRT-LLM from source into a Python package inside a Docker container.

Community
Intermediate
learningmatter-mitlearningmatter-mit
144

mat-amorphization

Generate amorphous atomic structures via melt-quench molecular dynamics from crystalline inputs.

Official
Advanced
learningmatter-mitlearningmatter-mit
144

ml-mlip-speed

Benchmark MLIP inference speed and memory across NaCl supercell sizes.

Official
Intermediate
tondevreltondevrel
19

ase

Build, manipulate, and analyze atomic structures for atomistic simulations.

Community
Advanced