computational-materials-science-guidelines - SKILL.md Agent Skill

name: Computational Materials Science Guidelines description: Create, run, and analyze simulations. Manage data and utilize computing resources.

Computational Materials Science Guidelines

Overview

Computational Materials Science (CMS) is an interdisciplinary field that involves defining problems through the creation or selection of models for material systems of interest, formulating system representations (from atoms to continuum), selecting methodologies/algorithms (e.g., DFT, MD, Phase-field, or FEM) to solve problems, and implementing models on various computing systems.

Typical CMS activities include preparing inputs, managing and analyzing outputs, interpreting results, accelerating discoveries, and optimizing specific targets or processes as needed.

Capabilities

Plan: literature, data, tools
Run: select models/programs, prepare inputs, run simulations
Analyze: monitor workflows, analyze outputs

How to use

Input data: provide simulation inputs based on the selected models/programs
Analysis: perform analyses for specific scales
Reproducibility: reuse predefined workflows or define a new one if needed
Publication: collect output data and use paper template to draft a report

Input format

Atomistic scale

Structures: xyz, cif, pdb, POSCAR, data.in
Parameters: run.in, INCAR, inp, POTCAR

Mesoscale

Structures & Parameters: opi, parameters.in, txt

Macroscale

Structures & Parameters: inp, mph

Output Format

Main Outputs: log, out, xml, OUTCAR (energy, forces, stress tensors)
Trajectories: dump, xtc (structural evolution)
Electronic Distribution: WAVECAR, CHGCAR, cube (wavefunctions and charge density)
Visualization in proper formats

Example usage

"Run an EOF simulation on the perovskite KNbO3"

"List candidates of dielectric materials for TFET NVM"

"Identify the most efficient catalysts and visualize the HER process at the atomic scale"

Scripts

rag.py: query contents using RAG
run_simulations.py: main engines for running simulations
analysis.py: analyze output data to provide insights

Best practices

Always cite sources when collecting information and data
Use workflow management tools to handle environment, packages, data provenance and errors
Decide which operations to perform based on specific scales and applications
Consider a broader context for analysis, even when the simulation was performed at a certain scale
Include computing resource usage and runtime in the final report

Limitations

Require accurate materials data and correct model selection
Many-body and relativistic effects may be crucial
Some problems may be unsolvable on classical computers