Computational Materials Science Agent

An Aider configuration that transforms the AI into an autonomous research agent for computational materials science. Pre-loads context files for molecular dynamics simulations (LAMMPS), quantum chemistry (Quantum ESPRESSO), HPC cluster management, literature search, materials databases, and data analysis.

What This Does

When you start Aider with this configuration, it automatically loads six domain-specific skill files covering the full materials science research workflow:

The agent understands materials science concepts, simulation tools, and research workflows without needing to be re-prompted.

Instructions

1. **Copy the configuration to your project root:**

```bash

cp configs/aider/.aider.conf.yml .aider.conf.yml

```

2. **Required project structure:**

Your project must have these context files at the specified paths:

```

AGENTS.md # Agent capabilities and goals

skills/lammps-simulation/SKILL.md # LAMMPS molecular dynamics

skills/quantum-espresso/SKILL.md # DFT quantum chemistry

skills/hpc-cluster/SKILL.md # HPC job management

skills/literature-search/SKILL.md # Research paper discovery

skills/materials-database/SKILL.md # Materials data queries

skills/data-analysis/SKILL.md # Simulation data analysis

```

3. **Choose your model:**

Uncomment one of the model lines in `.aider.conf.yml`:

```yaml

model: claude-3-5-sonnet # Best for reasoning (recommended)

# model: gpt-4o # Balanced performance

# model: deepseek/deepseek-coder # Code-focused, cost-effective

```

4. **Optional settings:**

- **Edit format** — Uncomment `edit-format: diff` to see unified diffs instead of full file rewrites

- **Auto-commits** — Set `auto-commits: true` to automatically commit changes (disabled by default for benchmarking workflows)

- **Git dirty commits** — Uncomment `dirty-commits: true` to allow commits with unstaged changes

- **Verbose output** — Uncomment `verbose: true` for detailed logging

5. **Start Aider:**

```bash

aider

```

The agent will load all skill files and be ready to assist with materials science research tasks.

Example Workflows

**Run a LAMMPS simulation:**

```

> Set up a LAMMPS NPT simulation for silicon carbide at 1500K

```

**Query a materials database:**

```

> Find all perovskites with bandgap between 1.5-2.0 eV from Materials Project

```

**Analyze simulation output:**

```

> Parse the LAMMPS trajectory, calculate radial distribution function, and plot

```

**Submit HPC job:**

```

> Create a SLURM script for 128-core Quantum ESPRESSO job with 48hr walltime

```

Constraints

Tips