dft-qe-workflow

Files (1)

• SKILL.md

SKILL.md (first lines)

# DFT QE Workflow

## Procedure

1. Confirm material, lattice, reference lattice constant, pseudopotential family,
   exchange-correlation functional, and whether the task is dry-run or live HPC.
2. Create a conventional-cell QE input using `ibrav=0`; avoid mixing `celldm`
   with Angstrom cell declarations.
3. Find pseudopotentials with `matsci_find_pseudopotential`.
4. For lattice constants, run one convergence grid first, then select
   `kspacing` and `ecutwfc` with `matsci_select_convergence`.
5. Generate EOS scale-factor inputs using the selected parameters.
6. Fit the EOS with `matsci_fit_lattice_constant` and report the result plus
   comparison to the reference value.

## Rules

- Electron `conv_thr` defaults to `1e-6`.
- Use smearing appropriate for metallic systems.
- Do not generate convergence tests for every structure unless explicitly asked.
- Record key decisions in `matsci_state`.

## Verification

Run `matsci_run_harness` with suite `sol27lc` for a deterministic dry-run check.

From chouyoudou/matsci-mcp, MIT.

More from chouyoudou/matsci-mcp