dpdata — конвертация данных квантово-химических расчётов

★ 7.5 · general

dpdata is a Claude Code skill that converts computational chemistry data between VASP, QE, CP2K, Gaussian, LAMMPS, and DeePMD formats using the dpdata Python library. It supports reading and writing over 10 format keys — including `vasp/outcar`, `deepmd/npy`, `lammps/dump`, and `extxyz` — and provides ready-to-run scripts for the most common workflows. With this skill, Claude can merge multiple OUTCAR files into a single dataset, filter trajectory frames by maximum force to remove unconverged steps, split data into train/valid subsets at any ratio, and run CLI commands like `dpdata convert` and `dpdata info`. It is essential for preparing DFT-based training data for machine learning interatomic potentials. Requires the `dpdata` package (`pip install dpdata`).