--- name: vasp description: Skills for VASP DFT calculations — input preparation, remote submission to Bohrium, and result collection. metadata: tools: - run_bash - run_skill_script - load_skill_resource dependent_skills: - bohrium - dpdisp tags: - vasp - dft - relaxation - scf - band-structure - dos - bohrium --- # VASP DFT Skill One script handles VASP-specific work; job submission is now delegated to the `dpdisp-submit` skill (DPDispatcher), which supports both Bohrium and standard Slurm/HPC clusters: | Script | Role | |---|---| | `vasp_tools.py` | Prepare input files; collect / read results Use the `run_skill_script` tool to execute it: - `skill_name`: `"vasp"` - `script_name`: `"vasp_prepare.py"` - `args`: the sub-command and flags as a single string Every command prints JSON to stdout and exits 0 on success, 1 on error. --- ## Mandatory workflow sequence 1. **Obtain a structure** — if the user has not supplied one, generate it first. 2. **Prepare inputs** — run the appropriate `vasp_tools.py prepare_*` command. 3. **Submit jobs** — pass the returned `calc_dir_list` to the `dpdisp-submit` skill by generating a `submission.json`. 4. **Collect / read results** — after the job finishes, run `collect_results` or `read_results`. Run exactly **one property step at a time**. Do not chain relaxation + SCF in a single step. --- ### prepare_relaxation Prepare structural relaxation (IBRION=2, ISIF=3, NSW=200). ``` run_skill_script(skill_name="vasp", script_name="vasp_tools.py", args="prepare_relaxation \ --structure \ [--frames 0 1 2] \ [--kpoints NX NY NZ] \ [--incar_tags '{\"ENCUT\": 600}'] \ [--potcar_map '{\"Bi\": \"Bi_d\"}']") ``` - `--structure`: structure file to read. Supported formats: - `*.extxyz` — may contain multiple frames; all frames are processed by default. - `*.vasp` (POSCAR format) — always contains exactly one frame. - `--frames`: integer indices to process (extxyz only); default is all frames. - `--kpoints`: explicit Gamma-centred mesh. Default: auto KPPRA density 40. --- ### prepare_scf Prepare a self-consistent field calculation (NSW=0, IBRION=-1). ``` run_skill_script(skill_name="vasp", script_name="vasp_tools.py", args="prepare_scf \ --structure \ [--frames 0 1 2] \ [--kpoints NX NY NZ] \ [--soc] \ [--incar_tags '{\"ENCUT\": 600}']") ``` - `--structure`: structure file to read. Supported formats: - `*.extxyz` — may contain multiple frames; all frames are processed by default. - `*.vasp` (POSCAR format) — always contains exactly one frame. - `--frames`: integer indices to process (extxyz only); default is all frames. - `--soc`: use `scf_soc` INCAR preset (ISPIN=2, LSORBIT=.TRUE.); default: `scf_nsoc`. - Default k-mesh: auto KPPRA density 40. - SCF outputs CHGCAR and WAVECAR when `LCHARG=true` / `LWAVE=true` (SOC preset only by default). --- ### prepare_nscf_kpath Prepare a non-self-consistent band-structure calculation along a k-path. Requires completed SCF directories that contain `CONTCAR` and `CHGCAR`. ``` run_skill_script(skill_name="vasp", script_name="vasp_tools.py", args="prepare_nscf_kpath \ --scf_dirs [ ...] \ [--kpath GMKG] \ [--n_kpoints 16] \ [--soc] \ [--incar_tags '{\"NBANDS\": 48}']") ``` - `--kpath`: explicit path string (e.g. `GMKG`). Default: auto from pymatgen `HighSymmKpath`. - `--n_kpoints`: points per segment (default 16). - Copies `CHGCAR` (and `WAVECAR` if present) from the SCF directory. --- ### prepare_nscf_uniform Prepare a non-self-consistent uniform-mesh calculation (for DOS). Requires completed SCF directories. ``` run_skill_script(skill_name="vasp", script_name="vasp_tools.py", args="prepare_nscf_uniform \ --scf_dirs [ ...] \ [--kpoints NX NY NZ] \ [--soc] \ [--incar_tags '{\"NEDOS\": 2000}']") ``` - Default k-mesh: auto KPPRA density 100. - Copies `CHGCAR` (and `WAVECAR` if present) from the SCF directory. --- ### collect_results Parse `OUTCAR` files and write all frames into a single extxyz (via dpdata). ``` run_skill_script(skill_name="vasp", script_name="vasp_tools.py", args="collect_results \ --dirs [ ...]") ``` Returns: ```json { "status": "success", "scf_result": "" } ``` --- ### read_results Read `vasprun.xml` / `OUTCAR` and return key scalar results as JSON. ``` run_skill_script(skill_name="vasp", script_name="vasp_tools.py", args="read_results \ --calc_type \ --calc_dir ") ``` | calc_type | Returned fields | |---|---| | `relaxation` | `structure`, `total_energy`, `max_force`, `stress`, `ionic_steps` | | `scf` | `structure`, `total_energy`, `efermi`, `band_gap`, `is_metal` | | `nscf` | `structure`, `efermi`, `is_metal`, `band_gap`, `cbm`, `vbm` | --- ## Submission ### `bohrium` skill (Recommended for Bohrium users) Submit jobs to Bohrium using the `bohrium` skill, which wraps the `bohr` CLI. This is the recommended submission method for users running on the Bohrium platform. ### `dpdisp` skill (Not recommended for Bohrium users) Submission is handled by the `dpdisp` skill (DPDispatcher), which supports Both Bohrium and standard Slurm/HPC clusters. See the `dpdisp` skill documentation for full details and schema. For VASP job submission on Bohrium platform using DPDispatcher, see: ``` load_skill_resource(skill_name="vasp", path="references/bohrium-submission.md") ``` --- ## Configuration file (config.yaml) `config.yaml` lives in the Skill references directory. It controls: - `work_dir` — where all calculation subdirectories are created (default `"vasp"`, resolved relative to the session directory). - `VASP_default_INCAR` — one sub-key per preset (`relaxation`, `scf_nsoc`, `scf_soc`, `nscf_nsoc`, `nscf_soc`). You can check the default INCAR settings by reading `config.yaml` with `load_skill_resource`: ``` load_skill_resource(skill_name="vasp", path="references/config.yaml") ``` To override individual tags for a single run without editing the file, use `--incar_tags` when running `prepare_*` sub-commands of the `vasp_tools.py` script. ```bash --incar_tags '{"ENCUT": 600, "EDIFF": 1e-6}' ```