원클릭으로 Manus에서 모든 스킬 실행

openmx-data-processing

스타7

포크1

업데이트2026년 4월 20일 14:33

Prepare split-local OpenMX preprocessing inputs and submit recognized OpenMX processing jobs through `inspire-openmx-submit`.

설치

Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.

Manus에서 실행

출처

cy0752

cy0752/computational-materials-skills

GitHub 저장소 열기 Creator 저장소 보기

다운로드

Manus에서 실행

OpenMX Data Processing

Prepare one split-local OpenMX workdir per dataset split. Do not run train and test processing against one shared in-place interfaces/openmx directory.
Seed each split-local workdir from <skill_root>/templates/openmx/:

poscar2openmx.yaml
run_openmx.sh
graph_data_gen.yaml
dat_gen.sh
graphgen.sh

Render the copied templates into the split-local workdir. Do not edit the skill-owned templates in place.
In run_openmx.sh, only render split-specific values such as SYSTEM_NAME, INPUT_ROOT, and DATASET_ROOT. Preserve the runtime environment contract.
Use inspire-openmx-submit for the actual OpenMX submission. inspire-openmx-submit is the authority for workspace, compute group, spec, image, and HPC submission shape.
Default split-local execution policy:

default OpenMX task budget is tasks=16, cpus-per-task=2
omp_threads defaults to 2
default material parallelism is therefore 16
jobs defaults to auto-calculation from the task's total CPUs
total CPUs must come from the submitted task budget
detect total CPUs from SLURM first, then from explicit task environment variables such as OPENMX_TOTAL_CPUS or OPENMX_TASKS * OPENMX_CPUS_PER_TASK
do not infer total CPUs from the local interactive session CPU set
explicit jobs / omp_threads overrides are still allowed

Submit the split-local entry command that runs <skill_root>/scripts/data_processing.sh --workdir <openmx_split_root> [--jobs <jobs>] [--omp-threads <omp_threads>].
Use as many CPUs as practical for the split, but do not exceed the number of data points.
Return the generated .npz path for the processed split so downstream HamGNN stages can consume it.

이 저장소의 다른 Skills

같은 저장소

abacus-data-processing

cy0752/computational-materials-skills

Prepare ABACUS calculation inputs from CIF/POSCAR with default settings and generate per-structure INPUT/KPT/STRU run directories.

2026-04-207

hamgnn-inference

cy0752/computational-materials-skills

Prepare and submit HamGNN inference jobs through the canonical HamGNN Inspire submit flow.

2026-04-207

hamgnn-training

cy0752/computational-materials-skills

Prepare HamGNN training configs and run locally or submit through the canonical HamGNN Inspire flow.

2026-04-207

remote-task-submit

cy0752/computational-materials-skills

Legacy compatibility wrapper for Inspire remote job submission. Redirect to `inspire-cli-training-submit` for the canonical generic submission flow.

2026-04-207

structure-openmx-hamgnn-training-pipeline

cy0752/computational-materials-skills

Structure-file-driven OpenMX/ABACUS-to-HamGNN training pipeline. Use when users provide a crystal structure file such as CIF or POSCAR and want to build a perturbed CIF dataset, run DFT preprocessing (OpenMX by default, ABACUS when explicitly requested), and train a HamGNN model, stopping at the trained checkpoint.

2026-04-207

name	openmx-data-processing
description	Prepare split-local OpenMX preprocessing inputs and submit recognized OpenMX processing jobs through `inspire-openmx-submit`.

OpenMX Data Processing

Prepare one split-local OpenMX workdir per dataset split. Do not run train and test processing against one shared in-place interfaces/openmx directory.
Seed each split-local workdir from <skill_root>/templates/openmx/:

poscar2openmx.yaml
run_openmx.sh
graph_data_gen.yaml
dat_gen.sh
graphgen.sh

Render the copied templates into the split-local workdir. Do not edit the skill-owned templates in place.
In run_openmx.sh, only render split-specific values such as SYSTEM_NAME, INPUT_ROOT, and DATASET_ROOT. Preserve the runtime environment contract.
Use inspire-openmx-submit for the actual OpenMX submission. inspire-openmx-submit is the authority for workspace, compute group, spec, image, and HPC submission shape.
Default split-local execution policy:

default OpenMX task budget is tasks=16, cpus-per-task=2
omp_threads defaults to 2
default material parallelism is therefore 16
jobs defaults to auto-calculation from the task's total CPUs
total CPUs must come from the submitted task budget
detect total CPUs from SLURM first, then from explicit task environment variables such as OPENMX_TOTAL_CPUS or OPENMX_TASKS * OPENMX_CPUS_PER_TASK
do not infer total CPUs from the local interactive session CPU set
explicit jobs / omp_threads overrides are still allowed

Submit the split-local entry command that runs <skill_root>/scripts/data_processing.sh --workdir <openmx_split_root> [--jobs <jobs>] [--omp-threads <omp_threads>].
Use as many CPUs as practical for the split, but do not exceed the number of data points.
Return the generated .npz path for the processed split so downstream HamGNN stages can consume it.