analysis-scripts
Write and run Python scripts to analyze quantum experiment data stored in HDF5 files. Use when the user asks to analyze experiment results, fit peaks or curves, extract features from measurement arrays, or when reusable analysis logic should be saved alongside an experiment for future reuse.
ソース情報
- リポジトリ
- NVIDIA/Quantum-Calibration-Agent-Blueprint
- ソースの最終更新活動
- 2026年4月20日 05:20
- 検出された SKILL.md の言語
- 英語
- スター
- 74
- フォーク
- 28
インストール方法
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
ソースファイルを確認
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
SKILL.md を表示中
SKILL.md
ソースの指示 · 読み取り専用プレビュー- name
- analysis-scripts
- description
- Write and run Python scripts to analyze quantum experiment data stored in HDF5 files. Use when the user asks to analyze experiment results, fit peaks or curves, extract features from measurement arrays, or when reusable analysis logic should be saved alongside an experiment for future reuse.
# Analysis Scripts
Write Python scripts to analyze experiment data. Scripts are saved alongside the experiment for reuse.
## Storage Location
Scripts go in a subfolder next to the experiment's HDF5 file:
```
data/experiments/
├── YYYYMMDD_HHMMSS_type.h5 # Experiment data
└── YYYYMMDD_HHMMSS_type_scripts/ # Scripts for this experiment
└── analysis_name.py
```
To find the scripts folder:
1. Get experiment: `lab(action="history_show", experiment_id="...")`
2. Response includes `file_path` (e.g., `.../20240315_143022_qubit_spectroscopy.h5`)
3. Scripts folder: replace `.h5` with `_scripts/`
## Script Template
```python
#!/usr/bin/env python
"""Description of what this script does."""
import sys
import json
from pathlib import Path
# Import core storage
from core import storage
DATA_DIR = Path(__file__).parent.parent.parent # data/experiments/
# Get experiment ID from command line
experiment_id = sys.argv[1]
# Load experiment data
exp = storage.load_experiment(experiment_id, DATA_DIR)
if exp is None:
print(json.dumps({"error": f"Experiment {experiment_id} not found"}))
sys.exit(1)
data = exp.to_dict()
# =============================================================================
# Your analysis code here
# =============================================================================
import numpy as np
# Available data:
# data['params'] - input parameters dict
# data['results'] - scalar results dict
# data['arrays'] - numerical arrays dict (e.g., frequencies, amplitudes)
# data['plots'] - plot data list
# Example:
# frequencies = np.array(data['arrays']['frequencies'])
# amplitudes = np.array(data['arrays']['amplitudes'])
result = {
"status": "success",
# your results here
}
print(json.dumps(result))
```
## Running a Script
```bash
python /path/to/script.py <experiment_id>
```
Example:
```bash
python data/experiments/20240315_143022_qubit_spectroscopy_scripts/fit_peak.py 20240315_143022_qubit_spectroscopy
```
## Workflow
1. User asks to analyze an experiment
2. Get experiment info: `lab(action="history_show", experiment_id="...")`
3. Check for existing scripts: `ls {file_path.replace('.h5', '_scripts/')}`
4. If script exists: run it
5. If not: write new script using template, then run it
## Finding Scripts from Similar Experiments
```bash
# Find experiments of same type
lab(action="history_list", filter_type="qubit_spectroscopy")
# Check each for scripts
ls data/experiments/20240315_143022_qubit_spectroscopy_scripts/
# Read a script to copy/adapt
cat data/experiments/.../fit_peak.py
```
GitHubで見る