DPGEN Web Monitor 0.4

A real-time web interface for monitoring DPGEN tasks.
This version is adapted for Quantum Espresso FP tasks.

Features

• 🚀 Tracks current stage via record.dpgen
• 🔍 FP task analysis: finished, running, failed, pending
• ⏱️ Time prediction from pw.out (CPU time)
• 📊 Live histogram of max_devi_f for the current active model_devi task
• 📈 Live training loss curve from lcurve.out for the current active training task
• 📚 Training History: View loss curves for all completed training tasks (cached)
• 🔄 Auto-refresh every 30 seconds for live data, every 5 minutes for the full page

Installation

Install dependencies:

pip install flask matplotlib numpy

Pull the repository.

How to Run

1. Run the application:

python app.py

2. Open in browser: http://127.0.0.1:5000 (local access only by default).

3. In the web interface:
       - Enter the path to your DPGEN process working directory.
       - Click "Set Directory".

Project Structure

DPGEN_WEB/  
├── app.py              # Main Flask application  
├── utils.py            # Helper functions for parsing and analysis  
├── templates/  
│   ├── index.html      # Main monitoring page  
│   └── train_history.html # Page for viewing completed training tasks  
├── static/  
│   ├── model_devi/     # Generated histograms for completed model_devi tasks  
│   └── train_plots/    # Cached plots for completed training tasks  
└── README.md

Notes

• Histograms & Plots:
      - Histograms for completed model_devi tasks are generated once and reused.
      - Plots for completed training tasks are generated once and cached in static/train_plots/.
      - Live plots (current training/model deviation) are generated dynamically on each refresh.
• Designed for Local Use: Do not expose this interface to public networks without proper security measures.
• No External Tracking: The application does not collect or transmit any data externally.
• Quantum Espresso Specific: The FP analysis (pw.out parsing) is tailored for Quantum Espresso. The provided param_qe.json and machine.json files are examples for running DPGEN with QE locally (1 GPU, 16 cores, system mpirun). You must adapt these parameter files for your specific task and system.

Feedback

Open an issue or pull request on GitHub if you:

• Found a bug
• Want to add a feature
• Have suggestions

Built with care for DPGEN users.