Group-4 Microlensing Filter using AMPEL

[Meet-Vyas-Dev]

🔧 Action Plan for Ampel Integration and Microlensing Filter Development

📚 Resources

• 🌐 Ampel Main Website
• 📖 Ampel-core Documentation
• 🧪 Ampel Sample Contribution Template

---

✅ Objectives

1. Thorough Documentation Review

   • Go through the Ampel documentation in detail.
   • Understand the framework architecture, especially T0 modules and filtering logic.

2. Fork and Modify Template

   • Fork the Ampel-contrib-sample repository.
   • Update the T0 module to enable ingestion of LSST alert streams.

3. Develop Initial Microlensing Filter

   • Create a filter module for detecting microlensing candidates.
   • Use 5-sigma SNR magnitude as an initial selection criterion.

4. Benchmarking and Collaborative Development

   • Collaborate with other teams to:
     • Incorporate Rubin OpSim and additional datasets.
     • Validate and benchmark filter performance against simulated alerts.

5. (Optional) Advanced Statistical Filtering

   • Enhance the filter by integrating Simulation-Based Inference (SBI).
   • Use Normalizing Flows for parameter estimation and model fitting of microlensing events.

---

📌 Notes

• Keep modularity in mind for easy integration into future pipelines.
• Maintain documentation of filter logic and T0 ingestion steps.

[Meet-Vyas-Dev]

Progress Update: AMPEL & Microlensing Pipeline

Resources

• RTModel GitHub Repository: https://github.com/valboz/RTModel
• RTModel Paper (A&A, 2024): https://www.aanda.org/articles/aa/pdf/2024/08/aa50450-24.pdf

Codebase Review

• Completed a thorough review of the AMPEL source code and associated documentation.
• Focused on understanding the framework architecture, especially the T0 modules and real-time filtering logic.

Pipeline Development

• Developed a lightweight pipeline to convert microlensing simulations (from the sprint) generated using Rubin OpSim into a format compatible with the RT Model.
• Ensured integration with AMPEL’s modular structure for scalability and reuse.

RT Model Execution

• The converted simulations are currently being processed by the RT Model.
• Model fits are in progress, and results will be analyzed upon completion.

---