This repository is part of a Multimessenger App designed to analyze data from different astronomical sources. In multimessenger astronomy, signals from various messengers—such as gravitational waves, radio waves, and electromagnetic waves—are combined to gain a more comprehensive understanding of astrophysical phenomena.
In multi-messenger astronomy, signals from different astrophysical messengers (e.g., gravitational waves and radio waves) provide complementary insights into the same cosmic events.
This module focuses on automating the end-to-end processing of radio follow-up observations for gravitational wave (GW) merger candidates, particularly binary neutron star (BNS) events.
The module performs:
-
Real-time ingestion and classification of GCN notices and circulars
-
AI-driven parsing of flux-time information from GCN circulars.
-
Privacy-enhancing, distributed fitting of radio light curves
These capabilities enable rapid and scalable integration of electromagnetic follow-ups with GW detections, facilitating timely and precise parameter estimation.
MMA_RadioWave/
├── GCNListener/ # Listens to GCN alerts and classifies circulars
├── AI_Parser/ # Extracts flux/time from radio circulars
├── FederatedFitting/ # Performs distributed light curve modeling
└── README.md
- Subscribes to GCN alerts from observatories and MMA_GravitationalWave module GW detection triggers.
- Detects potential matches between GCN alerts and GW merger events based on timestamp and metadata.
- For alerts with BNS probability > 50%, checks if it corresponds to an event of interest.
- Publishes “New GCN circular added” messages via the Octopus event fabric.
- A GCN classifier module categorizes the circular type (radio, optical, gamma-ray, etc.).
- For circulars classified as radio, applies a domain-specific AI parser.
- Extracts key observational parameters such as flux density, frequency, and observation time.
- Outputs structured metadata and saves it to a distributed datastore.
- Each parsed circular is linked to its associated GW event ID for cross-messenger correlation.
- Implements federated MCMC to fit radio afterglow light curves across distributed observation sites.
- Data never leaves the site — only posterior samples and log-likelihoods are shared.
- Supports progressive fitting, where model updates as new observations arrive.
- Produces final model parameters and credible intervals, which are published back to the event fabric.
Federated MCMC Architecture:
- A central server proposes parameters (
θ) at each iteration. - These are broadcast to data sites, which compute partial log-likelihoods using local data.
- The server aggregates the total posterior:
posterior ∝ prior + ∑ log-likelihoods - Accept/reject decisions guide parameter updates.
The Radio module enables low-latency, privacy-enhancing, and scalable radio follow-up for gravitational wave events. It:
- Reduces the time to generate radio constraints on jet structure and energetics.
- Maintains data locality while still allowing joint inference across institutions.
- Forms a critical link in the joint multimessenger analysis pipeline, feeding posterior distributions into the overlap module for combined cosmological inference.
Each subdirectory (GCNListener/, AI_Parser/, FederatedFitting/) contains modular scripts and configuration templates.
To run a typical radio follow-up analysis:
- Start the GCN listener to monitor alerts and classify circulars.
- When a radio circular is detected, use the AI parser to extract observations.
- Run federated fitting across participating sites to model the light curve.
- Feed the posterior samples into the OverlapAnalysis/ module to combine with GW posteriors.
This repo focuses on radio wave data. For gravitational wave and joint analysis, please visit Gravitational Wave Analysis Repo and GW-RW Joint Analysis Repo. Together, these repositories work within the multimessenger framework to capture and analyze various cosmic events.