Kilonova Filter & Classifier
Develop filters and machine-learning classification for kilonova candidates in Rubin alerts.
The kilonova filtering project develops candidate-selection and classification workflows using simulated light curves, colors and alert-image cutouts. Current work builds the training framework and model features, with planned integration into Rubin alert brokers and real-time analysis.
Maturity
Beta
Access
Public
No portal sign-in is required.
Maintainers
Ketan Sand · Mohammed Chamma
What it supports
Use cases
- Kilonova candidate selection
- Simulated training data
- Light-curve classification
- Alert-broker integration
Where it runs
Runtime homes
A simulated light-curve dataset exists. A deployed classifier or operational broker filter is a future milestone.
Runtime, inputs & outputs
- Distributions
- Runtime
- not-applicable
- Access
- public
- Inputs
- Not specified
- Outputs
- Not specified
Names & abbreviations
- CADC
- Canadian Astronomy Data Centre
- CanDIAPL
- Canadian Data Intensive Astrophysics Platform
Connected work
Related science context
Multi-messenger listeners and classifiersGala ingestion, common event schemas, transient-dashboard metrics and kilonova filtering are being developed with shared database and analysis interfaces.
Alert simulation, storage and monitoringLASS provides public alert simulation; FASTDB benchmarking, transient-dashboard development and kilonova-classifier training support the connected alert workflow.