Demonstrates a normalizing flow approach to estimate star cluster properties from broadband photometry, highlighting its efficiency in challenging scenarios.
Key Points
The aim is to improve the inference of star cluster properties like age, mass, and reddening from unresolved broadband photometry using normalizing flows.
Developed a dataset of synthetic photometric observations using CIGALE for mock star clusters.
Trained a conditional invertible neural network to predict posterior distributions for cluster parameters.
Evaluated network performance against the PHANGS Data Release 3 catalog.
Successfully predicted cluster parameters for the PHANGS catalog.
Findings show reasonable agreement between the network estimates and PHANGS data.
Demonstrated the usefulness of normalizing flow methods for efficient density estimation in complex scenarios.