Overview¶
The aim of Synference is to provide a flexible, modular, and user-friendly framework for performing SBI-based SED fitting. Synference leverages the Synthesizer package to generate synthetic observables and the LtU-ILI package to perform fast, amortized posterior inference.
Structure¶
The Synference package is split into two main components:
Library Generation: Tools to generate libraries of synthetic observables using the Synthesizer package.
SBI Training and Inference: Tools to train SBI models using the LtU-ILI package, and to perform posterior inference on observed data.
The flowchart below illustrates how the different components of Synference, such as defining observables and creating training arrays, interact: