Skip to content

SHINE: SHear INference Environment

SHINE Logo

A JAX-powered framework for probabilistic shear estimation in weak gravitational lensing.

Overview

SHINE treats shear measurement as a Bayesian inverse problem. Instead of measuring ellipticities and correcting for biases, SHINE generates forward models of the sky, convolves with the instrument response, and compares to observed data to infer posterior distributions of shear parameters.

Built on JAX, it leverages automatic differentiation and JIT compilation for fast, GPU-accelerated inference using NumPyro and JAX-GalSim.

Key Features

  • JAX-powered -- automatic differentiation and JIT compilation for optimal performance
  • Probabilistic inference -- full posterior distributions via NUTS/HMC
  • Differentiable rendering -- JAX-GalSim for end-to-end gradient flow
  • GPU acceleration -- seamless GPU support for large-scale analyses
  • Config-driven -- GalSim-compatible YAML with probabilistic extensions
  • Validation pipeline -- built-in bias measurement infrastructure

Quick Start

# Install from PyPI
pip install shine-wl

# Run inference
python -m shine.main --config configs/test_run.yaml

See the Getting Started guide for a full walkthrough.

Project Status

Early Development

SHINE is under active development. APIs may change between releases.

Organization

Born at CosmoStat (CEA / CNRS), built for the astro community.

Licensed under MIT.