hydra_tod

Bayesian calibration and map-making for radio intensity mapping experiments

hydra_tod implements joint Bayesian inference of sky temperature maps, instrument gains, and correlated noise parameters from radio telescope time-ordered data (TOD). It uses Gibbs sampling with MPI parallelization to efficiently handle multi-receiver, multi-scan datasets.

Features

• Flicker noise modeling: 1/f^alpha noise covariance via incomplete gamma functions, with polynomial emulators for fast evaluation
• Gibbs sampler: Joint calibration and map-making alternating between sky, gain, and noise parameter updates
• Multiple gain models: Linear, log-linear, and factorized parameterizations
• MPI parallelization: Distributed iterative linear solvers (conjugate gradient, MINRES) for large-scale problems
• Flexible noise sampling: MCMC (emcee) and NUTS (NumPyro) samplers for non-Gaussian noise posteriors
• Telescope simulation: Realistic MeerKAT scan simulations with sky models (GSM + point sources)
• Diagnostics: MCMC convergence diagnostics (ESS, R-hat), posterior visualization, sky map reconstruction

Data Model

The package models radio telescope observations as:

TOD = Tsys * (1 + n) * g

where:

• Tsys = Tsky + Tloc is the system temperature (sky + local components)
• n ~ 1/f^alpha is correlated flicker noise
• g is the time-varying instrument gain

Installation

# Basic installation
pip install hydra-tod

# With MPI support
pip install hydra-tod[mpi]

# With all optional dependencies
pip install hydra-tod[all]

# Development installation
git clone https://github.com/zzhang0123/flicker.git
cd flicker
pip install -e ".[dev]"

Quick Start

from hydra_tod.simulation import TODSimulation

# Create a simulated observation
sim = TODSimulation(
    nside=64,
    elevation=41.5,
    freq=750,
    alpha=2.0,
    ptsrc_path="gleam_nside512_K_allsky_408MHz.npy",
)

# Access simulated data
print(f"TOD shape: {sim.TOD_setting.shape}")
print(f"Sky pixels: {len(sim.pixel_indices)}")

Citation

If you use hydra_tod in your research, please cite:

@article{zhang2026joint,
  title={Joint Bayesian calibration and map-making for intensity mapping experiments},
  author={Zhang, Zheng and Bull, Philip and Santos, Mario G and Nasirudin, Ainulnabilah},
  journal={RAS Techniques and Instruments},
  pages={rzag024},
  year={2026},
  publisher={Oxford University Press}
}

Documentation

Full documentation is available at hydra-tod.readthedocs.io.

License

This project is licensed under the MIT License - see the LICENSE file for details.