AstroFlow Manual

AstroFlow is a GPU-accelerated single-pulse and fast radio burst search pipeline for FILTERBANK (.fil) and PSRFITS (.fits) observations. It combines CUDA dedispersion, RFI mitigation, AI-assisted candidate detection, and candidate plotting behind a YAML-driven command-line workflow.

This is the official AstroFlow manual. The online documentation is published at lintian233.github.io/astroflow, and the latest source version can be found at github.com/lintian233/astroflow.

Contents

• Installation
  • Requirements
  • Install From PyPI
  • Install a Pre-Release Build
  • Build From Source
  • First-Run Model Download
• Quick Start: Run astroflow on FRB180417
  • 1) Get the test file
  • 2) Create a run config
  • 3) Run
  • 4) View the results
  • 5) Troubleshooting
• YAML Configuration Reference
  • Minimal Single-File Example
  • Top-Level Options
  • Processing And GPU Options
  • Detection Options
  • Search Grid Options
  • RFI Options
  • Plotting Options
  • Mode Behavior
  • Output Layout
• Processing the FAST PREX FRB Dataset with AstroFlow
  • Prerequisites
  • Step 1: Download the Dataset
  • Step 2: Prepare Auxiliary Files
  • Step 3: Configure the Search Pipeline
  • Step 4: Execute the Pipeline
  • Step 5: Analyze the Output
• Dataset and Model
  • Dataset
  • Model

Overview

• CUDA-accelerated dedispersion and search over configurable DM, time, and frequency grids.

• AI candidate detection with the bundled yolov11n model or a user-provided model path.

• RFI mitigation through external masks, zero-DM filtering, and GPU-accelerated IQRM.

• Batch, directory, monitor, and dataset-validation modes.

• Candidate plots, logs, and optional CSV candidate summaries.

• Public training dataset and separately versioned model checkpoints.

Citation

If AstroFlow is useful in your research, cite:

ASTROFLOW: A Real-Time End-to-End Pipeline for Radio Single-Pulse Searches
https://doi.org/10.3847/1538-4365/ae4a26

Project Links

• Documentation: https://lintian233.github.io/astroflow/

• Source: https://github.com/lintian233/astroflow

• Issues: https://github.com/lintian233/astroflow/issues