import cuphoton
print(cuphoton.__version__)0.3.0.dev98+gc04d1a5cuPhoton User Guide
Install cuPhoton, run your first image difference, then follow task guides for each module.
Welcome
cuPhoton provides Python libraries and command-line workflows for GPU-accelerated imaging and signal analysis. Its workflows include image alignment and subtraction, source measurement, candidate classification, reconstruction, X-ray detector analysis, and signal correlation. This guide starts with time-domain astronomy: finding what changed between two images of the sky.
This guide shows you how to use cuPhoton. Start with the first three pages, then go to the guide for your task.
Get started
- 1 Install cuPhoton: clone the repository and pick a CPU or GPU profile.
- 2 Your first image difference: a five-minute tutorial. Make a sample image pair, then align, subtract, detect and measure what changed.
- 3 Concepts: the ideas every module shares.
Find your task
| I want to | Module | Guide |
|---|---|---|
| Read FITS images, also straight to GPU memory | Core, xDataReader | 4 Read FITS images |
| Put one image on another’s pixel grid, or stack images | xRep | 5 Align images |
| Match two images’ blur and subtract them | xPois | 6 Subtract two images |
| Measure sources and dipoles in many cutouts at once | xFit | 7 Measure candidates |
| Score candidates as real or bogus with a neural network | xScan | 8 Classify candidates |
| Run the shipped end-to-end pipeline on my own images | all of the above | 9 Run the transient chain |
| Propagate uncertainty and check error bars on my data | xPois, xFit, the transient chain | 10 Uncertainty Quantification |
| Reconstruct one sharp image from many frames | xImage | 11 Reconstruct from many frames |
| Analyze time-resolved X-ray detector data | xRay | 12 Analyze X-ray detector traces |
| Correlate two sampled signals, or radio recordings | xBeam | 13 Correlate signals |
| Run work on many GPUs or nodes | Core executors | 14 Scale out to many GPUs |
| Record what a run measured and review it in a dashboard | xLens | 15 Monitor runs |
Appendix A — Reference lists where to find the full API, CLI and data-format documentation in the repository.
About this guide
The executable Python examples use small synthetic inputs that each page generates. Run a page’s cells in order in the environment described in 1 Install cuPhoton. Commands that use your own files, a GPU, native extensions, or distributed launchers need the prerequisites and inputs described on their page.
The documentation build executes the CPU examples. GPU, native, and distributed command examples are instructions for those environments, rather than results of that build. Printed values and figures demonstrate the APIs; they do not establish performance or scientific accuracy on your data. 10 Uncertainty Quantification describes checks for your own noise model and measurements.
This copy was rendered with: