Multi Mode Source Extraction

Pushing the limits of SExtractor

This repository contains all of the work I did during my semester project in the spring semester of 2018, as part of my Physics Master ETH degree. For this I travelled to the Harvard-Smithsonian Center of Astrophysics in Cambridge, MA, to work with Sandro Tacchella¹, Daniel Eisenstein¹ and Ben Johnson¹ on source extraction methods suited for the upcoming JWST deep imaging data. Alexandre Refregier² supervised my work from ETH.

Content

• data folder contains:
  • HST extreme deep field fits files for various bands
  • 3D-HST catalog
  • a H band PSF
  • a pickled version of the final catalog
• *.ipynb files are Jupyter notebooks that document the research
• mmse.py contains the final MultiModeSourceExtractor class

Getting started

Before running any notebooks, download the HST XDF data

cd ./data/
wget $(cat hlsp_xdf_hst_download_v1.txt)

Jupyter notebooks

• Overshredding: analysis on SExtractor's parameters and how to get it to detect sub-structures
• Kernels: comparison of the detections obtained using different filtering kernels
• Asymmetries: analysis on the difference between peak value and barycentre of detected objects in order to set priors on final positions
• Multi Mode Source Extractor: example on how to use MMSE using H and I bands of HST XDF images
• Stacking: example on how to increase detection efficiency by stacking multiple H bands together

To see the notebooks, run jupyter notebook from the root directory of the project.

Slides

The Slides notebook contains a short presentation about my work. To see the presentation run

jupyter nbconvert Slides.ipynb --to slides --post serve

from the root directory of the project or run it as a normal notebook and start it with the RISE extension.

---

1. Harvard-Smithsonian Center of Astrophysics, Cambridge MA
2. Institute for Particle Physics and Astrophysics, ETH Zürich