Repository files navigation

We use convolutional neural networks (CNNs or convnets) to predictgalaxy properties using Sloan Digital Sky Survey (SDSS)gri images.Gas-phase metallicity, which is often estimated by using opticalspectroscopy, can also be estimated using our CNN.

We describe our methods in a paper: Wu & Boada (2019, MNRAS, 484, 4683;arXiv:1810.12913).

See also asimplified demo of our work.

• Usage
  • Dependencies
  • Data sets
  • Training and testing
• Citation

Download this repository by running

git clone https://github.com/jwuphysics/galaxy-cnns.gitcd galaxy-cnns

All analysis was performed inside the Jupyter notebooks using a Python3 environment.We use version 0.7.0 of thefastai machinelearning framework built atopPytorch. This can be installedby following the instructions on the Fastai README page. We will soon have workingexamples for Fastai version 1.0, which can be installed by running:

git clone https://github.com/fastai/fastai.gitcd fastai conda env create -f environment.yml

Note that you will need a GPU. If you don't have one, substitute the previous last linewith this instead:

conda env create -f environment-cpu.yml

Before executing any code (or running any notebooks), enter the environmentby runningconda activate fastai (orconda activate fastai-cpu).

If you encounter any errors, please feel free to reach out to me (@jwuphysics)or checkthis poston the fastai forums.

We queried theSDSS DR14 image cutout serviceusing the script./download_images.py in order to obtaingri images.

We queried theSDSS MPA-JHU DR7 catalogof spectral line and derived galaxy properties using the commands in theSQL script,./SDSS_sql_query.sql.

To run the our notebooks, make sure that you are in thefastai conda environment first,and then runjupyter notebook and enter the./notebook directory.

If you wish to reproduce all figures from the paper, run the notebooks in the./notebook/paper directory. You will first need to have downloaded all of thedata and executed the notebooks labeled"06. Predicting stellar mass in addition to metallicity.ipynb" and"10. The effects of resolution.ipynb" first (sorry about this disorganization -- thismay be cleaned up in a future update).

If you would like to reference our paper,please use the following citation, produced byNASA ADS:

@ARTICLE{2019MNRAS.484.4683W,       author = {{Wu}, John F. and {Boada}, Steven},        title = "{Using convolutional neural networks to predict galaxy metallicity from three-colour images}",      journal = {Monthly Notices of the Royal Astronomical Society},         year = "2019",        month = "Apr",       volume = {484},       number = {4},        pages = {4683-4694},          doi = {10.1093/mnras/stz333},archivePrefix = {arXiv},       eprint = {1810.12913}, primaryClass = {astro-ph.GA} }