Skip to content

Latest commit

 

History

History
233 lines (143 loc) · 8.7 KB

code_project.md

File metadata and controls

233 lines (143 loc) · 8.7 KB

PHYS265 Code Project (draft 2025)

See also: https://github.com/astroumd/PHYS265-spring25/blob/main/code_project.md

This Code Project is an alternative to taking the 3rd midterm exam on May 1st. In this project you will select a (python based) open source project. It is your task to evaluate this software package by using it (not reproducing it) and writing a report. Due date will be Thursday, May 2nd, at the usual 11.59pm time

You will need to learn how to install and use it, show some results and one or more figures to illustrate your findings. All of this will be summarized in a report presented in PDF. You also need to show your code example(s) how you exercised the code, and a README file to make your findings reproducable. Please name and date your report!

NOTE: We submit on github, and you will use a new folder code3 inside your existing github repository that you used for lab1 and lab2.

You are welcome to suggest your own project too. Either way, discuss with your instructor which package you choose and you may get additional guidelines. You can find some suggestions in our list below extracted from ASCL and JOSS.

Some packages may not work well on your particular computer. Some requires a lot of extra software, which make it too complex to install, or require a supercomputer. Select something simple, and always run it by your instructors for approval. We do not want you to waste time getting the package to run.

Some questions to answer in your report:

  • name of the package, and what does the package do?
  • how old is the package? does it have a geneology, i.e. what came before or after, what codes are related etc.
  • is it still maintained, and by the original authors?
  • evaluate how easy it was to install and use
  • is the source code available? (this should be a given)
  • in your example, do you use python code, or do they use an interface (e.g. web, command line tool)
  • is it pure python? or does it need accompanying C/C++/Fortran code?
  • does it install via the "standard" pip/conda, or is it more complex?
  • what is the input to the package? Just parameters, or dataset(s)?
  • what is the output of the package? Just parameters, or dataset(s)?
  • does the code provide any unit tests, regression or benchmarking?
  • how can you feel confident the code produce a reliable result? (see previous question)
  • what (main) python package(s) does it use (e.g. numpy, curve_fit, solve_ivp)
  • does the package produce figures, or are you on your own? Is matplotlib used?
  • if you prefer to use a jupyter notebook instead of a python script, that's ok.
  • if you used this code in a paper, do they give a preferred citation method?

Suggested projects

This list below is an extraction from https://ascl.net/code/search/python (a full list returns well over 600 codes). Another option would be to browse JOSS ( https://joss.theoj.org/papers/search?q=python ) the Journal of Open Source Software, also a sizeable list, but none of those are reported below.

Codes annotated with [peter] are codes that Peter uses from time to time.

  • PDRT: Photo Dissociation Region Toolbox - https://ascl.net/1102.022

    This package was written by two astronomers at UMD

  • PySpecKit: Python Spectroscopic Toolkit - https://ascl.net/1109.001 [peter]

    Fitting spectral lines

  • PyModelFit: Model-fitting Framework and GUI Tool - https://ascl.net/1109.010

    Seems only working in python2

  • PyEphem: Astronomical Ephemeris for Python - https://ascl.net/1112.014

    ok. no graphics here.

  • EzGal: A Flexible Interface for Stellar Population Synthesis Models - https://ascl.net/1208.021

    ok, but no example fits file?

  • pNbody: A python parallelized N-body reduction toolbox - https://ascl.net/1302.004

    A general "pip install pnbody" did not work, but downloading from github and using

       pip install ./pNbody

    worked though.

  • corner.py: Corner plots - https://ascl.net/1702.002

    Together with emcee a very popular package in the literature. Explores the covariant matrix graphically.

  • emcee: The MCMC Hammer (possibly most used code) - https://ascl.net/1303.002

    Markov chain Monte Carlo (MCMC) Ensemble sampler - probably one of the most downloaded codes used in papers. Together with the corner.py code, which plots up the covariance between variables.

  • Aegean: Compact source finding in radio images - https://ascl.net/1212.009

    Although python, more command line driven code

  • Galactus: Modeling and fitting of galaxies from neutral hydrogen (HI) cubes - https://ascl.net/1303.018

    complex build system, needs c++ and boost.

  • Astropy: Community Python library for astronomy - https://ascl.net/1304.002

    ok.

  • AstroAsciiData: ASCII table Python module - https://ascl.net/1311.003

    Only works with python2

  • SunPy: Python for Solar Physicists - https://ascl.net/1401.010

  • pyExtinction: Atmospheric extinction - https://ascl.net/1403.002

    It's still python2, why oh why

  • Gammapy: Python toolbox for gamma-ray astronomy - https://ascl.net/1711.014

    Needs data.

  • galpy: Galactic dynamics package - https://ascl.net/1411.008 [peter]

    ok

  • PyBDSF: Python Blob Detection and Source Finder - https://ascl.net/1502.007

    Needs fortran and boost, and some python packages. Source code has test image. Invented their own "ipython" shell

  • pYSOVAR: Lightcurves analysis - https://ascl.net/1503.008

    very author specific

  • PyTransit: Transit light curve modeling - https://ascl.net/1505.024

    OK

  • pyro: Python-based tutorial for computational methods for hydrodynamics - https://ascl.net/1507.018

    Regular pip install didn't work, but from source it did.

  • SavGolFilterCov: Savitzky Golay filter for data with error covariance - https://ascl.net/1601.012

    Just a code from a paper. no pip.

  • POPPY: Physical Optics Propagation in PYthon - https://ascl.net/1602.018

    OK

  • Lmfit: Non-Linear Least-Square Minimization and Curve-Fitting for Python - https://ascl.net/1606.014 [peter]

    ok, good alternative to curve_fit(). Here it would be interesting to show how you can fit something with curve_fit() and lmfit. Your instructors may give you a dataset to fit and compare results.

  • PRECESSION: Python toolbox for dynamics of spinning black-hole binaries - https://ascl.net/1611.004

  • Gala: Galactic astronomy and gravitational dynamics - https://ascl.net/1707.006 [peter]

    OK

  • PROFILER: 1D galaxy light profile decomposition - https://ascl.net/1705.010

    Seems old by now. Indeed, it's python 2 only

  • SETI-EC: SETI Encryption Code - https://ascl.net/1803.009

    A challenge!

  • pydftools: Distribution function fitting in Python - https://ascl.net/code/v/1940

    OK

  • Photon: Python tool for data plotting - https://ascl.net/1901.007

    A regular "pip install" claimed versions were not matching, but a manual "pip install -e ." in the source code worked, but then complains that my matplotlib has no mplDeprecation attribute.

  • oscode: Oscillatory ordinary differential equation solver - https://ascl.net/1908.012

    builds a wheel; mimics solve_ivp()

  • GWpy: Python package for studying data from gravitational-wave detectors - https://ascl.net/1912.016

    ok

  • seaborn: Statistical data visualization - https://ascl.net/2012.015 [peter]

  • Citlalicue: Create and manipulate stellar light curves - https://ascl.net/2202.014

    Lot of dependancies that were not listed (arviz, celerite, corner,. ...)

  • CosmosCanvas: Useful color maps for different astrophysical properties - https://ascl.net/2401.005

    OK

  • CONCEPT: COsmological N-body CodE in PyThon - https://ascl.net/2306.035

    Too compute intensive!!!

  • AMUSE: Astrophysical Multipurpose Software Environment - https://ascl.net/1107.007 [peter]

    Most likely too large and complex, it's a python package talking to legacy codes to run a set of simulations of different nature (n-body, stellar evolution, hydro dynamics etc.) Has a sister code called OMUSE.

Installation Guidelines

Most python packages can be installed with pip. Within spyder this would be the following command (but check the README file in the package for guidelines, sometimes a conda solution is given as well), e.g.

  !pip install galpy

Some packages will rely on you having a C compiler, and if that fails, it's better to find another package before spending too much time solving compiler and linker problems. There is also the danger that mixing conda and pip may mess up your python environment.

If have downloaded the source code via git, you can also install it as developer, i.e. while you would be editing their files, your modified code would be used!

  git clone https://github.com/jobovy/galpy
  pip install -e galpy

There were a few packages where the regular pip install did not work, but a source install like this did work. go figure.