SoCiS 2013 ideas

Specutils Summer of Code in Space 2013

Currently there is much effort going into the astropy project to build a framework to analyze astronomical data in python, as well as to foster interoperability between existing astrophysical analysis tools. Astropy is separated into a core-package and affiliated packages. The core package provides very fundamental data structures for the analysis (e.g. tables with meta data, units, ..). The affiliated packages, like specutils, provide specific functionality for a sub-field of data analysis.

Light is basically the only information we have about the cosmos. Thus, analyzing optical spectra is one of the most important tasks in astrophysics. Thus, a variety of python packages exists that address particular parts of spectroscopy (e.g. pymodelfit, pyspeckit, astropysics, pysynphot, pymidas, etc.). Our aim is to provide commonly used tasks as part of specutils, as well as provide data structures that simplify interoperability between the existing packages to avoid duplication of effort. We imagine, the project to be divided into subcomponents.

1. Develop the data structures required to support spectroscopic analysis, including unit conversions between spectroscopic systems and flux measurements.
2. Modelling and fitting of spectra by incorporating astropy models
3. A graphical interactive user interface to accomplish this task.

We would ask students to develop those subcomponents, write a documentation and unit tests.

Project 1. Building the Framework

The first major project needed is to build the general framework on top of astropy.nddata.

Much of the capability of specutils relies upon features already implemented in astropy. The major new features needed are spectral readers, i.e. from_ methods, to parse in different types of spectra into the proper units.

specutils aims to support reading FITS files subscribing to a wide variety of different header conventions. Ideally, each reader will be able to initialize both x and y units for any spectrum produced by standard data reduction pipelines.

Tools to create:

1. spectrum1D.from_fits_file (generic)
2. spectrum1D.from_table (generic)
3. spectrum1D.register_reader