Next: PyFITS, a FITS Module for Python
Up: Software Systems and Tools
Previous: The Faint Object Spectrograph Post-COSTAR Spectropolarimetry Correction
Table of Contents - Subject Index - Author Index - Search - PS reprint - PDF reprint

Pirzkal, N. & Hook, R. N. 1999, in ASP Conf. Ser., Vol. 172, Astronomical Data Analysis Software and Systems VIII, eds. D. M. Mehringer, R. L. Plante, & D. A. Roberts (San Francisco: ASP), 479

Python in Astronomy

Norbert Pirzkal, Richard N. Hook
ST-ECF, Karl-Schwarschild Str.2, Garching bei Munchen D-85748, Germany

Abstract:

We report on the use of Python to perform basic astronomical tasks. Python is a powerful, object-oriented scripting language that is easy to extend, free, and available for most computer platforms. Python, together with its Numerical module, and with the pFitsio module that is being developed by one of the authors, provides an attractive alternative to other languages such as Perl, IDL, and the IRAF CL, which are already widely used by the community. Simple examples of how Python can be used to explore and manipulate real astronomical data are given.

1. Why Python?

Python is an interpreted, interactive, object-oriented programming language. Invented in 1990 by Guido van Rossum, it shares many of its features with other modern languages such as Tcl, Perl and Java. Python has a simple, easy to learn syntax which emphasizes readability, and it draws many of its features from other popular languages such as C, C++, Modula-3, ABD, and others. We have found that Python satisfies the need for a high level programming language which fills the gap between regular shell scripting languages and compiled languages such as C. For tasks that are too complicated and too large to implement with regular shell scripts, but for which the extra overhead of coding everything in C is not warranted, Python offers a quick and elegant option. Some of the attractive features of Python include the use of transparent byte-code compilation for speed, automatic memory management and garbage collection, and a very powerful object oriented and modular design. Python ships with a large number of modules (See the Python standard documentation), and can be easily extended in C. With modules such as pFitsio or PyFits and the Numerical modules, astronomical images and tables can be easily accessed and manipulated as numerical arrays. We present small Python examples as illustrations of the language's power and encourage the reader to learn more about Python from the book by Lutz (1996).

2. Examples

3. Reading a List from a File and building Dictionaries

In this example (Fig. 1), we assume that we have a file called assoc.clear. This file contains a list of datasets from the HST STIS camera which have been previously determined to be combinable into longer integration time images that we wish to refer to as ``Associations''. The file is as follow:


ID       sub dataset   xoff  yoff   ra     dec  r d  exp
O46P3X010 1 O46P3X010  0.00  0.00 271.37 -19.90 0 1 300.00 ...
O46P3X010 1 O46P3Y010 -0.07 -0.02 271.37 -19.90 0 1 300.00 ...
...

We would like to be able to easily query all the members of a given association (``ID'' column), and to be able to access and potentially modify the information of the individual datasets that make up an association. We therefore build a Dictionary of datasets where each dataset's name serves as a key in this dictionary. Each entry of the dataset dictionary is in fact itself a dictionary which uses the original ASCII file column name as keys to hold the ASCII file original data. Hence, dataset['O46P47010']['xoff'] would return the value of 0.0. Another dictionary is then created and uses the names of the associations (Column 1 in the ASCII file) as a key. Each association key is then assigned a list of datasets from the dataset dictionary. Once the dictionaries are set up, which only takes a few lines (grey overlay region), one has the ability to manipulate and ``query'' those dictionaries in a very high level, natural, way.

**Figure 1:** Dictionary and List creation and manipulation, (see text for more description).
$\begin{figure} \epsscale{0.9} \plotone{pirzkaln1.eps} \end{figure}$

4. Reading FITS data into Arrays and manipulating those Arrays

Starting from the set of dictionaries we have built in our first example, we now proceed to read in each individual FITS file (Fig. 2). Notice that we now load both the pFitsio and the Numerical modules and how, once an image is stored in a two dimensional array, we can compute statistics on any part of this array. Our second example also shows the use of the powerful Numerical module's where() command which the IDL¹ user will recognize.

**Figure 2:** Array and FITS file access using Python, (see text for more description).
$\begin{figure} \epsscale{0.9} \plotone{pirzkaln2.eps} \end{figure}$

5. Availability and References

Python runs on virtually all computer platforms, including the JAVA virtual machine (JPython). One can download the compiled Python binaries or the Python source code from the Python Homepage. Exhaustive Python documentation, tutorials, JPython, and third party modules can also be found on the Python homepage. Information concerning the pFitsio and PyFits modules should be directed to the authors of this paper, and Paul Barrett (see Barrett & Bridgman 1999), respectively.

References

Barrett, P. E. & Bridgman, W. T. 1999, this volume, 483

Lutz, M. 1996, Programming Python, (Sebastopol: O'Reilly)

Footnotes

... IDL ¹: IDL is the trademark of Research Systems, Inc.

adass@ncsa.uiuc.edu