Tools for data analysis and
data generation
When faster
numerical computation and/or greater
dimensional abstraction than native Python affords is needed, NumPy is
a "standard" by which other tools can be judged.
Name 

NumPy 
Description 

NumPy
(Numeric Python) is an
extensive array data processing tools. It contains useful procedures
for numerical computation. It can be used for fast manipulation of
multidimensional arrays.
random 
random
generation tools 
linalg 
linear
algebra tools 
dft 
FFT
(fast Fourier transform)
routines 
testing 
Scipy
(Scientific Python)
testing tools 
NumPy also supports a FortrantoPython interface generator and other
Fortran compiler tools. 
Principal
Reference 

The SciPy (scientific
python) website: http://www.scipy.org/ 
Documentation 

http://numpy.scipy.org/numpy.pdf
//www.scipy.org/Documentation/ has a comprehensive list of documents.
update
20090127: Oliphant,
T.E. 2006. Guide to NumPy. This book
has been released into the public domain after three years of
restricted distribution. http://www.tramy.us/ 
Downloads 

http://www.scipy.org/
The SciPy community is quite active and they maintain a good range of
downloads. Make sure you get the right version for your hardware,
Operating System and SoniPy's current Python version. 
Version 

1.0.3 
License 

OSIApproved Open Source software 
Dependencies 

None 
Warnings 

The
correct version for HW and SW is
important. 
Initial
Comments 

Many
scientific routines have been
written using these tools. Over its evolution, NumPy has also been
known as both Numeric and Numarray and references to it as such can be
found both in the documentation and in thirdparty code. Earlierthis
year (2007) there was some discussion on the Python SIG about reasons
for a Numeric package still being available, however I believe that for
those using it for the first time, NumPy is the correct package. In
fact some of the code examples in the package download still try to
"import numeric". I have yet to find an example which doesn't run if
"numeric" is replaced with "numpy" in the python code. 
Do you have other
suggestions for tools in this
category? For example, there there is also a Python interface to Matlab
called PyLab. See details on the matplotlib home page. If you'd like to
contribute your experience, contact us via the FORUM link.
