Features of numarray

Numarray is a reimplementation of the Numeric module that provides Python with capabilities similar to Matlab, IDL, or Octave. It improves on Numeric by:

• Supporting access to byteswapped, misaligned, discontiguous, or type-mismatched arrays without requiring a complete temporary copy.
• Modifying the type coercion rules to avoid unexpected type inflation (eg., 2.0 × <Float32 array> should not result in a Float64 array).
• Implementing arrays as a family of classes rather than as an abstract C-type. Numarray provides python classes for handling numbers, fixed-length-strings, and records.
• Supporting access to fields of arrays of records as numeric arrays without copying the data to a new array.
• Supporting memory mapping of arrays.
• Generating most C-extension code using python scripts.
• Implementing its most complex functionality in Python.

Numarray to-do list:

• Package ports. Numarray currently provides ports only for RandomArray, LinearAlgebra, Mlab, and FFT.
• Support for extended precision types: Float128 and Complex128 are not yet supported.
• Numeric compatability.