Numarray

Numarray is a Python library that provides multi-dimensional, homogeneous arrays for numerical computing. It was developed as part of the Numeric Python project as an alternative to the earlier Numeric package, with the goal of delivering faster performance and more efficient memory use for large arrays.

Numarray offers an array object with a fixed data type stored contiguously in memory. It supports common numeric types such as integers, floating point numbers, and complex numbers, along with boolean values. The interface includes array creation, reshaping, indexing and slicing, as well as element-wise operations, reductions, and basic linear algebra routines. The implementation combines Python-level wrappers with performance-critical code written in C, aiming to minimize Python overhead while providing competitive speed on array computations. Numarray is designed to integrate with the broader scientific Python ecosystem and to work well with other libraries in the suite.

Historically, numarray was proposed as an improvement over Numeric, addressing certain performance and memory management concerns. In the mid-2000s, a new project known as NumPy emerged, combining the ideas and codebases of Numeric and numarray. NumPy eventually became the standard numerical array package in Python, offering a broader feature set and a larger community, while numarray was largely superseded and is now rarely used in new projects.

See also: Numeric, NumPy, SciPy.