cr-nimble 0.4.0

Iterative algorithms for numerical linear algebra with JAX

CR-Nimble

CR-Nimble consists of fast linear algebra and signal processing routines. Most of the routines have been implemented using Google JAX. Thus, they can be easily run on a variety of hardware (CPU, GPU, TPU).

Functionality includes:

• Utility functions for working with vectors, matrices and arrays
• Linear algebra functions
• Digital signal processing functions
• Data compression functions
• Test data generation functions

Installation

python -m pip install cr-nimble

For Windows, you can use unofficial JAX builds from here.

Import

import cr.nimble as crn

See documentation for library usage.

CR-Nimble is part of CR-Suite.

Related libraries:

• CR-Wavelets
• CR-Sparse