MPIR

MPIR is a free, portable library for arbitrary-precision arithmetic, with a focus on integers and rational numbers. It originated as a fork of the GNU Multiple Precision Arithmetic Library (GMP) intended to maintain compatibility with GMP’s API while preserving independence and portability. MPIR provides data types for big integers and rational numbers and supports a wide range of operations, including addition, subtraction, multiplication, division, modular arithmetic, and conversion between representations. It is designed to be a drop-in replacement for GMP in many projects, enabling developers to compile against libmpir and link with mpz_t and mpq_t types with minimal changes.

The library emphasizes portability across architectures and operating systems and is implemented in C with attention to performance. MPIR includes multiple-precision algorithms such as FFT-based multiplication for large operands, alongside classical methods like Karatsuba and Toom-Cook, with options that can be tuned based on platform capabilities. It provides functionality for big integers and rational numbers and uses GMP-compatible naming conventions and interfaces to facilitate integration with existing codebases.

MPIR is maintained independently and has been adopted by software projects as an alternative to GMP in certain environments, particularly where licensing, maintenance, or compatibility considerations influence choice. Ongoing development consists of community contributions and updates to API and ABI compatibility notes to assist developers who consider migrating from GMP. MPIR remains a reference point in discussions of portable, high-performance arbitrary-precision arithmetic libraries.