#MoonBit Arbitrary-Precision Math Library

#Introduction

mbmath is an active MoonBit port of core mpmath layers, including arbitrary-precision real (mpf), complex (mpc), interval (mpi/mpci) arithmetic, and selected special-function families.

Current baseline (2026-02-26):

Runtime test baseline: moon -C mbmath test --target native => 338 passed, 0 failed.
Python migration board: 36 files in 部分, 0 files still 未开始, 1 file marked 已迁移.
High-level numeric APIs now include Stage C (diff/quad/findroot/summation/levin/nsum/nprod) and Stage D (matrix/linalg/eigen/ode) with qr/qr_solve/cholesky/hessenberg/schur/eig/eighe/svd/cond subsets.
Stage E integration surface has started with migrated subsets of compatibility/pickle/torture/cli/visualization.

The project is usable on the implemented surface, but still not full-parity with upstream mpmath.

#Roadmap

Deepen Stage E parity

Expand CLI migration from current non-interactive subset to REPL/IPython semantics.
Expand visualization migration from current plot/cplot sampling adapters to backend-agnostic rendering hooks.

Upgrade partial files toward full parity

Continue promoting high-impact 部分 files to 已迁移 (P1/P2 first).

Reliability hardening

Continue removing remaining runtime try! and replace with structured raise propagation.
Add focused regression tests for difficult precision/conditioning edge cases.

Release preparation

Publish explicit parity matrix (supported/partial/unsupported functions).
Cut preview release with compatibility notes and known limitations.

Site source Report issue Packages Build queue Skills Statistics