multicalc is a small Rust library for single- and multi-variable calculus: numerical differentiation, integration, Jacobians and Hessians, vector calculus, and Taylor approximation. The goal is to run anywhere including bare metal, hence we always ship with zero heap allocation, no panics, and no_std compatible.

Highlights of the new version:

• Simpler API. Integrands are generic F: Fn(...) (monomorphized, no &dyn dispatch), and the redundant "number of integrations / order" runtime arguments are gone since they are just const-generic array lengths.
• Infinite and semi-infinite integration limits for the iterative integrators, via a domain transform, for convergent and decaying integrands.
• Bug fixes in the Gauss-Hermite/Laguerre kernels, a 3D line-integral component, the integration-limit checks, and the linear/quadratic approximation formulas and metrics.
• Also: a runnable example per module, criterion benchmarks, and CI covering the no_std target and MSRV.

What's next: forward-mode dual-number autodiff for exact derivatives, a fixed-size const-generic linear-algebra core, ODE solvers, symbolic math, and optimization, all keeping the no_std core.

Looking for contributors! I combed the backlog into small, self-contained issues, each with a clear scope, labels, and its dependencies linked, and tagged the approachable ones good first issue. If you are looking to contribute to open source rust, or you just enjoy numerical methods, const generics, and scientific computations in embedded space, I would love the help. Happy to answer design questions too.

• crates.io: https://crates.io/crates/multicalc
• Repo: https://github.com/kmolan/multicalc-rust
• Changelog: https://github.com/kmolan/multicalc-rust/blob/main/CHANGELOG.md
• Good first issues: https://github.com/kmolan/multicalc-rust/issues?q=is%3Aissue+is%3Aopen+label%3A%22good+first+issue%22

1 post - 1 participant

Read full topic