numdiff
Automatic and numerical differentiation.
Documentation
Please see https://docs.rs/numdiff.
Overview
This crate implements two different methods for evaluating derivatives in Rust:
- Automatic differentiation (forward-mode using first-order dual numbers).
- Numerical differentiation (using forward difference and central difference approximations).
This crate provides generic functions (for numerical differentiation) and macros (for automatic differentiation) to evaluate various types of derivatives of the following types of functions:
- Univariate, scalar-valued functions (
f: ℝ → ℝ) - Univariate, vector-valued functions (
f: ℝ → ℝᵐ) - Multivariate, scalar-valued functions (
f: ℝⁿ → ℝ) - Multivariate, vector-valued functions (
f: ℝⁿ → ℝᵐ)
These functions and macros are made generic over the choice of vector representation, as long as the vector type implements the linalg_traits::Vector trait. See the linalg_traits documentation for more information.