Optimization Solvers

A comprehensive Rust library implementing composable state-of-the-art numerical optimization algorithms with WebAssembly support for browser-based optimization.

Quick Start

use optimization_solvers::{GradientDescent, BFGS, Newton};

// Run optimization with any solver
let result = solver.minimize(objective_function, max_iterations);

📚 Documentation & Resources

• Documentation - Full Rust API documentation
• Examples - Complete examples for all optimization algorithms
• WASM Documentation - Browser-based optimization with WebAssembly
• Academic resources - Misc of papers/books about numerical optimization

🧮 Solver Categories

First-Order Methods

• Gradient Descent - Classic steepest descent
• Projected Gradient - Constrained optimization
• Coordinate Descent - Coordinate-wise optimization
• SPG - Spectral Projected Gradient
• P-Norm Descent - Lp-norm based descent

Quasi-Newton Methods

• BFGS - Broyden-Fletcher-Goldfarb-Shanno
• DFP - Davidon-Fletcher-Powell
• Broyden - Broyden's method
• L-BFGS-B - Limited-memory BFGS with bounds (enable lbfgsb feature flag)

Second-Order Methods

• Newton's Method - Classical Newton optimization
• Projected Newton - Constrained Newton
• SPN - Spectral Projected Newton

🚀 Getting Started

# Add to Cargo.toml
cargo add optimization-solvers

# Run examples
cargo run --example gradient_descent_example
cargo run --example bfgs_example

# Build for WebAssembly
cd wasm && ./build-wasm.sh

📦 Features

• Multiple Algorithms: 15+ optimization algorithms

• WebAssembly Support: Run in browsers with full performance

• Line Search Methods: Backtracking, More-Thuente, and more

• Bounded Optimization: Support for box constraints

• Comprehensive Examples: Ready-to-run examples for all solvers