lcit / non_lin_optim

Lightweight library for non-linear optimization written in modern C++.

non_lin_optim

Header only library for non-linear optimization using Eigen. I wrote this library to practice modern C++ and to rehearse optimization theory.

Features

• Numerical derivates (Jacobian, Hessian)
• Newton-Raphson
• Gaussian-Newton
• Gradient descent
• (TODO) Levenberg-Marquard

Usage

#include <non_lin_optim/optim.h>
#include <iostream>

using namespace non_lin_optim;  
    
Vector t(2);
t(0) = 1;
t(1) = 2;

Vector x(2); 
x(0) = 0.0;
x(1) = 0.0; 

auto func = [&](const Vector& x) -> Vector {
    Vector y_hat(2);
    for(int i=0; i<t.size(); ++i){
        y_hat(i) = pow(x(0)*t(i)-0.2, 2) + x(1);
    }
    return y_hat.array()-1; // return residuals
};    

auto optimizer = optim::GaussianNewton(func);  
auto info = optimizer.run(x);

std::cout << info << std::endl;
std::cout << x << std::endl;

Matrix J = numerical_deriv::JacobianApprox(func, x);
std::cout << J << std::endl;

Matrix H = numerical_deriv::HessianApprox(func_s, x);
std::cout << H << std::endl;

Build and run the examples

Use the following command give an example of how to build and run an example.

cmake -S examples -B build/examples/refine_camera_extrinsics
cmake --build build/refine_camera_extrinsics
./build/refine_camera_extrinsics/refine_camera_extrinsics --help

Build and run test suite

Use the following commands from the project's root directory to run the test suite.

cmake -S test -B build/test
cmake --build build/test
CTEST_OUTPUT_ON_FAILURE=1 cmake --build build/test --target test

# or simply call the executable: 
./build/test/Tests

To collect code coverage information, run CMake with the -DENABLE_TEST_COVERAGE=1 option.

Run clang-format

Use the following commands from the project's root directory to check and fix C++ and CMake source style. This requires clang-format, cmake-format and pyyaml to be installed on the current system.

cmake -S test -B build/test

# view changes
cmake --build build/test --target format

# apply changes
cmake --build build/test --target fix-format

See Format.cmake for details.

Build the documentation

The documentation is automatically built and published whenever a GitHub Release is created. To manually build documentation, call the following command.

cmake -S documentation -B build/doc
cmake --build build/doc --target GenerateDocs
# view the docs
open build/doc/doxygen/html/index.html

To build the documentation locally, you will need Doxygen, jinja2 and Pygments on installed your system.