Commit f36eebf2 authored by Jakub Klinkovský's avatar Jakub Klinkovský
Browse files

Update LICENSE and README files

parent ed685692

LICENSE.txt

+1 −0
Original line number Diff line number Diff line 
The MIT License (MIT)



Copyright (c) 2019 Fabian Meyer

Copyright (c) 2022 Jakub Klinkovský



Permission is hereby granted, free of charge, to any person obtaining a copy

of this software and associated documentation files (the "Software"), to deal

README.md

+14 −26
Original line number Diff line number Diff line 
# gradient-descent-cpp

# tnl-gradient-descent



![Cpp11](https://img.shields.io/badge/C%2B%2B-11-blue.svg)

![License](https://img.shields.io/packagist/l/doctrine/orm.svg)

![Travis Status](https://travis-ci.org/Rookfighter/gradient-descent-cpp.svg?branch=master)

![Appveyor Status](https://ci.appveyor.com/api/projects/status/66uh2rua4sijj4y9?svg=true)



gradient-descent-cpp is a header-only C++ library for gradient descent

optimization using the ```Eigen3``` library.

tnl-gradient-descent is a header-only C++ library for gradient descent

optimization using the Template Numerical Library.



## Install
@@ -21,20 +16,11 @@ cmake .. 
make install

```



The library requires ```Eigen3``` to be installed on your system.

In Debian based systems you can simply type



```bash

apt-get install libeigen3-dev

```



Make sure ```Eigen3``` can be found by your build system.



You can use the CMake Find module in ```cmake/``` to find the installed header.



## Usage



There are three steps to use gradient-descent-cpp:

There are three steps to use tnl-gradient-descent:



* Implement your objective function as functor

* Instantiate the gradient descent optimizer
@@ -51,9 +37,10 @@ struct Ackley 
    Ackley()

    { }



    double operator()(const Eigen::VectorXd &xval, Eigen::VectorXd &) const

    template <typename Vector>

    double operator()(const Vector &xval, Vector &) const

    {

        assert(xval.size() == 2);

        assert(xval.getSize() == 2);

        double x = xval(0);

        double y = xval(1);

        // Calculate ackley function, but no gradient. Let gradien be estimated
@@ -66,6 +53,9 @@ struct Ackley 


int main()

{

    // Define the vector type

    using Vector = TNL::Containers::Vector<double, TNL::Devices::Host, gdc::Index>;



    // Create optimizer object with Ackley functor as objective.

    //

    // You can specify a StepSize functor as template parameter.
@@ -75,8 +65,7 @@ int main() 
    // You can additionally specify a FiniteDifferences functor as template

    // parameter. There are Forward-, Backward- and CentralDifferences

    // available. (Default is CentralDifferences)

    gdc::GradientDescent<double, Ackley,

        gdc::WolfeBacktracking<double>> optimizer;

    gdc::GradientDescent<double, Ackley, gdc::WolfeBacktracking<double>> optimizer;



    // Set number of iterations as stop criterion.

    // Set it to 0 or negative for infinite iterations (default is 0).
@@ -101,8 +90,7 @@ int main() 
    optimizer.setVerbosity(4);



    // Set initial guess.

    Eigen::VectorXd initialGuess(2);

    initialGuess << -2.7, 2.2;

    Vector initialGuess = {-2.7, 2.2};



    // Start the optimization

    auto result = optimizer.minimize(initialGuess);
@@ -114,7 +102,7 @@ int main() 
    std::cout << "Final fval: " << result.fval << std::endl;



    // do something with final x-value

    std::cout << "Final xval: " << result.xval.transpose() << std::endl;

    std::cout << "Final xval: " << result.xval << std::endl;



    return 0;

}