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 Original line Diff line number Diff line 
The MIT License (MIT)

The MIT License (MIT)





Copyright (c) 2019 Fabian Meyer

Copyright (c) 2019 Fabian Meyer

Copyright (c) 2022 Jakub Klinkovský





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

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

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

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

README.md

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

# tnl-gradient-descent





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

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

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

optimization using the Template Numerical Library.

![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.





## Install

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

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.

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





## Usage

## 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

* Implement your objective function as functor

* Instantiate the gradient descent optimizer

* Instantiate the gradient descent optimizer
@@ -51,9 +37,10 @@ struct Ackley 
    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 x = xval(0);

        double y = xval(1);

        double y = xval(1);

        // Calculate ackley function, but no gradient. Let gradien be estimated

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




int main()

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.

    // Create optimizer object with Ackley functor as objective.

    //

    //

    // You can specify a StepSize functor as template parameter.

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

    // You can additionally specify a FiniteDifferences functor as template

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

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

    // available. (Default is CentralDifferences)

    // available. (Default is CentralDifferences)

    gdc::GradientDescent<double, Ackley,

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

        gdc::WolfeBacktracking<double>> optimizer;





    // Set number of iterations as stop criterion.

    // Set number of iterations as stop criterion.

    // Set it to 0 or negative for infinite iterations (default is 0).

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

    optimizer.setVerbosity(4);





    // Set initial guess.

    // Set initial guess.

    Eigen::VectorXd initialGuess(2);

    Vector initialGuess = {-2.7, 2.2};

    initialGuess << -2.7, 2.2;





    // Start the optimization

    // Start the optimization

    auto result = optimizer.minimize(initialGuess);

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

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





    // do something with final x-value

    // 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;

    return 0;

}

}