## Description This code is a one-dimensional discontinuous Galerkin time-domain code -- a discontinuous finite element method in time domain -- modelled after Hesthaven and Warburton \cite hesthaven2008nodal. The code differs from the given Matlab code in the way that it can be easily extended to a region based assignment of solution scheme parameters.

The code is applied to a mesh of a given 1D structure. The mesh needs to be in the current mesh format given by Gmsh.

Coding Guidelines

Though I do my best to follow the C++ Core Guidelines by Bjarne Stroustrup and Herb Sutter, there are a few points I would like to an extra emphasis on.

• In some cases there will be a competition between conciseness, readability and performance. In this case, I will put readability (where maintainability is a part of) first, and sustainability second. I will check for performance last as this will not be an issue for a 1D code (given that I have not written mindlessly wasteful pieces of code which need to be addressed). I will not attempt to write extra clever short code, as it might be hard to read. The simpler, the better.
• I preferably write short method units to improve readability (maintainability) and aid testing. Multiple function calls are (at least to me) not an issue.
• Whenever possible and practical, I do not write methods with side effects.
• The code within the files is ordered in a specific way: The most crucial information to the reader is put first. The importance of the implemented methods to the reader descends as the file progresses.
• I distinguish between interface and implementation, where the interface will contain the public members.
• I do my best to choose descriptive names.

Workflow

You will not see me opening any feature branches in this git repository. Obviously, because this code is only written by me, but also because I do believe that feature branching is counterproductive to CI. I will rather do a dark launch.

C++ Standard

The standard of this code is C++20, where I mostly make use of the \<set> library.

Requirements

• g++
• Boost
• BLAS
• LAPACK
• Armadillo
• ZLIB

Installation

In the following you find a command line description of the installation. The installation is pretty standard in this case.

Clone the git repository to obtain the code

git clone git@github.com:dan-nha/dgtd.git

Create a build directory in the code's root directory

mkdir build

and go into the newly created directory

cd build

Afterwards type the commands

cmake ..

and

make install

into your command line.

Usage

In the code's root directory you will find an binary file called dgtd lying in the folder bin. Execute the binary by giving it a mesh file produced by Gmsh as argument, e.g.

./dgtd my.msh

You can find example meshes under examples.

Documentation

I rather give some motivation for certain design choices, and the context of certain methods than describe what a part of code is doing. In this regard, I hope the given names and the implementation to be verbose enough to be able to abstain from lengthy documentation. This might not always be possible though.

To build the documentation go to the doc directory and type the following command.

doxygen Doxyfile.in

Open doc/doxygen/html/index.html with a browser to view the documentation.