Garamon Generator

Garamon (Geometric Algebra Recursive and Adaptative Monster) is a generator of C++ libraries dedicated to Geometric Algebra. From a configuration file, the library generator generate the source code of the specified Algebra as well as a compatible sample code and some documentation.

For help or discussion related to Garamon, you can join us on Discord.

Features

* MIT Licence allows free use in all software (including GPL and commercial)
* multi-platform (Windows, Linux, Unix, Mac)

Install

Dependencies

* 'Eigen 3.3.4'  or more [Eigen](http://eigen.tuxfamily.org)
* 'CLI11 2.7.2'  or more [CLI11](https://github.com/CLIUtils/CLI11)
* 'CMake 3.24' or more
* 'Catch2 3' or more [Catch2](https://github.com/catchorg/Catch2) (only for the unit tests, downloaded if not found)
* 'pybind11' (only for building the Python bindings of a generated library, install with 'pip install pybind11'; requires CMake 3.18 or more)

Compiler tested

* gcc 5.4.0
* clang 4
* apple-clang 900.0.39.2
* MinGW 7.2.0
* MSVC 19.14.26430.0 (Visual Studio 15.7.3)

Install for Linux-Mac from the terminal

* Check the dependencies
* From Garamon Generator root directory
    * 'mkdir build'
    * 'cd build'
    * 'cmake ..'
    * check that the cmake output has no errors
    * 'make'
    * the binary executable is on the 'build' directory

Install for Windows with MinGW, using Windows Power Shell

* Check the dependencies
* From Garamon Generator root directory
    * 'mkdir build'
    * 'cd build'
    * 'cmake -G "MinGW Makefiles" ..'
    * check that the cmake output has no errors
	* 'mingw32-make'
    * the binary executable is on the 'build' directory

Install for Windows with Visual Studio 15 2017 Win64, using Windows Power Shell or cmd

* Check the dependencies
* From Garamon Generator root directory
    * 'mkdir build'
    * 'cd build'
    * 'cmake -G "Visual Studio 15 2017 Win64" ..'
    * check that the cmake output has no errors
	* open the file garamon_generator.sln with Visual Studio
	* generate the project "ALL_BUILD" with Release configuration
    * the binary executable is on the 'Release' directory

Usage

~/garamon/build$ ./garamon_generator --help
GARAMON: Geometric Algebra Recursive and Adaptative MONster Generator

./garamon_generator [OPTIONS]

OPTIONS:
  -h,  --help                      Print this help message and exit
  -c,  --config REQUIRED           Configuration file, e.g. "garamon/conf/c3ga.conf"
  -t,  --template REQUIRED         Template directory, e.g. "garamon/data/"
  -o,  --output REQUIRED           Output directory, e.g. "garamon/build/output/"

~/garamon/build$

Generate a library

* define the algebra to generate: chose a configuration file (.conf) on the 'conf' directory or create your own.
* run the binary executable (from the 'build' directory) with the configuration file as argument
  > ./garamon_generator -c ../conf/ga.conf -t ../data/ -o ./output/
* the generated library is located in 'build/output' directory
* to install the generated library, see its README.md

Run the Python binding sample for a specific algebra

Let us consider the considered algebra is CGA of R3 corresponding to the configuration file c3ga.conf.

* Check the dependencies (in particular `pip install pybind11`, Python 3 development headers and CMake 3.18 or more)
* From Garamon Generator root directory
	* 'mkdir build'
	* 'cd build'
	* 'cmake ..'
	* 'make'
    * './garamon_generator -c ../conf/c3ga.conf -t ../data/ -o ./output/'
	* 'cd output/garamon_c3ga/'
	* 'cmake -S . -B build -DBUILD_PYTHON=ON' (¹)
	* 'cmake --build build --config Release'
	* 'cmake --build build --config Release --target python_sample' (runs 'sample/sample.py' with the just-built module)
	* (optional) 'cmake --install build --config Release --component python' (²) to make the module importable from anywhere, then:
		* 'cd sample'
		* 'python sample.py'

(1) The cmake -S ... command requires the Eigen3 library. It will fail if it can't find it. You might need to add the -DCMAKE_PREFIX_PATH=/path/to/eigen3 parameter to point to where the library is.

(2) The module is installed into the site-packages of the Python interpreter found by CMake, which usually requires admin rights (sudo, or an elevated terminal on Windows). Either install to your per-user site-packages instead, by adding -DPYTHON_INSTALL_DIR="$(python -m site --user-site)" to the cmake -S ... command, or activate a virtual environment (python -m venv .venv) before the cmake -S ... command: CMake then uses it and installs the module there.

Tests

The unit tests use Catch2 (v3) and CTest. They are built by default (CMake option GARAMON_BUILD_TESTS):

~/garamon$ cmake -S . -B build
~/garamon$ cmake --build build
~/garamon$ ctest --test-dir build --output-on-failure

• generator_tests: unit tests of the generator itself (configuration parser, metric analysis, product tables, code generation, ...)
• <algebra>_tests: the algebras c3ga, c5ga, e3ga and p3ga, as well as two test algebras with a non-normalized and a non-orthogonal metric (tests/conf/), are generated at build time; each generated library is compared with an independent reference implementation of the geometric algebra (tests/algebras/reference/CliffordReference.hpp)
• <algebra>: sample: the generated sample program is compiled and run.

To run a part of the tests, e.g. the c3ga dual tests: build/tests/c3ga_tests "[dual]", or ctest --test-dir build -R c3ga.

Notes

Authors

* Vincent Nozick (Universite Paris-Est Marne-la-Vallee, France)
* Stephane Breuils (Universite Savoie Mont-Blanc, Japan)

Contact

* vincent.nozick (at) u-pem.fr

Reference

If you use Garamon for research purpose, please cite the following paper:

@Article{breuils_garamon_2019,
author="Breuils, St{\'e}phane and Nozick, Vincent and Fuchs, Laurent",
title="Garamon: A Geometric Algebra Library Generator",
journal="Advances in Applied Clifford Algebras",
year="2019",
month="Jul",
day="22",
volume="29",
number="4",
pages="69",
issn="1661-4909",
doi="10.1007/s00006-019-0987-7",
url="https://doi.org/10.1007/s00006-019-0987-7"
}