samyak-268 / linear-algebra

Implementation of some common Linear Algebra routines

Linear Algebra Library

This library hosts C++ implementation of some of the routines which are most frequently encountered in Linear Algebra.

Installation

The project can be configured using CMake. The steps for the same are given below:

• Create a directory where you want the binaries to build. Let's call it the build directory.

mkdir build

• Change the working directory to the build directory and run CMake as follows:

cd build
cmake ..

• After the configuration is complete, the Makefile is generated inside the build directory. You can now compile the project and run the executable as follows:

make
./LinearAlgebra

How to Contribute

Each new module should go into a different directory whose name must be synonymous with the module functionality.

Let us suppose, you want to add a module for Singular Value Decomposition (S.V.D.) and you have implemented two files for this purpose: svd.h and svd.cpp. Both these files should reside inside a separate directory named svd which would also contain a CMakeLists.txt. Add the follwing line to the CMakeLists.txtresiding inside the module directory :

add_library(SVD svd.cpp)

This instruction tells CMake to build a library from the function implementations in svd.cpp.

To make use of this library, we add a few statements to the top level CMakeLists.txt file, which are given below:

include_directories("${PROJECT_SOURCE_DIR}/svd")
add_subdirectory(svd)

The include_directory() command tells CMake to look inside the svd directory for the header files for this library and the add_subdirectory() call ensures that the svd library gets built.

The final step would be to add (link) the library with the executable where it's actually being used. Say, you want to reference some functions from the svd module in main.cpp. Adding the following line to the top level CMakeLists.txt serves this purpose:

# LinearAlgebra is the name of the executable built from main.cpp
# SVD is the name of the executable built from the library modules
target_link_libraries(LinearAlgebra SVD)

The module is now ready to be used with main.cpp with the inclusion of a simple #include "svd.h"

linear-algebra's Issues

Incorporating gaussj_AbKDs.cpp into gaussj.cpp

The Gauss-Jordan elimination method implemented in gaussj_AbKDs.cpp should be incorporated within gaussj.h and gaussj.cpp. Try to re-use as many features already implemented in the module as possible to avoid potential re-inventing the wheel.

Also, double comparison should be modified to use numeric_limits<double>::epsilon() in <limits> header.