This is a library written in C++, using STL containers and algorithms in order to provide implementations for some of the core structures of the linear algebra.
C++MIT
Francois
This is a library written in C++, using STL containers and algorithms in order to provide implementations for some of the core structures of the linear algebra, starting with the Matrix and Determinants. Relevant concepts and structures are being added. This library attempts to draw a coherence between the mathematical properties of these structures and their respective implementations in programming. At the same time, it provides some convenient methods to create, transform, and utilize these structures. It uses the algebra namespace to provide implementations for all of its features. C++ static assertion is implemented in order to restrict its use only for the datatypes representing real numbers for the time being. A separate extension is being defined in order to explore more possibilities of working with custom datatypes, involving numeric computations.
Furthermore, it is an interface library which uses CMake build system to build and integrate the googletest in order to enable testing. The testing framework is closely integrated with the library during the CMake configuration time because the associated test suite is used both to validate the library features and at the same time acts as a viable reference of how to use them. All cases of input parameters which violate the mathematical logic behind the structures are handled with std::exception and its child types. Following are the core datatypes of the library.
algebra::ElementSequence
ElementSequence is the most fundamental structure in the algebra namespace which, represents any one dimensional sequence of real numbers. ElementSequence uses protected initializers in order to prevent users from instantiating this type. Therefore, it rather works as an internal base type which encapsulates the management and interaction with the data structure it uses behind the scene.
algebra::Row & algebra::Column
Both Row and Column datatypes inherit from the ElementSequence and represent the rows and the columns of a matrix or determinant. Therefore, they independently support several convenient arithmetic operations while at the same time sharing the same data structure behind the scene. Defining these types individually makes it easier to implement several other more complex operations e.g., multiplication of two matrices. Both algebra::Row and algebra::Column offer public initializers and some of the arithmetic operations they support are
* Addition and subtraction of two rows or columns of equal lengths from each other.
* Multiplication between a row and a column of equal length.
* Addition of a scalar to each element of the row or column.
* Subtraction of a scalar to and from each element of the row or column.
* Multiplication of a scalar to each element of the row or column.
* Division of each element of the row or column by a scalar.
* Modular division of each element of the row or column by a scalar.
algebra::MultiSequence
Similar to the ElementSequence, MultiSequence represents a two dimensional sequence of real numbers and manages the internal data structure. It enforces some constraints and offers some properties on the two-dimensional sequence which qualify itself as the base type of matrix and determinant. It provides access to its internal data structure in terms of an std::vector of algebra::Row and an std::vector of algebra::Column. It implements the subscript feature for elements in the two-dimensional sequence by index. It also defines the equality and inequality of two MultiSequence types. These features are used repeatedly further in matrix and determinant. Furthermore, it implements the mathematical concepts of elementary operations on matrices and determinants. These operations are further shared by both of its child types.
algebra::Matrix
Inheriting directly from the MultiSequence, matrix datatype encapsulates the classification of matrices as well as implements fundamental operations between two matrices such as
* Addition and subtraction of two matrices.
* Multiplication of a matrix by a scalar.
* Multiplication of two matrices which are compatible for the operation.
* Adjoint of a matrix (possible upto order 3 for now).
* Inverse of a matrix (possible upto order 3 for now).
* Transpose of a matrix.
* Elementary Row and Column operations on a matrix.
Matrix datatype can be instantiated using the C++ construct std::initializer_list like this
const algebra::Matrix<int> matrix1 = {
{1,2,4,0},
{3,3,0,1},
{6,2,8,1},
{2,7,0,5}
};
//Alternate types that can be used for initialization
std::vector<std::vector<T>> //T is type of a real number
std::vector<algebra::Row<T>> //T is type of a real number
std::vector<algebra::Column<T>> //T is type of a real number//e.g.const std::vector<algebra::Column<double>> columns = {{65.4,36.12,90.23,65.78},{47.1,93.15,81.32,43.98}};
const algebra::Matrix<double> matrix2 = columns;
//will result in matrix65.447.136.1293.1590.2381.3265.7843.98
Arithmetic and logical operations between two matrices could be performed like this
const algebra::Matrix<int> givenMatrix1 = {
{1,2,3,4},
{5,6,7,8},
{9,10,11,12}
};
const algebra::Matrix<int> givenMatrix2 = {
{13,14,15,16},
{17,18,19,20},
{21,22,23,24}
};
const algebra::Matrix<int> sumOfIntMatrices = givenMatrix1 + givenMatrix2;
//will result a matrix141618202224262830323436const algebra::Matrix<int> differenceOfTwoMatrices = givenMatrix1 - givenMatrix2;
//will result in a matrix
-12 -12 -12 -12
-12 -12 -12 -12
-12 -12 -12 -12const algebra::Matrix<int> productMatrix1 = givenMatrix1 * 4;
//will result in a matrix4812162024283236404448const algebra::Matrix<int> givenMatrix3 = {
{1,2,3,4},
{5,6,7,8},
{9,10,11,12}
};
const algebra::Matrix<int> givenMatrix4 = {
{1,2,3,4},
{5,6,7,8},
{9,10,11,12},
{13,14,15,16}
};
const algebra::Matrix<int> productMatrix2 = givenMatrix3 * givenMatrix4;
//will result in matrix of order (3,4)90100110120202228254280314356398440//If two matrics are provided which are not compatible for multiplication in terms of their respective orders an exception would be thrown with the appropriate reason.const algebra::Matrix<int> givenMatrix5 = {
{1,2,3,4},
{5,6,7,8},
{9,10,11,12}
};
const algebra::Matrix<int> transposeMatrix5 = givenMatrix5.Transpose();
//transposeMatrix5 will result in the matrix of order (4,3)159261037114812
algebra::Matrix datatype also provides several operations with the help of algebra::Determinant datatype e.g., deriving adjoint and inverse of a matrix. The library incorporates a detailed test suite and an example directory which demonstrate all possible operations with an algebra::Matrix datatype.
algebra::Determinant
Likewise the Matrix, algebra::Determinant datatype also inherits from the MultiSequence base type. Therefore, they share the same internal data structure and some of the properties and operations with the Matrix e.g., subscripts for elements, order and the elementary operations on rows and columns. The main reason to establish algebra::Determinant as a type of its own is to maintain the coherence with the mathematical structures. Furthermore, constructors of determinants enforce further constraints on the two-dimensional sequence compared to matrices. Determinant datatype defines some of its relevant constructs e.g., value, minors, and co-factors. These constructs, in turn, are used to define an invertible matrix and compute the inverse of a matrix. The library incorporates a detailed test suite and an example directory which demonstrate all possible operations with an algebra:: Determinant datatype.
For the time being the initialization of the algebra::Determinant datatype and all other related operations e.g., determining minors, co-factors, adjoint matrices, and inverse matrices are possible up to order 3 and below. Work is in progress to improve and overcome this limitation. This limitation does not apply to other operations with matrices.
Both algebra::Matrix and algebra::Determinant datatypes support basic row and column elementary operations like this
All possible arithmetic and logical operations with the algebra::Matrix and algebra::Determinant datatypes are validated and documented in the integrated test suite and examples.
Utility Methods
The library offers several utility methods to create and transform matrices. Some of these methods are inspired by other libraries. The utility test suite and examples directory demonstrate the use of these methods vividly. Some of these methods are given below
//algebra::Zeros(<size_t>,<size_t>) can be used to create a null matrix of specified order.const algebra::Matrix<int> null_matrix = algebra::Zeros(3,4);
//will result in a null matrix of order(3,4)000000000000//algebra::Ints(<size_t>,<size_t>,<int>) can be used to create a matrix of specified order and the given integer const algebra::Matrix<int> custom_matrix = algebra::Ints(4,4,5);
//will result in 5555555555555555//algebra::Identity(<size_t>) can be used to create an integer identity matrix of given order const algebra::Matrix<int> identity = algebra::Identity(4);
//will result in1000010000100001const algebra::Matrix<int> matrix1 = {
{1,2,3,4},
{5,6,7,8},
{9,10,11,12}
};
//Yields each row of the matrix in each iteration and applies the passed in function on the row. Permissible operations on algebra::Row<T> are listed above.//Does not modify the actual matrix.
std::vector<algebra::Row<int>> modified_rows;
modified_rows.reserve(3);
matrix1.For_EachRow([&](const algebra::Row<int>& _row){
modified_rows.emplace_back(2 * _row);
});
//std::vector modified_rows now contains {{2,4,6,8},{10,12,14,16},{18,20,22,24}}//Yields each column of the matrix in each iteration and applies the passed in function on the column. Permissible operations on algebra::Column<T> are listed above.//Does not modify the actual matrix.
std::vector<algebra::Column<int>> modified_columns;
modified_columns.reserve(4);
matrix1.For_EachColumn([&](const algebra::Column<int>& _column){
modified_columns.emplace_back(_column % 2);
});
//std::vector modified_columns now contains {{1,1,1},{0,0,0},{1,1,1},{0,0,0}}const algebra::Matrix<double> matrix2 = {
{1.2,3.4,5.6},
{7.8,9.10,11.12},
{13.14,15.16,17.18}
};
//Function algebra::Matrix<T> Map(const std::function<T(const T&)>& _lambda) transforms a matrix depending on its actual datatype and passed in function where T is type of real number.//Does not modify the actual matrix. const algebra::Matrix<double> sqrt_matrix = matrix2.Map([&](constdouble& _element) {
returnstd::sqrt(_element);
});
//Matrix sqrt_matrix would look like1.095451.843912.366432.792853.016623.334673.624913.893584.14488//Sample utility method, implemented utilizing algebra::Determinant datatype to compute the area of a triangle represented by the non-collinear 2D coordinate points defined with std::vector<std::pair<int,int>>.const std::pair<int,int> givenPoint1 = std::make_pair<int,int>(3,8);
const std::pair<int,int> givenPoint2 = std::make_pair<int,int>(-4,2);
const std::pair<int,int> givenPoint3 = std::make_pair<int,int>(5,1);
constdouble computedAreaOfTriangle = algebra::DeterminantUtility_AreaOfTriangle({givenPoint1,givenPoint2,givenPoint3});
//Yields the value 30.5
Apart from the aforementioned examples, several other utility methods are provided to wrap and unwrap an algebra::Matrix from and to simpler one-dimensional sequences e.g., flat C++ array and std::vector. Wrapping and unwrapping can be done along rows or along columns using enum types defined in algebra namespace. Several examples of the same are validated and demonstrated in the utility test suite and examples directory.
Installation
In this section, two ways are described to integrate Francois using CMake. First using the git submodule and the second using CMake FetchContent module.
//A simple CMake project setup
src/main.cpp
build
CMakeLists.txt
//#1 Integration using git submodule //
//Creating the root level git repository and adding the Francois as submodule
$ git init
$ git submodule add https://github.com/sarkarchandan/Francois.git extern/Francois
//A simple CMake configuration going inside CMakeLists.txt
cmake_minimum_required(VERSION 3.1...3.13)
if(${CMAKE_VERSION}VERSION_LESS 3.13)
cmake_policy(VERSION${CMAKE_MAJOR_VERSION}.${CMAKE_MINOR_VERSION})
endif()
project(Installation VERSION 1.0 DESCRIPTION "Installation of Francois"LANGUAGES CXX)
#Adding extern/Francois as subdirectoryadd_subdirectory(extern/Francois)
set(SOURCES src/main.cpp)
add_executable(${PROJECT_NAME}${SOURCES})
target_link_libraries(${PROJECT_NAME} Francois)
//#2 Integration using CMake FetchContent module //
//A simple CMake configuration going inside CMakeLists.txt
cmake_minimum_required(VERSION 3.1...3.13)
if(${CMAKE_VERSION}VERSION_LESS 3.13)
cmake_policy(VERSION${CMAKE_MAJOR_VERSION}.${CMAKE_MINOR_VERSION})
endif()
project(Installation VERSION 1.0 DESCRIPTION "Installation of Francois"LANGUAGES CXX)
include(FetchContent)
FetchContent_Declare(
Francois
GIT_REPOSITORY https://github.com/sarkarchandan/Francois.git
)
FetchContent_GetProperties(Francois)
if(NOT Francois_POPULATED)
FetchContent_Populate(Francois)
add_subdirectory(${francois_SOURCE_DIR}${francois_BINARY_DIR})
endif()
set(SOURCES src/main.cpp)
add_executable(${PROJECT_NAME}${SOURCES})
target_link_libraries(${PROJECT_NAME} Francois)
//Assuming the out of source build
$ cd build
$ cmake -G "Unix Makefiles" ..
$ make
Tests
Francois provides a grouped and illustrative test suite. The test suite, on the other hand, works as secondary documentation of the library features. It uses the googletest framework and CTest driver program to execute the tests. In this section, we describe the execution of the test suite assuming the steps described in the Installation section are performed. Based on which CMake integration option is used the location for the build directory of Francois may differ. Samples of both cases are given.
//#1 Assuming that git submodule option is taken to link the library //
$ cd build/extern/Francois/
$ make
$ ctest
//#2 Assuming that the CMake FetchContent module is used to link the library //
$ cd build/_deps/francois-build/
$ make
$ ctest
Examples
A humble set of examples is provided in order to demonstrate several features of the library. More are being added. Here are the steps to execute the examples.
//#1 Assuming that git submodule option is taken to link the library //
$ cd build/extern/Francois/
$ make
$ ./Example
//#2 Assuming that the CMake FetchContent module is used to link the library //
$ cd build/_deps/francois-build/
$ make
$ ./Example
Dependencies
CMake version 3.11
C++14
The library uses CMake FetchContent module in order to integrate googletest framework at configuration time. FetchContent is introduced with CMake version 3.11. The library uses features from C++ standard 14 in several places.
Authors
Chandan Sarkar
I am a grad student in computer science and learning C++ programming. This library is being created as part of my educational endeavors. I welcome all feedback and suggestions to improve my work.
License
This project is licensed under the MIT License - see the LICENSE.md file for details.