Yet Another Vector-Matrix Library

As the title says, this is just another vector/matrix library for numerical calculations in C++. The impetus for this library stemmed from my dissatisfaction with the current barrage of vector/matrix libraries. If the library was not a toy program without sanity checks, it was too specialized to $R^3$, relied on non-standard compiler behavior, or had a non-standard Open Source license. Thus, I wrote my own with tests and a means to simply install.

My end goal is to have light weight wrapper around a C “array” (which are just pointers and not real arrays (I'm a Fortran programmer)) that understands the usual mathematical operations and interoperates with OpenGL calls. This means two primary things. First, all of the common mathematical operations must be transparently supported. This includes addition, subtraction, multiplication, and division with scalars, vectors, and matrices. The scalars should broadcast element-wise as in Fortran. Second, the underlying data must be tightly packed.

The advantage of needing interoperability with OpenGL is the arrays must be in column major order. This buys us the same ordering as Fortran and, therefore, access to the usual BLAS libraries.

This library is a header only library that provides a mathematical vector and matrix template class for C++. The dimensions and types are template parameters so that an arbitrary length vector or sized matrix may be used. The library also provides specializations for $R^2$, $R^3$, and $R^4$ via the templates vec2, vec3, and vec4 respectively.

Installation

The library can be installed by simply placing the headers in the include directory on your compiler’s search path. Alternatively, you can use CMake to install the library using

$ mkdir build
$ cd build
$ cmake ..
# make install

You can run the tests using

$ make test

You can disable building the tests using

$ cmake -DENABLE_TESTS:BOOL=Off ..

Usage

Using the library is as simple as including the headers and declaring your vector

#include "vecmat/vector.hpp"
#include "vecmat/matrix.hpp"

...

vecmat::vector<3, float> xhat {{1.0, 0.0, 0.0}};
vecmat::vector<3, float> yhat {{0.0, 1.0, 0.0}};
vecmat::vec3<float> zhat = vecmat::cross(xhat, yhat);

vecmat::matrix<3, 3, float> M {{1.0, 2.0, 3.0,
                                4.0, 5.0, 6.0,
                                7.0, 8.0, 9.0}};

vecmat::vector<3, float> c = 1.1 * xhat + 2.8 * yhat + 4.4 * zhat;
vecmat::vector<3, float> d = vecmat::dot(M, c);

...

Note the vector and matrix use the aggregate style initialization. Further, the elements of the matrix are stored in column-major order. This means the matrix in the above is stored as

M =
\begin{pmatrix}
    1.0 & 4.0 & 7.0 \\
    2.0 & 5.0 & 8.0 \\
    3.0 & 6.0 & 9.0
\end{pmatrix}.

A major underlying assumption is the type contained in the vectors and matrices are real numeric types. Complex numbers are not considered and non-numeric types do not make sense mathematically. The elements are stored in a C array named data that can be passed to functions that need a pointer to contiguous column major data. Although the data may be accessed directly, the usual [i] operator is provided for abstracting away the access. Additionally, the (i) operator access the elements on the vector and a symmetric (i, j) operator access the i-th row and j-th column of a matrix.

License

This work is licensed under the 3-clause BSD license.

kprussing/vecmat

Yet Another Vector-Matrix Library

Installation

Usage

License