Matrices are a fundamental building block in many computational applications, particularly in the realm of scientific computing (e.g. CFD – Computational Fluid Dynamics) and machine learning. The architecture of a matrix class must be carefully crafted to:<\/p>\n\n\n\n
In this post we’re going to create a matrix class that can perform:<\/p>\n\n\n\n
In later posts, we will use this matrix class to create a simple neural network that will learn how to perform the XOR function, and then we’ll cover how we can use the GPU to perform our matrix operations in parallel.<\/p>\n\n\n\n
So how are we going to represent the actual matrix? For now, we will only be working with M<\/em> x N<\/em> matrices (i.e. a 2 dimensional array), where M<\/em> is the number of rows, and N<\/em> is the number of columns. We could use a raw array, an A one dimensional array (or vector) is often the most practical data structure to use in a matrix class. If using a one dimensional array to represent a two dimensional matrix, the index can be calculated using:<\/p>\n\n\n\nstd::vector<\/code>, or even more complex data types.<\/p>\n\n\n\n