The result of multiplying a simple vector by a matrix

is a geometric transformation. Here’s a 90° anti clockwise rotation

That is to say

the resulting vector is a linear combination of columns of times rows of (four-ways-to-look-at-matrix-multiplication).

The role of each matrix element in the transformation is better understood by interpreting matrix vector multiplication as basis transformation. For example, in a 2 dimensional plane with and axis, the basis is defined as

where the coefficients identify the amount of movement—the standard unit—that happens on their respective axis, given unit of movement along the corresponding axis in the multiplied vector, and it can be read like

for every unit along the axis move along the axis and along , and for every unit along the axis move along the axis and along

In other words, each matrix column corresponds to an input dimension—what input dimension gets transformed—each matrix row to an output dimension—how the input dimension is transformed

A look at the matrix representation of the familiar Cartesian plane makes it clearer

The first column is the axis, the second the .

The initial transformation can thus be interpreted as moving unit along the axis (and along the axis) for every unit of movement along the axis and along the axis (and along the axis) for every unit on the axis.