Is it true that all affine transformations have a last row of the form `(0, 0, .., 0, 1)`

, and therefore could be stored in a n by (n+1) matrix?

Hi! Yes, all affine transformations can be represented as `n × (n + 1)`

matrices. However not all `n × (n + 1)`

matrices can be interpreted as affine transforms. For example the matrix filled with zeros does not represent an affine transform because that would not be a bijection.

My motivation is to make composing them faster. Was wondering if you could reduce the number of ops required if you used an alternative repr.