Hacker News Reader

Probably wise: there are too many ways to do it, each with different behavior for ill-conditioned matrices.

GLSL includes an inverse for fixed-sized (2,3,4) matrices, which can return NaNs for ill-conditioned inputs. But it's fine for inverting transformation matrices.

At least there’s always LAPACK if you need it!

Are there any matrix operations in the built-ins?

Most numerical operations work on array elements, so things like A+B work as expected with matrices. Not A*B, because that is element-wise multiplication and the matrix product is matmul().

The Euclidean norm norm2() should also work as expected as well.

matmul