Matrix Inversion ANA

[Paul Dale]

This algorithm might be faster but I have doubts it will be as numerically stable as the builtin function. There is no pivoting involved which is a good sign it will be unstable for some inputs. I'm also unsure it will necessarily be more resource efficient, you need a second matrix of the same size as the input and it is possible to do a matrix inversion insitu with a temporary vector of length n.

The 34S firmware does both of these -- it uses the Doolittle LU decomposition with partial pivoting as the basis for its matrix inversion. This is what HP used in 15C and I suspect it is what they still use -- writing numeric code is difficult and theirs was designed by one of, if not, the best.

Writing programs like this is always a bit of fun


- Pauli