Making sense of vectorizing GP/PARI code (Episode #6) 

One Minute One-liners for the curious: https://www.youtube.com/@1minute1liners

In this YouTube short from One Minute One-liners for the curious(@1minute1liner), I compare different ways to vectorize GP/PARI code using an erfinv approximation function as the template. I start with a one-liner using the approximation from Winitzki and written in Matlab:

u=log(1-x.^2);u1=2/(pi*0.147)+u/2;y=sign(x).*sqrt(-u1+sqrt(u1.^2-u/0.147));

Below are three illustrative functions using different vectorizing methods in GP/PARI.

Links to supporting documents

• Erfinv approximations: https://www.academia.edu/9730974/A_handy_approximation_for_the_error_function_and_its_inverse?auto=download