Approximating the Inverse Normal

Beasley-Springer-Moro algorithm for approximating the inverse normal.

Description

Applying the inverse transform method to the normal distribution entails evaluation of the inverse normal. This is the Beasley-Springer-Moro algorithm for approximating the inverse normal.

Input: u, a sacalar or matrix with elements between 0 and 1
Output: x, an approximation for the inverse normal at u

Reference:
Pau Glasserman, Monte Carlo methods in financial engineering, vol. 53 of applications of Mathematics (New York),
Springer-Verlag, new York, 2004, p.67-68

• MATLAB release: MATLAB 7.10 (2010a)

[Ben Petschel]

any comments on the speed or accuracy compared to norminv? e.g. at u=0.917 the values differ by about 1-e9.

[Aris]

Amazing implementation! Thank you very much!