What is the best way to apply inverse Gaussian CDF to an array of floating point values?
I was looking for something similar to MKL's v?CdfNormInv. There seems to be no direct equivalent in ACML.
Relevant question. What is underlying algorithm in DRANDGAUSSIAN? If it inverse CDF, I might be able to use it as an replacement for v?CdfNormInv.
The ACML DRANDGAUSSIAN routine uses the Box Muller method to generate the distribution. It will not do what you are after.
The NAG Statistical libray provides a Inverse CDF. There are other options, AS241 from Statlib, and GSL has this capability.