Just in case this informations is useful…
I am using this function to get the significance and I found out that after about a value of 5.47955939024333727e+00 for sigma it starts returning 0’s.
I am going to use the Erfc to scan over sigma values, and match until match them with the p-value I have.
there is a large numerical error in the function for small value of prob. We need to fix this. I would recommend you to use instead
double s = ROOT::Math::normal_quantile_c(prob/2,1.0)/ sqrt (2); // = TMath::ErfcInverse(prob);
which returns the inverse of the complement cumulative normal distribution
(i.e. right tail integral of the gaussian)
Thank you for your report