News for hackers. Stuff that matters.

A comment on Herbie: Automatically Improving Floating Point Accuracy.

Re:Herbie: Automatically Improving Floating Point Accuracy by ImaCake on Thursday August 06, 2020 @05:58PM (#24075918)
This seems like a great tool for certain statistical applications. I am learning a bit about the controversey surrounding extremely small p-values in Genome Wide Association Studies (think a million n ANOVA) and machine precision. I guess tools like this would be very useful to make sure your statistical tests can more accurately calculate p-values approaching the limits of machine precision.
[ Parent | on HN ]
Re:Herbie: Automatically Improving Floating Point Accuracy by contravariant on Thursday August 06, 2020 @07:06PM (#24076354)
If your p-values are approaching machine precision then you don't need better floating point precision you need to be using the log-likelihood (admittedly for gaussian samples this is the variance which requires some careful handling).

Also if your p-values are approaching machine precision, then maybe p-values aren't the most meaningful metric.

[ Parent | on HN ]