John Femiani wrote:
I noticed one file (http://www.boost.org/boost/random/normal_distribution.hpp) asked for mathematical constants (pi, e, phi, etc) in boost.
Which constants other than perhaps those would the boost community want? Is there interest in a library of mathematical constants?
It also seems that there is no statistics library (i.e. functions to compute mean, variance, moments, skew, kurtosis, median, mode, quartiles, etc. from samples or histograms). Would the boost community be interested in these sorts of functions?
There's some activity in this area: Eric Neibler's accumuators lib is up for review: includes calculation of various statistics from sample data (mean variance etc): see the vault for code and docs. Paul Bristow and I have been working on special functions / statistical distributions: www.johnmaddock.co.uk/toolkit. This will let you perform statistical tests for significance etc once you have the basic stats. I hope to get this in the review queue in a week or so (just polishing docs now). Numeric constants have a long history around here: you should find some from Paul Bristow in the vault, and a few in the Math.Toolkit above, which we've deliberately avoided documenting! The basic problem is that no one around here could agree on an interface :-( It's one of those bicycle shed issues that we don't want to be held up by. If you think you can cut through the design problems and satisfy the competing demands on numeric constants (look through the list archives and you should find lots of discussion) then go for it :-) Otherwise I suspect Paul and I could always use more help on the toolit code: more of the TR1 special functions are needed (I have some notes that may be useful if that interests you), plus we could really use some "guinnea pig" testers/documentation checkers should anyone want to volunteer! Cheers, John Maddock.