"From ... to" method for creating beta distribution

[erwald]

Currently you can create normal and lognormal distributions using x, y, and credibility parameters, e.g.:

def norm(x=None, y=None, credibility=90, mean=None, sd=None,
         lclip=None, rclip=None):

but the beta distribution initializer only takes alpha/beta values:

def beta(a, b):

It might be nice to also have the previous interface available for beta, like here.

• x and y would need to be in the 0 <= i <= 1 range.
• It'd either be a breaking change, or lead to inconsistent distribution creation interfaces. I.e. x, y, and credibility can't be the default arguments -- like they are for norm and lognormal -- without making this a breaking change.
• Apparently calculating this is nontrivial, and maybe not that fast, which could be another problem.

[peterhurford]

I would like to see https://nunosempere.com/blog/2023/03/15/fit-beta/ implemented in squigglepy as an additional interface to beta. I'll have to think about how to best handle the important trade-offs you mentioned. Adding lclip and rclip sound reasonable too.

[NunoSempere]

You can find an embryonic implementation of this in this pull request. Embryonic because it works in python, but the glue code for integrating it into Squigglepy is missing. Also see more details in the README.md in the pull request.