Adding skew normal random variable

[saona-raimundo]

Background

What is your motivation?

Skew normal distribution is a variant of the normal distribution that allows a preference to some side of the mode. It has many applications and it is a very simple variant to work with. Personally, I used it in a biological model and implemented the sampling method to do some Monte Carlo computations.

What type of application is this?

Numerical simulation

Feature request

I would like to add SkewNormal to the rand_distr crate. This would implement Distribution<F> where F:Float, just as Normal does.

As for the sampling algorithm, I propose to follow this paper. The method is simple, but there is no benchmark against other algorithms. That would be a reason to choose another algorithm. Other algorithms might have performance issues in extreme parameters.

I volunteer to make a PR. I wanted to ask first about it.

[dhardy]

That sounds fine to me, and the algorithm looks straightforward to implement.

It would be good to have some tests (see #1121).

Aside: would you care to review #1136? (No pressure.)

[saona-raimundo]

That sounds fine to me, and the algorithm looks straightforward to implement.

Great! Thank you!

It would be good to have some tests (see #1121).

I will include tests like these ones then.

Aside: would you care to review #1136? (No pressure.)

Sure, no problem :) I will not take more than a week.

[vks]

Implemented in #1174.