[Feature]: add support for Gumbel Distribution

[wiseaidev]

Hey everyone! I have been using this library for quite a while now, and I noticed that It doesn't support some distribution functions like Exponential and Gumbel. I think one way to add support for missing distributions is by migrating from rand to rand_distr. What do you think?

Edit 1: I am thinking about a relationship between, for instance, Poisson and Exponential Distributions. Probably there is one and I can compute one from another. Know what I am saying?

Edit 2: Oh, wait! It is called Exp.

[YeungOnion]

This could be useful.

This distribution would probably need to implement the following traits

• derive
  • Debug
  • Copy
  • Clone
  • PartialEq
• distribution::Continuous<f64, f64>
• distribution::ContinuousCDF<f64, f64>
• statistics::{Min, Max, Mean, Median, Mode}
• ::rand::distributions::Distribution<f64>
• statistics::Distribution

all methods seem like they should be based on closed form expressions, perhaps with considerations for numerical stability of

• double exponential, especially with pdf
• double logarithms, if directly using them in inverse_cdf

[YeungOnion]

Since it wasn't straightforward to find the exponential distribution as Exp could be a short PR to add a doc alias for "exponential". See rustdoc docs on #[doc] and in our crate, #174 and eb0745e

[soumyasen1809]

Hi @YeungOnion I am new to Rust and would want to take a shot at this issue. Is it possible to work on adding Gumbel distribution?

[YeungOnion]

Absolutely! Feel free to ping me here for any help. Also, as you're new, I'd like to mention the Rust Language forum.

For consistency, I request that the error type you create for this distribution should have an error message from Display with language that describes the unexpected input, as opposed to what it is not. Some truncated examples from other error messages in the project.

• "is NAN or is outside [0,1]" instead of "must be on [0,1]"
• "shape and scale are both inf" instead of "must have finite shape or finite scale"

[soumyasen1809]

@YeungOnion I have a PR for this issue
#285

[YeungOnion]

Closed by #285