How to sample multiple self defined random variable

General Usage
1

I have defined the following Distribution,

using Distributions

struct  ComplexNormal <:ContinuousUnivariateDistribution
    μ::ComplexF64
    σ::Float64
    function ComplexNormal(μ=0,σ=1) new(μ,σ) end
end


function Base.rand(d::ComplexNormal)
    randn(ComplexF64)*d.σ+d.μ
end

I’m trying to figure out how to make rand(ComplexNormal(),2) and rand(ComplexNormal(),2,2) work.

According to the documentation, The package already implements a vectorized version of rand! and rand that repeatedly calls the he scalar version to generate multiple samples. Thus I probably do not need to define Base.rand(::ComplexNormal, n::Int), but after googling for 1 hour, I couldn’t figure out how to do it.

Any help will be greatly appreciated, thank you!

2

https://docs.julialang.org/en/v1/stdlib/Random/#Generating-random-values-of-custom-types

3 Likes
3

Thank you, certainly gets me closer to what I wanted.

using Distributions,  Random

struct  ComplexNormal <:ContinuousUnivariateDistribution
    μ::ComplexF64
    σ::Float64
    function ComplexNormal(μ=0,σ=1) new(μ,σ) end
end

Base.rand(rng::AbstractRNG, d::ComplexNormal) =randn(ComplexF64)*d.σ+d.μ;

rand(ComplexNormal(1+2im,1)) generates a single element works.
rand(ComplexNormal(1+2im,1),2) don’t quiet, as it gives
InexactError: Float64(0.37414329354088083 + 1.039498256924925im)

This is probably outside of what I asked in this post, I’m going to start a new one.

4

I am wondering if you actually read the docs though. You are probably looking for SamplerTrivial, as in

https://docs.julialang.org/en/v1/stdlib/Random/#A-simple-sampler-without-pre-computed-data

1 Like
5

Yes, I do noticed that I can achieve what I want with the following

using Random
struct CNormal
    μ::ComplexF64
    σ::Float64 
end

Random.rand(rng::AbstractRNG, d::Random.SamplerTrivial{CNormal}) = randn(rng,ComplexF64)*d[].σ+d[].μ


rand(CNormal(0,1),2)

However, I have not yet figure out how to make this work if I want CNormal <:ContinuousUnivariateDistribution

The following is a solution, but there probably is a way to make the above work

using Distributions, Random

struct  ComplexNormal <:ContinuousUnivariateDistribution
    μ::ComplexF64
    σ::Float64
    function ComplexNormal(μ=0,σ=1) new(μ,σ) end
end

Base.rand(rng::AbstractRNG, d::ComplexNormal) = randn(rng,ComplexF64)*d.σ+d.μ;
Base.eltype(::Type{ComplexNormal}) = ComplexF64