Add type Sampler & redefine EigvalsSampler & EigvecsSampler

src/distributions.jl

@@ -1,36 +1,35 @@
 using Distributions: Distribution
 using LinearAlgebra: qr
-using Random: AbstractRNG, Sampler, default_rng
+using Random: AbstractRNG, default_rng
 
 import Random: rand
 
 export EigvalsSampler, EigvecsSampler
 
-struct EigvalsSampler{R<:AbstractRNG,T,D<:Distribution} <: Sampler{T}
+abstract type Sampler{R<:AbstractRNG,D<:Distribution} end
+
+struct EigvalsSampler{R,D} <: Sampler{R,D}
     rng::R
-    type::T
     dist::D
-    function EigvalsSampler(dist, type=Float64, rng=default_rng())
-        return new{typeof(rng),typeof(type),typeof(dist)}(rng, type, dist)
+    function EigvalsSampler(dist, rng=default_rng())
+        return new{typeof(rng),typeof(dist)}(rng, dist)
     end
 end
 
-struct EigvecsSampler{R<:AbstractRNG,T,D<:Distribution} <: Sampler{T}
+struct EigvecsSampler{R,D} <: Sampler{R,D}
     rng::R
-    type::T
     dist::D
-    EigvecsSampler(dist, type=Float64, rng=default_rng()) =
-        new{typeof(rng),typeof(type),typeof(dist)}(rng, type, dist)
+    EigvecsSampler(dist, rng=default_rng()) = new{typeof(rng),typeof(dist)}(rng, dist)
 end
 
 function Hamiltonian(s₁::EigvalsSampler, s₂::EigvecsSampler, m::Integer)
     evals, evecs = sort(rand(s₁, m)), collect(rand(s₂, m, m))
     return Hamiltonian(Eigen(evals, evecs))
 end
 
-rand(s::EigvalsSampler, dims::Integer...) = rand(s.rng, s.dist, dims...) .* oneunit(s.type)
+rand(s::EigvalsSampler, dims::Integer...) = rand(s.rng, s.dist, dims...)
 function rand(s::EigvecsSampler, m::Integer, n::Integer)
-    matrix = rand(s.rng, s.dist, m, n) .* oneunit(s.type)
+    matrix = rand(s.rng, s.dist, m, n)
     q, _ = qr(matrix)
     return collect(q)
 end