julia> using AdvancedVI, Distributions; mv = MvLocationScale(zeros(2), 1.0, Normal());
julia> rand(mv)
2-element Vector{Float64}:
1.1396187240493345
-0.23614109843834447
julia> rand(mv, 3)
2×3 Matrix{Float64}:
-0.843996 0.73772 -1.05117
0.356957 -0.909233 0.829167
julia> rand(mv, 3, 4)
3×4 Matrix{Vector{Float64}}:
[-0.138124, 0.335119] [-1.36683, -0.0922746] [-0.965939, -0.229622] [1.70725, -0.243339]
[1.20617, -0.659821] [-0.0097997, -1.38867] [0.460145, 0.352632] [-0.785098, -1.82992]
[0.726249, 2.11884] [0.634132, 0.491341] [0.718743, -0.603134] [0.572106, -2.25754]I would have expected either the second one to return a vector-of-vectors, or the last one to return a 2x3x4 array, but right now they're kind of inconsistent.
That said, while writing up this issue, I discovered that Distributions.MvNormal has exactly the same behaviour! So I'm not sure if this is an AdvancedVI issue per se...
I would have expected either the second one to return a vector-of-vectors, or the last one to return a 2x3x4 array, but right now they're kind of inconsistent.
That said, while writing up this issue, I discovered that
Distributions.MvNormalhas exactly the same behaviour! So I'm not sure if this is an AdvancedVI issue per se...