Unable to sample from VonMisesFisher

[nishadgothoskar]

I'm encountering issues when sampling from a VonMisesFischer centered on the identity orientation.

julia> rand(Distributions.VonMisesFisher([1.0, 0.0, 0.0, 0.0], 100.0))
4-element Vector{Float64}:
 NaN
 NaN
 NaN
 NaN

Digging deeper, I found the issue is here:

julia> s = Distributions.sampler(Distributions.VonMisesFisher([1.0, 0.0, 0.0, 0.0], 100.0))
Distributions.VonMisesFisherSampler(4, 100.0, 0.007499578172454263, 0.9851124936725868, 87.9465773626279, [NaN, NaN, NaN, NaN])

where the NaN vector is output by this function:

Distributions.jl/src/samplers/vonmisesfisher.jl

Line 87 in d6985ad

function _vmf_householder_vec(μ::Vector{Float64})

Any help here would be appreciated!

(Unrelated issue related to VMF is the following, where it should be abs(norm(v) - 1.0) < 1.0e-12

Distributions.jl/src/utils.jl

Line 14 in b4ce282

isunitvec(v::AbstractVector) = (norm(v) - 1.0) < 1.0e-12

[devmotion]

Thanks for the issue. It seems to be a duplicate of #1423.