NaN Particles

[mauro3]

This seems inconsistent:

julia> NaN ± NaN
ERROR: ArgumentError: Normal: the condition σ >= zero(σ) is not satisfied.

julia> NaN ± Inf
Particles{Float64,2000}
 NaN ± NaN

[baggepinnen]

It comes from this

julia> std(rand(Normal(0,Inf), 10))
NaN

julia> Normal(0,NaN)
ERROR: ArgumentError: Normal: the condition σ >= zero(σ) is not satisfied.

[mauro3]

It's feels a bit "wat?", but it's ok to close this.

[baggepinnen]

Do you have an application where the semantics for infinite ra dom numbers is important? If not we can probably close this issue as I'm not sure if it makes sense to modify the behavior to give something other than nan.

[mauro3]

No, my worry was more that NaN \pm NaN errors. I got data at times which contains NaNs (arguably I should use missing) and that is how I came across this. But I can work around it. Well, it might be worth having a coherent story for 1.0.