Moritz Schauer

Oh, there is no fault in coordinating these kind of things. So I would like to keep in mind, that there are plenty of connections, so for example a Data...

If you find this tedious to work out because of different parametrisations and notations here it is already in Julia https://discourse.julialang.org/t/seven-lines-of-julia-examples-sought/50416/26?u=mschauer

It is arguably also introducing a bug: ``` julia> eltype(dict) Pair{Int64, Int64} ``` ``` julia> collect(dict) 80200-element Vector{Pair{Int64, Int64}}: ``` ``` julia> sort(dict, by=x->x.second) ERROR: type Int64 has no field...

Data point: even though I commented above, I forgot about this issue and now I notice that I have code which relies on this observable behaviour.

Ah, great! I plan to add the Cholesky factorization, I am also using FixedSizedArrays for filtering and MCMC in the widest sense. I could also add the `randn()`-functor from https://github.com/mschauer/Bridge.jl/blob/master/src/fsa.jl,...

I can answer some of this from memory: - just use `(Σ+0.0000000001*I)` - `rand(Vec{d,Float64})` is native, see src/constructors.jl#L142 - `\` uses native inversion

Ah, yes. For now I would rather add `randn(Vec{d,Float64})` from https://github.com/mschauer/Bridge.jl/blob/master/src/fsa.jl ``` immutable RandnFunctor{T}

@michaellindon You see, it is work in progress, but your "trying the ice to see if it holds" is very helpful for us.

I make a PR this weekend if c42f does not beat me to it.

Actually, it should be a different function name alltogether because it modifies the second argument as well and therefore does not follow the push! or append! interface at all.