Andreas Noack

This might be useful for the gaussian distribution, but I it doesn't really generalize well as other distributions don't have as nice properties when conditioning. Please also take a look...

Please search the issue tracker for several similar issues. I think this issue might be redundant.

Which version of Distributions are you using? On master I'm getting ```julia julia> typeof(rand(Exponential(θ))) == typeof(θ) true ``` and that is a consequence of the commit that you link to...

> in our code base we basically implicitly tested that `eltype(distribution) == typeof(rand(distribution))` I think it's best to avoid `eltype(::Distribution)` completely. Please refer to the many long discussion in the...

My point is that this is an open issue and part of the discussion here is just repeating arguments from other issues. > obtaining the exact sample type without having...

We have five test dataset where we could reproduce the saddle point. When running the test script above with those datasets, `COBYLA` always avoids the saddle point solutions that `BOBYQA`...

Because of #833, I revisited this issue. I just tried out the PRIMA backend and the issue persists. I.e. ```julia N = 500 _rng = Random.seed!(Random.default_rng(), 124) fts = map(1:N)...

The model might not make sense, but you are in a better position to tell the FDA than I am. However, even if the model doesn't make sense, it seems...

There are two separate issues here. Both are interesting but I think it might be worthwhile to consider them separately 1. Does the model make sense and does it do...

@ararslan I think the value would have to be sufficiently small to not impose a restriction on the model since the components cannot be negative. The estimated sigmas in the...