Bart van Erp

For a project that I am working on, I would like to be able to send out mixture distributions as messages. Although this PR mentions Gaussian mixture messages only, it...

I am not sure whether this should be a flag for Gaussian nodes. I think it is better to always constrain the `cholinv` method. Cholesky by definition only applies for...

I agree that sometimes we end up with improper distributions, from which the covariance/precision matrix is not positive definite. However, this does not mean that the matrix is not Hermitian...

I am not sure whether I am following entirely here, so better to discuss this in person. >There are some matrices that do have a proper inverse, but they are...

This issue may be solved by the addons in issue #172 and PR #215. Basically (once completed) it should be possible to save extra information in messages and marginals. This...

In #215 we introduce so-called `addons`. These yield extra information and are propagated in the message. A `message` is now (approximately) defined as: ```julia struct Message{D,A} distribution::D addons::A end ```...

Seems interesting, thanks for sharing!

Thanks for starting this PR. The error was caused because the `out.jl` file for the `Exponential` node was not included under `prototypes.jl`. Hence, the package did not load the rule...

@bvdmitri Can we close this PR? I think the exponential mapping will be resolved with the `Delta` node. If someone needs an `Exponential` distribution, then we will create an issue...

Also check for `*(Adjoint{ T, StandardBasisVector{T}}, AbstractMatrix)` and extend get/set for `Adjoint`