Carlos Parada

While working through the NestedSamplers integration for Turing, I needed a way to get quantiles for the prior. This is a bit awkward for multivariate distributions, where I need a...

Add Elliptical Distributions

2

We should probably add a type that lets users specify an arbitrary elliptical distribution in the vein of `MvNormal` by providing a generator, a location vector, and a scale matrix....

Combine Basic functions shared by Univariate/Multivariate

1

While working on #1442 I was bitten by method ambiguities caused by having separate (identical) methods for univariate and multivariate distributions. For instance, I can't define one `logpdf` method for...

Error thrown when using `kldivergence` with infinite values

3

``` julia> kldivergence(TDist(1), Normal(0, 1)) ERROR: DomainError with -0.9999999999999964: integrand produced NaN in the interval (-1.0, -0.9999999999999929) ``` It makes sense why this happens (the divergence is infinite), but I...

Allow Reflections by Removing Positivity Constraint for `LocationScale`

1

Many asymmetric distributions are more easily expressed with a location-scale transformation that has a negative scale parameter. For instance, the lower tail of a variable can be modeled as a...

How does BAT differ from Turing?

10

Hi! Found this on Github and I'd like to say, BAT.jl looks really interesting, and I'm always happy to see new tools for Bayesian analysis in Julia! I have a...

Dropping this paper on autocorrelation-reducing techniques for HMC by sampling with anticorrelated proposals here: http://proceedings.mlr.press/v108/piponi20a/piponi20a.pdf Closely related to the QMC suggestion in another issue earlier, but more likely to yield...

Evaluating at endpoints of open interval

1

The code seems to be creating errors by trying to evaluate at endpoints of an open interval, when the function might not exist at these endpoints: ``` julia> d (0.0..3.0...

Fan plots are a great addition to any statistical visualization package. Rather than emphasizing a single level of confidence, like a 95% interval, a fan chart displays a continuum of...