rand5

Certainly happy to help- with the caveat that I don't have any prior experience working with these file types.

Speaking of labeled components and connectivity, I think that for people transitioning from MatLab, equivalents to bwperim, regionprops, and imfill would be highly desirable.

+1 I would also find this functionality incredibly useful.

I'm having trouble seeing that this is an issue with the corners. What struck me was not that the results were different if `θ = -2.5` versus `θ = 2.5`,...

That is a good point. I just don't see any reason for it to blow up to `Inf` in the region (approx 1.14 to 4.15) it does based on the...

Sorry for the confusion--I'm taking the equation for the convolution, $f_{Z}(z) = \int_{-\infty}^{\infty} f_{XY}(x,z-x)dx$, and looking at the integrand when setting $z=0$, i.e. $f_{XY}(x,-x)$ as a function of $x$. If...

I tried a few other copulas (Clayton, Gumbel, Joe) and am getting similar domain errors. Would this need to be addressed for `C::ArchimedeanCopula` as well?

Here is what I was using to check the Archimedian Copulas ``` X1 = Normal() d = 2 θ = 1. Cs = [ClaytonCopula(d,θ), FrankCopula(d,θ), GumbelCopula(d,θ), JoeCopula(d,θ)] #Copulas to check....

@lrnv I'm using v0.1.27 and still getting domain errors: ``` X1 = Gumbel(15.,5.) X2 = MixtureModel([Normal(-6.,0.1),X1],[0.9,0.1]) θ = 2.5 d= 2 C = GumbelCopula(d,θ) D1 = SklarDist(C,(X1,X2)) N = 100...

``` pdf(D1,[-100.,100.]) # = NaN ``` Edit: this error is being thrown despite the fact the test code above it works fine