Carlos Parada

> I'm feeling a bit under the weather today and will try to respond more tomorrow or the next day, but just wanted to thank everyone for chiming in here....

It looks like PackageCompiler.jl might not work for our specific use case, so I think having both R and Julia versions is the best way to move ahead with this.

I've built the code to fit the Generalized Pareto distribution in Julia; let me know if/how/where you'd like to see it.

> > I've built the code to fit the Generalized Pareto distribution in Julia; let me know if/how/where you'd like to see it. > > Awesome, thanks! Do you have...

> > The code fitting the distribution is [here](https://github.com/ParadaCarleton/JuLOOa/blob/main/README.md), but I haven't built any code hooking it up to anything else since I'm still trying to figure out the internals...

> That seems fine, I doubt it matters in practice. I forget why we used `x[floor(n/4 + .5)]`. @avehtari do you remember if we chose to use that instead of...

> > I forget why we used x[floor(n/4 + .5)]. @avehtari do you remember if we chose to use that instead of using quantile() for a particular reason? > >...

In any case, having benchmarked the Julia code, a run of `gpdfit()` in Julia takes about a third as long on average as in R when fitting samples of size...

> > but I think this only requires changing one line in the documentation > > It's implemented and documented in more than one place, so there is more work....

> . If you want see more proof why the Julia approach is not as easy/straight forward as you think, check [this](https://discourse.julialang.org/t/julia-mojo-mandelbrot-benchmark/103638) out. The only version, contract the most people...