VineCopulas.jl, toured with KaimonSlate.jl

This post highlights two recent Julia projects:

• VineCopulas.jl, developed by Santiago Jimenez;
• KaimonSlate.jl, developed by Kahli Burke.

Kahli used KaimonSlate to produce the video and PDF below about VineCopulas.jl.

A new package for vine copulas in Julia

Vine copulas build high-dimensional dependence models from bivariate copulas. Each edge of a sequence of trees can carry its own pair-copula, allowing different dependence strengths, asymmetries, and tail behaviours within the same model.

Santiago’s VineCopulas.jl is a pure-Julia implementation built on top of my baby Copulas.jl. It supports C-vines, D-vines, and general R-vines, and integrates them with the familiar Julia distribution workflow.

The package can construct explicit vine models, evaluate densities, simulate observations, and apply Rosenblatt and inverse Rosenblatt transforms. It also provides pair-copula conditional primitives, truncated C- and D-vines, and exchange helpers for R-vine matrices.

The video gives a quick sense of the objects involved:

The visual tour

The companion document covers the bivariate building blocks, C-vines, D-vines and R-vines, the Rosenblatt transform, truncation, quasi-Monte Carlo simulation, and an application to basin-wide flood risk. It includes both Julia code and visualizations.

Download the 16-page VineCopulas.jl visual tour (PDF)

KaimonSlate.jl

KaimonSlate is a reactive computing environment for Julia supporting code, prose, equations, plots, widgets, media, presentations, and exported documents. More examples are available in Kahli’s Slate portfolio.

Kudos to both of them! 🙂

Links: VineCopulas.jl repository · VineCopulas.jl documentation · VineCopulas.jl announcement · Kahli Burke’s Slate portfolio · KaimonSlate.jl repository · KaimonSlate.jl documentation · KaimonSlate.jl announcement