Maximum likelihood estimation of vine copulas models with automatic differentiation.
C++ R Shell
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
DVineAD
.gitignore
LICENSE
README
Vine-AD.pdf
buildvine
reload.R

README

The goal of this project is to demonstrate that by using automatic differentiation
tools, the maximum likelihood estimation of vine copula models can be done significantly 
faster than the current practice. This package is associated to the submitted (as a code snippet)
article to the Journal of Statistical Software "Faster Estimation of High-Dimensional Vine 
Copulas with Automatic Differentiation" authored by Hongbo Dong, Wei Li and Haijun Li (Washington
State University). A pdf file of the submitted version is Vine-AD.pdf.

If you have successfully installed the ADOL-C software, you will then be able to build/install our
R package DVineAD with the bash script buildvine. It will ask you for location information of your
ADOL-C (https://projects.coin-or.org/ADOL-C) installation. Note that for now we only tested the code
on linux systems. We have seen compiling issues when installing ADOL-C on Mac and Windows systems.
We think such issues can be fixed with some efforts, but we have not yet persued further.