This repo is to accompany the paper "Entropic Trace Estimates for Log Determinants", which can be found here. The work is a part of an on going collaboration between the MLRG at Oxford (UK) and the Department of Data Science at Eurecom (France).
Currently the code is only the main parts used on the paper. However, over the coming weeks we aim to have some easy to full worked demos.
The UFL datasets can be downloaded from here.
In order to recreate the GMRF experiments we modified the third party code from the paper "Efficient Computation of Gaussian Likelihoods for Stationary Markov Random Field Models", Guinness and Ipsen 2015.
For any queries related to the code please do not hesitate to contact the authors.