Simulations and Experiments for Graphical Continuous Lyapunov Models (2020)

This repository contains the files for replicating the simulations and the examples in Graphical Continuous Lyapunov Models, G. Varando and N. R. Hansen 2020.

Required packages

The following packages must be installed:

• gclm
• glmnet
• glasso
• igraph
• MASS
• ggplot2

All packages are available from CRAN.

Simulations

Run the simulations

To run the simulations just move in the directory and run the R script simulate.R. The simulations can take up to a couple of days, depending on the machine. The script simulate.R accepts arguments if executed from a terminal with Rscript to allow for a simple parallelization of the simulations using, for example, tmux.

The exact simulations as in the paper are the ones that can be executed from the script run_simulate.sh.

Plot results

Before plotting the results you need to compute the evaluations and aggregate the results with the R script aggregate_results.R or directly using the script run_aggreagate.sh.

Then you can execute the plotting scripts plot_simulated.R, plot_marginalized.R and plot_times.R to generate the same plots as in the paper.

Protein-signaling network

First run the script proteins.R to generate the results.

The estimated graphs can be obtained running plot_proteins.R. Graphs are saved both as pdf and as tikz format.