simulations/main_simulation/README.md

In mlondschien/hdcd: High - Dimensional Change Point Detection

README

This folder contains files required to replicate the main simulation results (Tables 1, 2) of Change point detection for graphical models in the presence of missing values. The file simulation.R needs to be called with 14 arguments comment, n, p, Network (one out of "RandomNetwork", "ChainNetwork"), NA_method (one out of "loh_wainwright_bias_correction", "complete_observations", "pairwise_covariance" and "average_imputation"), optimizer (either "line_search" for BS or "section_search" for OBS), deletion (either "mcar" or "blockwise"), delete_fraction, seed, nrep, n_of_segments, segment_lengths, e.g.

R --vanilla --slave <simulation.R rfcd_9 500 100 RandomNetwork pairwise_imputation section_search mcar 0.1 2 100 4 70 120 120 190 >

Note the trailing whitespace.

The results of the simulation will be stored in an rds file, names via the parameters.

The files simulation_line.txt and simulation_section.txt contain bash code to run simulations for different setups.

The aggregate_simulation_results.R file contains code to read in the rds files from the simulations, aggregate the simulation results, and produce latex code for tables 1 and 2. Note that it might be necessary to adjust directory.

mlondschien/hdcd documentation built on Jan. 5, 2021, 11:26 p.m.