/LGCP-normConst-simulations

Results and source code for Jullum, M. (2020) Investigating mesh based approximation methods for the normalization constant in the Cox process likelihood, Stat, in press

Primary LanguageHTML

LGCP normalization constant approximation - simulation source code

This repository contains the results and source code for the simulation experiments in the paper

Jullum, M. (2020) Investigating mesh based approximation methods for the normalization constant in the Cox process likelihood, Stat, in press, doi.org/10.1002/sta4.285

The preprint of the paper is available here, and the supporting information (preprint) here

Main files

  • source_code/help_functions.R Functions containing the implementation of all the differnet approximation methods for the integral in question, in addition to a series of help functions related to the simulation experiments.
  • source_code/full_simulations_script.R The script used to carry out the full simulation experiment in the paper. The script also contains the short test version that quickly runs trough some parameter combinations (to check that the script works as intended).
  • source_code/permutation_tests.R The script used to carry out the permutation tests mentioned in the paper.
  • source_code/illustrating_mesh_approximated_field.R A self-contained script for illustating how to approximate a given field on specific mesh using the finite element method (FEM).

Full result table for simulations experiments

A sort and searchable table with the full simulation results is available here.

Permutation tests

A sort and searchable table with results for permutation tests of pairwise differences between all combinations of the approximation methods, for each parameter combination is available here.

Any questions about the paper, simulations or scripts can be raised by opening a GitHub issue, or sending an email to the author jullum@nr.no.