This repository contains the code for the analyses of
Novak & Stouffer (2021) Geometric complexity and the information-theoretic comparison of functional-response models. Frontiers in Ecology and Evolution (bioRxiv)
Please email Mark Novak (mark.novak@oregonstate.edu) and Daniel Stouffer (daniel.stouffer@canterbury.ac.nz) with any questions.
All analyses were performed in Mathematica v. 12.1.1.0 using its notebook (.nb) format.
The notebooks for recreating the main analyses and figures of the paper are:
- GeomComp_Compute.nb computes the geometric complexities of all considered models for all considered experimental designs of the specified design-type.
- GeomComp_Plotting.nb produces the visualizations of the paper's main results.
The two notebooks are used repeatedly for each experimental design type (i.e. for analyzing logarithmic (= GoldenRatio) or Arithmetic prey and predator abundance spacings) as well as for the analyses using alternative minimum or maximum number-of-prey-eaten constraints. Which design type or prey-eaten constraint is to be analyzed is specified manually at the top of the GeomComp_Compute.nb notebook (for design type) or in the argument of the GeomComplex[] function (for the prey-eaten constraint) defined in that notebook. Which results are to be plotted by GeomComp_Plotting.nb is specified by changing the directory locations of the results and export folders.
Additional notebooks:
- ParamComp_2ndTerm.nb produces Fig. 1 visualizing FIA's parametric complexity term as a function of sample size.
- GeomComp_H2MM.nb produces Fig. 2 comparing Holling and Michaelis-Menten formulations.
- Identifiability.nb produces Fig. S.1 visualizing the identifiability of all considered models as a function of experimental design.
- GeomComp_H2H3_binom.nb computes the geometric complexities of Rogers' Type II and Type III models assuming a binomial likelihood and non-replacement experimental designs.
- GeomComp_Plotting_binom.nb produces the figures corresponding to the ananlysis of Rogers' Type II and Type III models.
Further notebooks in the other directory were used for visualizing or exploring specific functional-response models or were useful in the development of the primary analyses.
The docs directory contains pdf printouts of the main Mathematica notebook files.
The figs directory contains all figures of the main text and supplementary materials, organized by design type.
The results directory contains all files in .txt format exported by GeomComp_Compute.nb, organized by design type. The results include the geometric complexity estimates for each model. The directory also includes Design__.txt files that summarize all considered experimental designs (i.e. prey and predator abundance combinations) as depicted in the paper's supplementary materials.
All code is provided "as is" and without warranty. If you know of more efficient or elegant ways of doing anything (of which there are most definitely many), we’d love to learn from you.