indirectfacilitation

The goal of indirectfacilitation is to …

Installation

You can install indirectfacilitation from github with:

# install.packages("devtools")
devtools::install_github("alaindanet/indirect_facilitation_model")

Example

This is a basic example which shows you how to solve a common problem:

• Load the packages:

devtools::load_all()

library(simecol)
library(ggplot2)
library(tibble)
library(magrittr)

• Load the model and specify the parameters:

mod <- two_facilitation_model()
mod

# Show parameters:
parms(mod)

# Tweak parameters:
parms(mod)["g"] <- 0.2
parms(mod)["gamma1"] <- 0.10
parms(mod)["u"] <- 5 
times(mod) <- c(from = 0, to = 1000, by = 1)

• Launch a simulation and plot the result:

mod_run <- sim(mod)
plotnp(mod_run)

Todo

Paper

• Fig 1: methods
  • Add temporal dynamic CA + PA
• Fig 2: multi-states
  • Add arrows
• Fig 3: bifurcation details
• Fig 4: Clustering cellular automata, C++ et Cnp (f(\gamma, u)) and (f(\gamma, u))
  • (Change space between panel)[tidyverse/ggplot2#2464 (comment)]
  • (Change margin of the plot for plot_grid)[wilkelab/cowplot#31 (comment)]
  • Fix B and E legend decrease a bit the lowest value (some points disapeared in fig B right panel, en haut à gauche)

Run

1. Run pair-approximation model with automated end of run
2. Replace the file with use_data()

Implementation

• Clustering:
  • plot NP
  • plot (x2 - x1) / x1 for comparison CA/PA
  • run CA for gradient facilitation/dispersion
• plot: separate lines (create groups: e.g: together & < threshold, together & >= threshold, low_together …)
  • Create groups
  • Plot groups (linetype problem)
• Check if simulations have reached stability
  • Implementation of a custom solver (“lsodar”)
  • Test the implementation
• sim_multi():
  • Save a simecol object
  • Save a time argument
  • Save init values
  • Save baseline parameters
  • Save param_combination
  • param_combination: matrix of parameter combination
  • be able to sim from a sim_multi object
• Generalize run_scenarii_gradient and run_2d_gradient: f(b,g)
  • check_param(model_spec, names(gradient)))
  • Add check for argument conformity (check_param, check_inits) of valid starting values (scenario = “custom”), or a named list (provided by init_scenarii())
  • Add methods: init.scenarii(), param.scenarii() (S4 methods) https://stackoverflow.com/questions/12100856/combining-s4-and-s3-methods-in-a-single-function
  • See here
• Compare the effect of dispersal, facilitation strength, paturâge, aridité on clustering
• Document functions