Personality 2D

This is a small Rcpp model tying together a number of interesting tools to get at a persistent question of the co-variation of two orthogonal but related 'animal personality' traits.

Running the model

Clone the repository using SSH by running git clone git@github.com:pratikunterwegs/noisewalker.git
In R, build the package using devtools::build()
In R, install the package using devtools::install()
Try out the model using the script scripts/check_noisewalker.R

Landscape

The spatially explicit resource landscape is modelled using noise, provided by FastNoiseLite.

Competition

Competition among animals is modelled as a cost of proximity to other animals, and inter-individual distances are calculated using boost Rtrees.

Fast-ish/er Bernoulli distributions

Animal movement and behaviour is probabilistic, with hudreds of thousands of calls to std::bernoulli_distribution, not to mention setting up this distribution. This is (hopefully) speeded up by using the GNU Scientific Library random number distributions.

Simulations launched from, and data returned to, `R`

The simulations are disguised as easy-to-use R functions, using Rcpp to obscure the scary C++ code that lies beneath. The simulation data are also returned to R, as well known objects (lists and data.frames).

Using Rcpp extensions for Windows compatibility

GSL and Boost are ridiculously difficult to use on Windows (coming from Linux) --- but Rcpp extension packages BH (Boost Headers) and RcppGSL allow for this to happen seamlessly.

GSL on Windows

However, some manual tinkering is required to get GSL libraries on Windows: Download local323.zip from http://www.stats.ox.ac.uk/pub/Rtools/libs.html and place contents of local323/lib/x64 in C:\local\lib.

GSL works out of the box on Linux systems with libgsl installed.

In R, make sure the GSL path is known by running:

Sys.setenv("LIB_GSL" = "C:/local323")

pratikunterwegs/noisewalker