watanabe-j / eigvaldisprcpp

Rcpp extension of eigvaldisp

eigvaldispRcpp

Rcpp extension of eigvaldisp

This package extends the R package eigvaldisp by fast C++ functions using Rcpp-related packages. This is to speed-up evaluation of the approximate variance of the relative eigenvalue variance of correlation matrices with eigvaldisp::Var.VRR(), which takes prohibitively long time in R, even with vectorization and parallelization (see eigvaldisp:::AVar.VRR_pfd()).

This extension is provided as a separate package to minimize the dependency of eigvaldisp. Practically, one would not need this extension unless interested in applying eigvaldisp::Var.VRR() to large correlation matrices (p > 100 or so).

Installation

# install.packages("devtools")
devtools::install_github("watanabe-j/eigvaldispRcpp")

This package has the following dependencies:

Depends: eigvaldisp
Imports: Rcpp, RcppParallel
LinkingTo: Rcpp, RcppArmadillo, RcppEigen, RcppParallel
Suggests: testthat (>= 3.0.0)
SystemRequirements: GNU make

In case you are unfamiliar with Rcpp, it requires a working C++ compiler. Most Linux and macOS users would already have one. Windows users will require Rtools, which comes with GNU make. See http://www.rcpp.org/ or https://cran.r-project.org/package=Rcpp for details.

See also

Watanabe, J. (2021). eigvaldisp: R package for statistics of eigenvalue dispersion. https://github.com/watanabe-j/eigvaldisp.

Watanabe, J. (2022). Statistics of eigenvalue dispersion indices: quantifying the magnitude of phenotypic integration. Evolution, 76, 4–28. doi:10.1111/evo.14382.