Community Detection for Evolving Multiplex Networks

Multiplex temporal community detection with customizable interlayer coupling. Runs Louvain or Leiden community detection on each network layer and constructs interlayer ties using Jaccard similarity, overlap coefficient, node-strength weighted variants, or direct node identity links, and also provides a two-stage snapshot-and-match tracker that aligns independently detected per-layer communities across time with the Hungarian assignment algorithm. Supports user-specified layer connectivity via the layer_links argument, enabling adjacent-only temporal coupling that avoids the long-range pooling problem in standard multislice approaches.

dynamic_multiplex

dynamic_multiplex is an R package for multiplex community modeling with customizable interlayer ties.

Why this package

Standard multislice approaches (Mucha et al. 2010) connect all layers to all layers, meaning community structure at distant time periods influences assignments everywhere. When applied to temporal data, this creates a pooling problem: the community configuration at 2010 affects the structure detected at 1950. dynamic_multiplex provides explicit control over which layers influence which layers via a layer_links argument (from, to, weight), defaulting to adjacent-only temporal coupling.

Functions

Note: Louvain in igraph is undirected. When directed = TRUE and algorithm = "louvain", layers are collapsed to undirected weighted graphs before clustering.

• fit_multilayer_jaccard()
  • Fits Louvain or Leiden communities on each layer (supports directed layers).
  • Builds interlayer ties between communities using Jaccard similarity across selected layer pairs.
• fit_multilayer_overlap()
  • Fits Louvain or Leiden communities on each layer (supports directed layers).
  • Builds interlayer ties between communities using overlap coefficient across selected layer pairs.
• fit_multilayer_weighted_jaccard()
  • Fits Louvain or Leiden communities on each layer (supports directed layers).
  • Builds interlayer ties using node-strength weighted Jaccard similarity across selected layer pairs.
• fit_multilayer_weighted_overlap()
  • Fits Louvain or Leiden communities on each layer (supports directed layers).
  • Builds interlayer ties using node-strength weighted overlap coefficient across selected layer pairs.
• fit_multilayer_identity_ties()
  • Fits Louvain or Leiden communities on each layer (supports directed layers).
  • Builds interlayer ties only for the same node across selected adjacent layers.
• simulate_and_fit_multilayer()
  • Simulates multiplex layers from a planted partition process and runs one of the three fitting strategies.

Installation

Development install:

# install.packages("remotes")
remotes::install_local(".")

GitHub install (recommended for collaborators):

# install.packages("remotes")
remotes::install_github("jfedgerton/dynamic_multiplex", subdir = "r_code")

Quick example

library(dynamicmultiplex)

sim <- simulate_and_fit_multilayer(
  directed = TRUE,
  n_nodes = 50,
  n_layers = 4,
  n_communities = 3,
  fit_type = "jaccard",
  algorithm = "louvain",
  seed = 123
)

head(sim$fit$interlayer_ties)

# Custom layer influence map
custom_links <- data.frame(
  from = c(1, 2),
  to = c(2, 4),
  weight = c(1, 0.6)
)

fit_overlap <- fit_multilayer_overlap(
  sim$layers,
  algorithm = "leiden",
  layer_links = custom_links,
  min_similarity = 0.1,
  add_self_loops = TRUE,
  self_loop_multiplier = 1
)

Plotting options

# 1) Static series of network panels
plot_multilayer_series(sim$layers, fit = sim$fit, directed = TRUE, palette = "Dark2")

# 2) GIF animation with colorblind-friendly community colors (`Set2`/`Dark2` from RColorBrewer)
animate_multilayer_gif(
  sim$layers,
  fit = sim$fit,
  output_file = "multilayer_communities.gif",
  directed = TRUE,
  fps = 2,
  palette = "Set2"
)

# 3) Alluvial plot for community flow across temporal layers
plot_multilayer_alluvial(sim$fit, max_nodes = 100, palette = "Dark2")