villager: A framework for designing and executing agent-based models in R

Abstract

Villager is an agent-based modeling framework: it prescribes a convention and interface for modelers to create and run agent-based models (ABM). The framework is aimed at researchers in the social sciences who are focused on modeling human populations. The key features of villager are:1. Scalability: villager makes extensive use of the R6 class system (Chang, 2020), enabling the power of reference semantics without the hurdles of manual memory management. This enabled an architecture design where user-supplied functions are run within the framework. The reference semantics also enable cheaper memory operations by allowing for the mutation of agents in-place rather than costly copy semantics. 2. Extensibility: villager exposes a number of classes that can be extended by domain scientists to provide flexibility in experiment design. The extended classes can be "plugged" into the villager framework and run seamlessly.Together, these two features allow researchers to design ABMs with flexible requirements-both functionally and computationally.

epiworldR: Fast Agent-Based Epi Models

“…Several alternatives to epiworldR exist and provide researchers with a range of options, each with its own unique features and strengths, enabling the exploration and analysis of infectious disease dynamics through agent-based modeling. Below is a manually curated table of existing alternatives including ABM (Ma, 2023), abmR (Gochanour et al, 2023), cystiSim (Devleesschauwer & Braae, 2016), villager (Thelen et al, 2022)…”

Hello World! Building computational models to represent social and organizational theory

“…Similarly,Wilensky and Rand's (2015) textbook offers an accessible and comprehensive reference for using the NetLogo modeling software to develop agent-based models.7 We recommend using the RStudio IDE when programming models in R. RStudio can be freely downloaded from https://www.rstudio.com/products/rstudio/download and runs on all major operating systems.8 The present paper focuses on building models using only the functionality provided in base R. However, packages are emerging within the R ecosystem that attempt to provide more direct capabilities for developing computational models and running simulations. For example, the simecol(Petzoldt & Rinke, 2007) and villager(Thelen et al, 2022) packages offer architectures and helper functions for writing and simulating system dynamics and agent-based models, respectively, within R. Alternatively, the RNetLogo package(Thiele, 2014) provides an interface for operating, executing, and communicating with models programmed in NetLogo through R that allows one to use the programming language/philosophy of the former and the analytic tools/capabilities of the latter from the same software application.9 The hashtag (#) symbols surrounding the headings for each of the five sections shown in Figure1are used in the R programming language to indicate code that should be ignored by the computer. More precisely,…”