WorldDynamics.jl: A Julia Package for Developing and Simulating Integrated Assessment Models

Pierluigi Crescenzi, Emanuele Natale, Aurora Rossi, Paulo Bruno S. SerafimJ. Open Source Softw 2024 · DOI · DBLP

Introduction

Integrated Assessment Models (IAMs) serve as critical tools for understanding the long-term dynamics between human activity and the Earth's biophysical systems. Since the 1970s, models like World3—commissioned by the Club of Rome—have provided foundational insights into global sustainability and the potential for societal collapse under resource constraints. However, the reliance on proprietary "black box" software and legacy programming paradigms has historically hindered transparency, modularity, and the application of modern high-performance computing (HPC) techniques.

WorldDynamics.jl addresses these challenges by providing an open-source Julia package designed to modernize the simulation of IAMs. By leveraging the Julia ecosystem, the package enables researchers to move beyond rigid frameworks toward a more flexible, high-performance environment for global modeling.

Methodology and System Architecture

The core innovation of WorldDynamics.jl lies in its transition from simple finite-difference equations to Differential-Algebraic Equations (DAEs). This shift allows for more stable and accurate simulations using state-of-the-art numerical solvers provided by DifferentialEquations.jl.

Furthermore, the package adopts a compositional modeling approach via ModelingToolkit.jl. This architecture treats different global sectors—such as Capital, Population, Pollution, and Agriculture—as modular components.

Causal loop diagram showing the modular sectors (Agriculture, Pollution, Capital, etc.) within the WorldDynamics.jl framework.

As illustrated in the system diagram above, these sectors can be developed, tested, and modified independently before being integrated into a unified system. This modularity allows researchers to swap out outdated components (e.g., a 1970s agriculture model) for contemporary versions without necessitating a complete rewrite of the simulation logic.

Results and Validation

A primary goal of WorldDynamics.jl is to ensure functional parity with historical models. The package includes parameters and interpolation tables to replicate the "reference runs" of Forrester’s World1 and World2, as well as the various iterations of World3 (1974, 1994, 2003).

Replication of the World3 model's standard run, demonstrating functional parity with historical simulations.

As shown in the replication of World3's standard run, the Julia implementation successfully overlays the original plots. This verification proves that the modern DAE-based approach is consistent with historical methods while offering the benefits of open-source transparency. Users can inspect every aspect of the simulation, from solver tolerances to the non-linear functions describing relationships like the impact of pollution on life expectancy.

Future Directions

While the current package focuses on the "Club of Rome" family of models, it provides a foundation for future advancements in global modeling. Planned developments include the integration of the Earth4All (2022) model and the use of data-driven techniques to "discover" new model structures from historical data. Additionally, the integration with JuMP.jl will facilitate policy optimization, helping researchers identify the most effective paths toward achieving global sustainability goals.

Summary updated

BibTeX
@article{crescenzi2024worlddynamics,
  author = {Pierluigi Crescenzi and Emanuele Natale and Aurora Rossi and Paulo Bruno S. Serafim and Aurora Rossi and Paulo Bruno S. Serafim and Pierluigi Crescenzi},
  title = {{WorldDynamics.jl: A Julia Package for Developing and Simulating Integrated Assessment Models}},
  journal = {J. Open Source Softw.},
  year = {2024},
  doi = {10.21105/joss.05772},
}
Figure