Model systems as stock-and-flow models in R. Stock-and-flow models are foundational to system dynamics, and help to understand systems complicated by nonlinearities, delays, and feedback loops. sdbuildR aims to make stock-and-flow modelling accessible and effortless, enabling you to dedicate your expertise to what matters most: building insightful, high-quality models. Get started at https://kcevers.github.io/sdbuildR/!
Features
- Accessibility: Get started with stock-and-flow models with limited knowledge.
- Flexibility: Modify your models easily and robustly.
- Insight Maker integration: Import models from Insight Maker.
- Julia backend: Use Julia for high-performance simulations, without any knowledge of Julia.
- Unit support: Use standard or custom units to verify and interpret your model.
All package capabilities are described in the vignettes:
-
vignette("build")Learn how to build, modify, and debug stock-and-flow models. -
vignette("units")Learn why and how to use (custom) units.
Installation
You can install the development version of sdbuildR like so:
# install.packages("remotes")
remotes::install_github("KCEvers/sdbuildR")sdbuildR offers two simulation engines: R and Julia. If you would like to use units and delay functions, you will need to set up the Julia environment. Run use_julia() to install Julia and required packages (initial setup may take 5–15 minutes):
sdbuildR::use_julia()Limitations
Non-negative stocks and flows are minimally supported, Setting stocks to non-negative will not adjust flows. Hard-coding non-negativity is not recommended, as it may mask model misspecification. Stocks and flows that logically cannot be negative (e.g., animals or deaths) should remain non-negative through appropriate model equations and parameters to avoid unexpected results. In addition, sdbuildR does not support vectorized operations.
Other System Dynamics software
sdbuildR is heavily based on common System Dynamics software such as Vensim, Powersim, Stella, and Insight Maker. To translate xmile models to R, see the R package readsdr. To build stock-and-flow models with the R package deSolve, the book System Dynamics Modelling with R by Jim Duggan will prove useful. In Python, stock-and-flow models are supported by PySD.
Troubleshooting
sdbuildR is under active development. While thoroughly tested, the package may have bugs, particularly in complex model translations. We encourage users to report issues on GitHub (https://github.com/KCEvers/sdbuildR/issues) - your input helps the package improve! Use debugger() to diagnose model errors, and use the vignettes for guidance.