apsimR

apsimR is a typed R interface to APSIM Next Generation, the process-based agricultural systems simulator. It drives the simulator for prediction over scenario designs, parameter calibration, global sensitivity analysis, Gaussian-process emulation and validation against observations, and it can manufacture known-causal-structure test datasets for grading causal-inference methods. Every verb returns a versioned, provenance-complete apsim_manifest object, or a typed apsim_abstention when the simulator or its runtime is absent -- apsimR installs, loads and its tests run with no simulator present.

Related packages

• apsimx inspects, reads, edits and runs APSIM Next Generation and APSIM Classic simulation files.
• sensitivity screens factors and runs global sensitivity analysis on the outputs of any model.
• BayesianTools provides MCMC and SMC samplers and diagnostics, with a focus on calibrating complex system models.

apsimx is the closest: it gives detailed control over individual APSIM files. apsimR works one level up, running a simulation over a scenario design and covering calibration, sensitivity analysis, emulation and validation in one interface. Its results carry the inputs and simulator version that produced them, and it returns a typed abstention instead of an error when APSIM is not installed.

Install

# install.packages("remotes")
remotes::install_github("max578/apsimR")

Simulator-backed verbs additionally need a working install of APSIM Next Generation (>= 2024.1) and the .NET 8.0 runtime; apsim_available() reports whether apsimR can find one.

Minimal example

library(apsimR)

# Works with no simulator installed: every simulator-dependent verb
# abstains with a typed reason instead of erroring.
apsim_available()

# With APSIM installed, run the bundled Wheat example and read the yield
# report back as a provenance-complete manifest.
if (apsim_available()) {
  m <- apsim_predict(apsim_example("Wheat"))
  m
}

See vignette("apsimR") for prediction, calibration, sensitivity analysis, emulation, validation and the causal ground-truth generator end to end.

Contributing

Bug reports and suggestions are welcome as GitHub issues.

Citation

citation("apsimR")

Licence

MIT + file LICENSE. See LICENSE / LICENSE.md.