GitHub release (latest SemVer including pre-releases) CRAN/METACRAN

Implements an n-dimensional parameter space partitioning algorithm for evaluating the global behaviour of formal computational models as described by Pitt, Kim, Navarro and Myung (2006)

If you want to quickly catchup with parameter space partitioning, click here.

To quickly install the latest version:


As of the version 0.5-dev, psp recieved significant improvements. As the graph below, the C++ implementation can be 100x faster than the simple R implementation.

Since the original R package came out, we have been struggling with one simple thing: performance. Parameter space partitioning can take a long time if the models are complex or have many parameters. The more parameter the model has, the larger the space we need to explore becomes.

While there are various ways to go about it, the first step was to re-implement our parameter space partitioning routine in C++. The goal was to significantly improve single-core performance, which must come before any parallelisation.

This was a big step and I am incredibly happy that it was a succesful one.