News

samOptiPro: make your model runs faster

As DIASPARA rapidly comes to an end, it is time to present some of our results. Let’s start with the R package samOptiPro. This package aims to enhance the computational performance of Bayesian stock assessment models. Any expert knows how problematic the computation time required to run such models can be.

The idea behind samOptiPro is to help users identify computational bottlenecks and propose possible solutions to address them. It provides a full set of tools to identify problematic nodes, propose and test solutions, and run benchmarks. Since there is no one-size-fits-all solution, using such tools to explore a wide range of options is critical.

The efficiency of the package is showcased on different models used by ICES expert working groups and was applied in practice during the last Working Group on North Atlantic Salmon meeting. The package takes advantage of the flexibility offered by the NIMBLE platform, including the recently developed Hamiltonian Monte Carlo sampler.

If you want to test it for your own model, the package includes a detailed vignette describing the process and the main functions.