NIMBLE

Nimble is a system for building and sharing analysis methods for statistical models mostly for computationally-intensive and hierarchical methods and hence you can turn the BUGS code into model projects which will be used for whatever algorithm you want. It is built in R but uses C++ to compile your models and algorithms.its components include; an initial library of algorithms for models written in bugs which can be customized from R or used directly before being compiled and run, language embedded in R for programming algorithms for models, and a system for using models written in bugs model language as programmable objects in R.

You can use and customize Nimble’s statistical algorithm since it provides MCMC, sequential Monte Carlo, and others. The Nimble algorithms are written in order to adapt to different statistical model.

For instance when using MCMC, Nimble assigns a default set of sampler choices but you can customize the sampler from R. with Nimble, you can compile your models and algorithms for fast execution since it generates C++ code which is customized to your model and algorithms. To use Nimble, you will not be required to have knowledge of C++. Nimble provides R function which calls the compiled algorithm and you get output back in R.

With Nimble, you can also write your own algorithms in nimble whereby writing new statistical methods using NimbleFunctions in R is the same as writing R function. The nimble compiler makes Nimble functions run efficiently. Nimblefunction also allows programmers to control how a particular algorithm should adapt to each model or variables it is applied to.

You will also compile numerical work in R via C++ without coding any C++ since this Nimblefunction doesn't need to use BUGS models so you use them to speed up many kinds of numerical computations for any other purpose.