After dwelling happily on our GitHub repositories for an extended period of time, rTRNG has now finally made it to CRAN.
We are very happy to get this out just ahead of the R in Finance in Chicago, where the functionality had been show-cased with an applied example two years ago. The main changes since then are related to dependency upgrades and other maintenance, so if you watched the package presentation at useR!2017 and have tried it out before, it will likely feel very familiar. In case you’ve never heard about it, no worries, we will be back with a second post very soon, which will discuss functionality and technical details. In the meantime, below is a short story about the integration of Tina’s Random Number Generator1 with R :).
What started as a series of extensions to a large-scale Credit Risk simulation application on a client project several years ago, evolved into more investigative and exploratory work. This involved the methodological modeling side (with topics such as importance sampling and stochastic recovery rates), as well as the software engineering, or more specifically high-performance computing side, with a focus on parallelization and efficient random number generation. In our search for libraries that could assist in this quest, we came across TRNG, which made the impression of a high-quality, yet overseeable project. We gave it a try and thanks to the smoothness of Rcpp and RcppParallel, things worked out in a breeze, and we had developed a nice custom solution bringing high value to our client.
Our experience was so positive, that we felt it was worth sharing this with the wider community, and that’s where we decided to make a new R package on GitHub. Now most long-time R users will claim that a package that isn’t on CRAN is only half a package, living in the shadows, never fully accredited. Being aware of this, we set the aim for CRAN, though side-tracked with lots of other exciting and demanding projects, things ended up taking a bit longer… but now we are happy to say: the wait is over!
We sincerely hope that some of you will fall in love with this package similarly as we did.
Who’s Tina? Tina is no acronym and is the name of a Linux cluster at the Institute of Theoretical Physics at the University Magdeburg in Germany, where TRNG was developed. ↩