The optimizations here are uncannily similar to the ones I performed for an algorithm we developed for application in bioinformatics. It began with a statistics

The optimizations here are uncannily similar to the ones I performed for an algorithm we developed for application in bioinformatics. It began with a statistics post-doc who knew some R. It was very brute-force. https://arxiv.org/pdf/1509.00549.pdf https://arxiv.org/pdf/1509.00549.pdf (Results on page 28) I suspect we would have seen similar speedups with larger vectors. If you look at the tweaks to the algorithm, they're relatively minor, but with very significant gains. Sorry about the code syntax; the audience for the paper were statisticians.