First it is really good c++ code. Really good to read.
I work on something similar what I can release in some months. It has a focus on time and space-efficiency (your code is in some places slower, this is why I couldn't use it for my project). You allocate the array with zeros. But if you would define array slates in size log n you don't have initialize everything instantly (what saves you lot of time for big arrays) (there was a paper describing this technique but I don't remember which).

also some documentation would be nice.