"Thomas de Groot" <tDOTdegroot@interDOTnlANOTHERDOTnet> wrote:
> Just found this now. I am going to consult this in detail. Excellent work.
> Thank you very much indeed.

Well, I guess I need to post a revised version; some of the observations &
guesses turned out wrong, as I found out by trying to optimize the code
accordingly.

For example, to my great dismay the low error factor can *not* be used for the
same effect as nearest reuse count when alway sample is off.

I still don't really understand *why* that is though (or, to be precise, I do
understand why it can make *some* difference, but I have no clue why it gives
the results it does).

Post a reply to this message