Algo Movelist

Description

The idea is to take a candidate for a hit (as provided by AlgoFinalpos or AlgoHash), i.e. a position on the board where a certain pattern could possibly match, and then to play through the game in order to decide whether this is really a hit. Of course, in practice, we take a list of all candidates for a given game, so that we have to play through the game only once. Furthermore, instead of using the SGF file, we use a "move list" where all the moves and captures are stored - by avoiding to compute whether any stones are captured with some move (and which ones) we save a lot of time.

Further Improvements

I do not see much room for improvement of this particular algorithm.