> My questions are: > > - How does this speed compare to your implementations of > Bouzy's 5/21 or your own implementations, I'm sure Many Faces' influence function is faster, even though it is not incremental, like Mark's. I've described my algorithm before so no details now. It does a single pass expansion from each stone. White and black influence are tracked separately and combined in a nonlinear manner to get the final score. Influence is reflected from the edge of the board (the same as your magnetic analogy). > > - Does my function look reasonable (I am not a Go player..) > The sole black stone on the right edge is not guaranteed > territory, but what is most important is that the overall > balance is correct. That black stone does not signify > "territory", it means that on that part of the board, there > will be at least 1 black stone or at least 1 black territory > point. For white, there is no such "guarantee" (with optimum > play of both). It doesn't look reasonable to me, since it gives territory to dead groups. I do life and death first, and the strength of the influence depends on the strength of the group. Without this, there is no way to get accurate territory from any kind of influence function, although it might be useful for other purposes, like estimating connection strength or moyo potential. The function used for influence is not as important as the code that draws conclusions from it, sicne no influence function can determine territory by itself. > > - What do you think about doing away with the difference > between "moyo" and "almost certain territory" and instead > trying to classify more of the "moyo" into "territory"? I am > less interested in "sphere of influence" than I am in "will > become B/W when vigorously defended". The least ambiguity, the better? You can't play good go on the full board if you ignore moyo. Your opponent can always trade small secure territory for big moyo. If you don't see the potential territory in the moyo, your score will be way off. > > - Does any of you use an influence function in a tree search > or do you limit it to the first ply? I use it in every full board evaluation. I've attached the same position with MFGO's influence (-50 is secure black territory and +50 is secure white territory). The dead black group in the lower right is white territory. The unsettled stones in the upper right and left center get intermediate values. The white bottom is solid territory to the right, and not solid to the left. The connecting points of the one point jumps in the center don't get full value since the opponent can put a live stone there. The upper side is not territoty yet since it is open at the side and there is a 3-3 invasion possible. I think if you use influence to get a territory score you need to get good values near dead stones, unsettled fights, and moyo. -David > > _______________________________________________ > computer-go mailing list > computer-go@xxxxxxxxxxxxxxxxx > http://www.computer-go.org/mailman/listinfo/computer-go/ >
Attachment:
influence.GIF
Description: GIF image
_______________________________________________ computer-go mailing list computer-go@xxxxxxxxxxxxxxxxx http://www.computer-go.org/mailman/listinfo/computer-go/
