Robert Jasiek asked: > In this paper, does "semeai" stand for "one of 'capturing race', > 'unsettled life and death situation', 'basic ko fight', 'multiple ko > fight', 'ladder breaking' 'etc.'"?
It stands for "localizing multiple peaks" in a histogram of many playouts. Have a look at http://www.althofer.de/crazy-shadows.html In the s econd block, at move 138, you see two peaks. They come from a local fight. In the right peak are those games collected where Black won the fight, in the left peak White was that winner. > Is the identification of a cluster proven only with stochastic > probability, Of course, it is a Monte Carlo "proof" only: it is based on the outcomes of the random playouts. > How do the methods affect simulations runtime? Not this much when I interpret the algorithm(s) correctly. (The Paderborn guys have no problems at all with this. They have massive parallel machines, but problems with the speedup when more than 32 processors are in team work...) ***************************** Let me use the opportunity to propose another wish for a feature: At the end of a playout, typically the neutral (Seki) points are recognized correctly. For me as a user it would be nice when a commercial bot (CrazyStone or Zen or ManyFaces or MyGoFriend or ...) would be able to show "Seki clouds", i.e., marking those points which are frequently Seki points. Ingo. Ps. Sorry for writing this as a new message. But for months already I do not receive the mails from the list, and an attempt to register from another mailing address failed. _______________________________________________ Computer-go mailing list Computer-go@dvandva.org http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
