RE: [computer-go] Interesting endgame case
I checked this position again, and Many Faces finds j3 after a few thousand playouts, but even at a million playouts (28 seconds) it's only 61%. From: computer-go-boun...@computer-go.org [mailto:computer-go-boun...@computer-go.org] On Behalf Of Brian Sheppard Sent: Saturday, August 15, 2009 8:13 AM To: computer-go@computer-go.org Subject: [computer-go] Interesting endgame case assuming komi 7.5 and Chinese rule, playing at J3 white will win. After J3, white has 35. It only needs to win the ko or takes two dames. If black fills the dame, it loses the ko. If it fills the ko, white can take two dames. Yes, Chinese and 7.5. I basically figured that J3 was just another in a long series of stupid moves by Pebbles, but when it insisted that J3 was winning no matter how long it searched then I decided to look into it. J3 looks stupid because it fills territory that O already owns. J3 wins because it is reverse sente (I think this is the right terminology) because H4 is no longer sente for X. It also gains a move in the semeai against X's dead stones on the left, so O gains *two* ko threats. Just curious: how did you find J3? 1 2 3 4 5 6 7 8 9 A - - O - - - - - - B X X X O X - - - - C O O X O X X - - - D O - O X X - X X - E - O O O X X O X X F - X O - X O O O X G - X O - X O O - O H O X O - X X O O - J - - - O X - X O - O to play ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Interesting endgame case
On Sat, Aug 15, 2009 at 09:13:02AM -0600, Brian Sheppard wrote: assuming komi 7.5 and Chinese rule, playing at J3 white will win. After J3, white has 35. It only needs to win the ko or takes two dames. If black fills the dame, it loses the ko. If it fills the ko, white can take two dames. Yes, Chinese and 7.5. I basically figured that J3 was just another in a long series of stupid moves by Pebbles, but when it insisted that J3 was winning no matter how long it searched then I decided to look into it. J3 looks stupid because it fills territory that O already owns. J3 wins because it is reverse sente (I think this is the right terminology) because H4 is no longer sente for X. It also gains a move in the semeai against X's dead stones on the left, so O gains *two* ko threats. Just curious: how did you find J3? 1 2 3 4 5 6 7 8 9 A - - O - - - - - - B X X X O X - - - - C O O X O X X - - - D O - O X X - X X - E - O O O X X O X X F - X O - X O O O X G - X O - X O O - O H O X O - X X O O - J - - - O X - X O - O to play I don't see how J3 works, black still can win the ko, can't he? J3 - H4 - G4 - F4 - J6 now they play the ko: H9 - J9 - J5 black takes A4 - A5 - J6 white takes F1 - J2 - J5 black takes A2 - A1 - J6 white takes E1 - G1 - J5 black takes and white has no threats. I admit I didn't see F1 immediately and went at E1 right away, maybe that's the hard part for the bots? -- Petr Pasky Baudis A lot of people have my books on their bookshelves. That's the problem, they need to read them. -- Don Knuth ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Interesting endgame case
Petr Baudis wrote: 1 2 3 4 5 6 7 8 9 A - - O - - - - - - B X X X O X - - - - C O O X O X X - - - D O - O X X - X X - E - O O O X X O X X F - X O - X O O O X G - X O - X O O - O H O X O - X X O O - J - - - O X - X O - O to play I don't see how J3 works, black still can win the ko, can't he? J3 - H4 - G4 - F4 - J6 Here's your mistake, don't exchange G4-F4. To begin with white can save G4 for use as a ko threat but in this case that is also wrong. By keeping G4 and F4 open white has enough outer liberties so that black only gets a single ko threat in the lower left (E1). now they play the ko: H9 - J9 - J5 black takes A4 - A5 - J6 white takes F1 - J2 - J5 black takes A2 - A1 - J6 white takes E1 - G1 - J5 black takes and white has no threats. I admit I didn't see F1 immediately and went at E1 right away, maybe that's the hard part for the bots? With two outer liberties F1 can be ignored. /Gunnar ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] Interesting endgame case
Valkyria plays J3 or capture the ko, with capturing the ko as the most common choice. But it always thinks for some thousands of playouts before it sees a clear win. Magnus Quoting David Fotland fotl...@smart-games.com: I checked this position again, and Many Faces finds j3 after a few thousand playouts, but even at a million playouts (28 seconds) it's only 61%. From: computer-go-boun...@computer-go.org [mailto:computer-go-boun...@computer-go.org] On Behalf Of Brian Sheppard Sent: Saturday, August 15, 2009 8:13 AM To: computer-go@computer-go.org Subject: [computer-go] Interesting endgame case assuming komi 7.5 and Chinese rule, playing at J3 white will win. After J3, white has 35. It only needs to win the ko or takes two dames. If black fills the dame, it loses the ko. If it fills the ko, white can take two dames. Yes, Chinese and 7.5. I basically figured that J3 was just another in a long series of stupid moves by Pebbles, but when it insisted that J3 was winning no matter how long it searched then I decided to look into it. J3 looks stupid because it fills territory that O already owns. J3 wins because it is reverse sente (I think this is the right terminology) because H4 is no longer sente for X. It also gains a move in the semeai against X's dead stones on the left, so O gains *two* ko threats. Just curious: how did you find J3? 1 2 3 4 5 6 7 8 9 A - - O - - - - - - B X X X O X - - - - C O O X O X X - - - D O - O X X - X X - E - O O O X X O X X F - X O - X O O O X G - X O - X O O - O H O X O - X X O O - J - - - O X - X O - O to play -- Magnus Persson Berlin, Germany ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
[computer-go] Interesting endgame case
1 2 3 4 5 6 7 8 9 A - - O - - - - - - B X X X O X - - - - C O O X O X X - - - D O - O X X - X X - E - O O O X X O X X F - X O - X O O O X G - X O - X O O - O H O X O - X X O O - J - - - O X - X O - O to play I analyzed this position as a loss for O, as follows: O has 22 points at left, and will gain 1 point from F4/G4/H4 O has 13 points at right, with a Ko to potentially gain 2 O has two ways to win: the Ko on J6/J7, or gain two of F4/G4/H4 Since H4 is sente for X, O should start with H4. This looks winning for O because O will gain another point from F4/G4. Unfortunately, X can win both F4 and G4 by using his doomed stones on F2. Some variations: J2-J3-F1-E1-F4-G1-G4, and J2-F1-G1-E1-F4-J3-G4. If H4 does not win, then O must start his ko immediately with J6. X has ko threats with H9, H4, E1 (or F1). O has Ko threats with A4 and then A2 and H4 (which threatens to win by taking *two* of the dame from F4/G4/H4. Since X has more ko threats, it seems like O should not win. Extensive simulations supported this theory. But now I have a new version of Pebbles, and it says that O should win. What do you think? ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Interesting endgame case
In message 18d95a4b0a2748e39855f882ed6ce...@inspirone1705, Brian Sheppard sheppar...@aol.com writes 1 2 3 4 5 6 7 8 9 A - - O - - - - - - B X X X O X - - - - C O O X O X X - - - D O - O X X - X X - E - O O O X X O X X F - X O - X O O O X G - X O - X O O - O H O X O - X X O O - J - - - O X - X O - O to play I analyzed this position as a loss for O, as follows: What rules are you using? What is the komi? Nick O has 22 points at left, and will gain 1 point from F4/G4/H4 O has 13 points at right, with a Ko to potentially gain 2 O has two ways to win: the Ko on J6/J7, or gain two of F4/G4/H4 Since H4 is sente for X, O should start with H4. This looks winning for O because O will gain another point from F4/G4. Unfortunately, X can win both F4 and G4 by using his doomed stones on F2. Some variations: J2-J3-F1-E1-F4-G1-G4, and J2-F1-G1-E1-F4-J3-G4. If H4 does not win, then O must start his ko immediately with J6. X has ko threats with H9, H4, E1 (or F1). O has Ko threats with A4 and then A2 and H4 (which threatens to win by taking *two* of the dame from F4/G4/H4. Since X has more ko threats, it seems like O should not win. Extensive simulations supported this theory. But now I have a new version of Pebbles, and it says that O should win. What do you think? ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -- Nick Weddn...@maproom.co.uk ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
Re: [computer-go] Interesting endgame case
assuming komi 7.5 and Chinese rule, playing at J3 white will win. After J3, white has 35. It only needs to win the ko or takes two dames. If black fills the dame, it loses the ko. If it fills the ko, white can take two dames. On Fri, Aug 14, 2009 at 12:56 PM, Nick Wedd n...@maproom.co.uk wrote: In message 18d95a4b0a2748e39855f882ed6ce...@inspirone1705, Brian Sheppard sheppar...@aol.com writes 1 2 3 4 5 6 7 8 9 A - - O - - - - - - B X X X O X - - - - C O O X O X X - - - D O - O X X - X X - E - O O O X X O X X F - X O - X O O O X G - X O - X O O - O H O X O - X X O O - J - - - O X - X O - O to play I analyzed this position as a loss for O, as follows: What rules are you using? What is the komi? Nick O has 22 points at left, and will gain 1 point from F4/G4/H4 O has 13 points at right, with a Ko to potentially gain 2 O has two ways to win: the Ko on J6/J7, or gain two of F4/G4/H4 Since H4 is sente for X, O should start with H4. This looks winning for O because O will gain another point from F4/G4. Unfortunately, X can win both F4 and G4 by using his doomed stones on F2. Some variations: J2-J3-F1-E1-F4-G1-G4, and J2-F1-G1-E1-F4-J3-G4. If H4 does not win, then O must start his ko immediately with J6. X has ko threats with H9, H4, E1 (or F1). O has Ko threats with A4 and then A2 and H4 (which threatens to win by taking *two* of the dame from F4/G4/H4. Since X has more ko threats, it seems like O should not win. Extensive simulations supported this theory. But now I have a new version of Pebbles, and it says that O should win. What do you think? ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ -- Nick Weddn...@maproom.co.uk ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
RE: [computer-go] Interesting endgame case
Many Faces thinks X wins if X moves first (85%), but unclear if white moves first (51%) with no change after huge numbers of playouts. Good test position. I've seen from KGS games that Many Faces does not handle endgame kos very well. This is another example. David -Original Message- From: computer-go-boun...@computer-go.org [mailto:computer-go- boun...@computer-go.org] On Behalf Of Brian Sheppard Sent: Friday, August 14, 2009 9:46 AM To: computer-go@computer-go.org Subject: [computer-go] Interesting endgame case 1 2 3 4 5 6 7 8 9 A - - O - - - - - - B X X X O X - - - - C O O X O X X - - - D O - O X X - X X - E - O O O X X O X X F - X O - X O O O X G - X O - X O O - O H O X O - X X O O - J - - - O X - X O - O to play I analyzed this position as a loss for O, as follows: O has 22 points at left, and will gain 1 point from F4/G4/H4 O has 13 points at right, with a Ko to potentially gain 2 O has two ways to win: the Ko on J6/J7, or gain two of F4/G4/H4 Since H4 is sente for X, O should start with H4. This looks winning for O because O will gain another point from F4/G4. Unfortunately, X can win both F4 and G4 by using his doomed stones on F2. Some variations: J2-J3-F1-E1-F4-G1-G4, and J2-F1-G1-E1-F4-J3-G4. If H4 does not win, then O must start his ko immediately with J6. X has ko threats with H9, H4, E1 (or F1). O has Ko threats with A4 and then A2 and H4 (which threatens to win by taking *two* of the dame from F4/G4/H4. Since X has more ko threats, it seems like O should not win. Extensive simulations supported this theory. But now I have a new version of Pebbles, and it says that O should win. What do you think? ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/ ___ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/