Life In 19x19

Life in 19x19. Go, Weiqi, Baduk... Thats the life.
https://www.lifein19x19.com/

Basic endgame problem -- feedback requested

https://www.lifein19x19.com/viewtopic.php?t=3056

Page 4 of 5

Re: Basic endgame problem -- feedback requested

Posted: Tue Feb 01, 2011 10:52 am
by mw42
Dusk Eagle wrote:I imagine that with experience, the value of certain moves would just be memorized naturally over time. If you constantly refer back to something (like, to take a mathematical example, the value of sin (0)), it eventually just gets ingrained in your memory.

There was something confusing me about calculating the value of one of the positions on the board, but as I typed it out to ask I figured it out. I'll just leave it here in case I forget or if anyone's curious to see my thought process.
Click Here To Show Diagram Code
[go]$$B
$$ -------------
$$ . O X . . X .
$$ . O O X X X .
$$ . . . . . . .[/go]

If white plays first, he has one point. If black plays first, he has one point. So, if I'm not mistaken, the count here is 0 and a play gains 1 point.

Click Here To Show Diagram Code
[go]$$B
$$ -------------
$$ . O . O O X .
$$ . O O X X X .
$$ . . . . . . .[/go]

In this case, if white plays first, he has no points, and if black plays first, he has 2 (from captures) + 0 (from the resulting position) = 2 points. So black has one point in this position, and a play here seems like it gains one point.

So, if black were to capture, and then white recaptures, it seems like they both gain one point, which should leave the count unchanged. What I was missing originally is that the count in the second diagram is +1. I overlooked this and was thinking of it as zero, which made me question how miai counting could say that the score after black captures and white recaptures is zero when clearly black gained a point.


Click Here To Show Diagram Code
[go]$$B
$$ -------------
$$ . O X . . X .
$$ . O O X X X .
$$ . . . . . . .[/go]


If white plays he gains a point, if black plays he gains a point. From the perspective of white, we consider a black gain as a "negative" gain for white. So, from white's perspective it is a 1p-(-1p)=2p play.

Re: Basic endgame problem -- feedback requested

Posted: Tue Feb 01, 2011 11:12 am
by Magicwand
daniel_the_smith wrote:I'm still curious, for those who can do these problems: Do you all have a table of values memorized, do you calculate them out each time, or do you just read everything out and compare results?


basic endgame values given in the example is in the head and it only takes few sec to verify if not sure.
i think i am speaking for all dan level players. correct me if i am wrong.

Re: Basic endgame problem -- feedback requested

Posted: Tue Feb 01, 2011 11:21 am
by Magicwand
Click Here To Show Diagram Code
[go]$$B
$$ -------------
$$ . O X a . X .
$$ . O O X X X .
$$ . . . . . . .[/go]


Click Here To Show Diagram Code
[go]$$B
$$ -------------
$$ . 2 1 3 . X .
$$ O O O X X X .
$$ . . . . . . .[/go]


two above endgame are equal in value 2 point gote play. (or bill's 1 point gain for players)

below is also 2 point play. (bill will count as 1 point gain).
Click Here To Show Diagram Code
[go]$$B
$$ -------------
$$ . O . O O X .
$$ . O O X X X .
$$ . . . . . . .[/go]

Re: Basic endgame problem -- feedback requested

Posted: Tue Feb 01, 2011 11:59 am
by Violence
All I can do is be able to do the most basic of calculations. Recognize a 1 pt gote, a 2 pt gote, a 3 pt gote, and so on and so forth.

The rest is reading.

Re: Basic endgame problem -- feedback requested

Posted: Tue Feb 01, 2011 12:56 pm
by John Fairbairn
I'm still curious, for those who can do these problems: Do you all have a table of values memorized, do you calculate them out each time, or do you just read everything out and compare results?


Someone told me that Rob van Zeijst (insei level) had memorised the counts of about 1,000 endgame positions. I have no idea if either the story or figure is true, but the fact that the story is out there suggests there's at least a grain of truth in it.

At the other extreme, I do know one strong pro who claimed not to know how to count a yose ko.

Re: Basic endgame problem -- feedback requested

Posted: Tue Feb 01, 2011 2:19 pm
by Dusk Eagle
mw42 wrote:
Dusk Eagle wrote:I imagine that with experience, the value of certain moves would just be memorized naturally over time. If you constantly refer back to something (like, to take a mathematical example, the value of sin (0)), it eventually just gets ingrained in your memory.

There was something confusing me about calculating the value of one of the positions on the board, but as I typed it out to ask I figured it out. I'll just leave it here in case I forget or if anyone's curious to see my thought process.
Click Here To Show Diagram Code
[go]$$B
$$ -------------
$$ . O X . . X .
$$ . O O X X X .
$$ . . . . . . .[/go]

If white plays first, he has one point. If black plays first, he has one point. So, if I'm not mistaken, the count here is 0 and a play gains 1 point.

Click Here To Show Diagram Code
[go]$$B
$$ -------------
$$ . O . O O X .
$$ . O O X X X .
$$ . . . . . . .[/go]

In this case, if white plays first, he has no points, and if black plays first, he has 2 (from captures) + 0 (from the resulting position) = 2 points. So black has one point in this position, and a play here seems like it gains one point.

So, if black were to capture, and then white recaptures, it seems like they both gain one point, which should leave the count unchanged. What I was missing originally is that the count in the second diagram is +1. I overlooked this and was thinking of it as zero, which made me question how miai counting could say that the score after black captures and white recaptures is zero when clearly black gained a point.


Click Here To Show Diagram Code
[go]$$B
$$ -------------
$$ . O X . . X .
$$ . O O X X X .
$$ . . . . . . .[/go]


If white plays he gains a point, if black plays he gains a point. From the perspective of white, we consider a black gain as a "negative" gain for white. So, from white's perspective it is a 1p-(-1p)=2p play.


Yes, but it is a gote play for both white and black, and thus the [sl=localtally]local tally[/sl] is 2. So a play in the area is worth:
Count / Tally
= 2 points / 2
= 1 point.

This is using the [sl=MiaiCounting]Miai Counting[/sl] method, which I believe I am using correctly (let me know if I'm doing something wrong).

Re: Basic endgame problem -- feedback requested

Posted: Tue Feb 01, 2011 2:37 pm
by Magicwand
some comment about below shape.
if white play this two point gote play as below then black will have 1/2 point ko threat at the end.

Click Here To Show Diagram Code
[go]$$W
$$ -------------
$$ . O X 1 . X .
$$ . O O X X X .
$$ . . . . . . .[/go]


for that reason below 2 point gote is better than above because of 1 ko threat.
Click Here To Show Diagram Code
[go]$$W
$$ -------------
$$ . O O 1 O X .
$$ . O O X X X .
$$ . . . . . . .[/go]


although they equal in value they are not same and can make difference in the outcome.
i know for fact that professionals count such difference and make decision accordingly.

Re: Basic endgame problem -- feedback requested

Posted: Tue Feb 01, 2011 2:47 pm
by Bill Spight
John Fairbairn wrote:At the other extreme, I do know one strong pro who claimed not to know how to count a yose ko.


Only a few people in the world know how to count a yose ko. :)

Re: Basic endgame problem -- feedback requested

Posted: Tue Feb 01, 2011 3:16 pm
by Perception
My endgame is terrible so I'm not going to attempt this right now. I did however get The Endgame from the Elementary Go Series a couple days ago so when I've read that I'll give it a shot (probably in a week or two).

Also if anyone knows of a Japanese book about the endgame, I remember seeing it mentioned on godiscussions, that had full board 11x11 problems I'd like to get a copy of it. If you know the title or ISBN or even better where I could find a copy (I think it's out of print) that would be great. I'm going to go try to find the original post about it in the godiscussions archive.

edit: I found the post, the book is a Yose Dictionary (新編ヨセ辞典) by Kano Yoshinori (加納 嘉徳). I don't know where to find it since it's out of print.

Re: Basic endgame problem -- feedback requested

Posted: Tue Feb 01, 2011 3:52 pm
by mw42
Dusk Eagle wrote:Yes, but it is a gote play for both white and black, and thus the [sl=localtally]local tally[/sl] is 2. So a play in the area is worth:
Count / Tally
= 2 points / 2
= 1 point.

This is using the [sl=MiaiCounting]Miai Counting[/sl] method, which I believe I am using correctly (let me know if I'm doing something wrong).


Yes, you are using it correctly, but I'd prefer to say 2p gote play then a 1p play as you did. I thought you were counting wrong, sorry. :-)

Re: Basic endgame problem -- feedback requested

Posted: Tue Feb 01, 2011 7:27 pm
by Bill Spight
Click Here To Show Diagram Code
[go]$$W White to play. Japanese rules. 6.5 komi.
$$ ---------------------------
$$ | . . X O . . . O O . . . . |
$$ | X . X O O O O . O . X X . |
$$ | . X O . . . . O O O X O . |
$$ | . X X X X . X . X O . X . |
$$ | O O O O X . X . X X . X . |
$$ | . . O X . X . X O X . X . |
$$ | . . O X . X . X O X X X . |
$$ | . O O . O X . X O . O X . |
$$ | . X O O X X X O O X . X . |
$$ | . O . O O X X . O X X X O |
$$ | . . O . . . X O O O O X X |
$$ | . . O O X X X O . O O X . |
$$ | . . O . O O X O O . O . . |
$$ ---------------------------[/go]


First thing we do, we kill all the miai. ;) (We can do that because there are no kos. The same player might get both miai when the opponent takes and wins the ko.)

Click Here To Show Diagram Code
[go]$$W White to play. Japanese rules. 6.5 komi.
$$ ---------------------------
$$ | . . X O . . . O O 5 3 4 . |
$$ | X . X O O O O . O . X X . |
$$ | . X O 1 . a . O O O X O . |
$$ | 6 X X X X . X . X O b X . |
$$ | O O O O X . X . X X . X . |
$$ | . . O X . X . X O X . X . |
$$ | . . O X . X . X O X X X . |
$$ | . O O . O X . X O . O X . |
$$ | . X O O X X X O O X . X . |
$$ | . O . O O X X . O X X X O |
$$ | . . O . . . X O O O O X X |
$$ | . . O O X X X O . O O X . |
$$ | . . O . O O X O O . O 2 . |
$$ ---------------------------[/go]


:w1: and :b2: are miai. The result is 2 points (we take Black's point of view), whichever player gets which. :w3: - :w5: and :b6: are also miai. The result is 1 point, whichever player gets which.

So we can assume that one player gets one while the other player gets the other, and go ahead and make the exchange.

"a" and "b" are also miai, but to know that you have to know that they are not sente. I know that, but I will assume that the solver does not know that yet. :)

Click Here To Show Diagram Code
[go]$$W White to play. Japanese rules. 6.5 komi. {Miai removed.}
$$ ---------------------------
$$ | . . X O . . . O O O O X . |
$$ | X . X O O O O . O . X X . |
$$ | . X O O . . . O O O X O . |
$$ | X X X X X . X . X O . X . |
$$ | O O O O X . X . X X . X . |
$$ | . . O X . X . X O X . X . |
$$ | . . O X 1 X . X O X X X . |
$$ | . O O a O X . X O 2 O X . |
$$ | . X O O X X X O O X . X . |
$$ | . O . O O X X . O X X X O |
$$ | . . O . . . X O O O O X X |
$$ | . . O O X X X O . O O X . |
$$ | . . O b O O X O O . O X . |
$$ ---------------------------[/go]


As has been pointed out, our eyeballs tell us that :w1: and :b2: are the largest and next largest plays on the board. (If we want to count, :w1: gains 2 points and :b2: gains 1.5 points.) Now, Wa also gains 2 points and would also lead to a White win, but, with no kos it is never better than :w1:. Again, our eyeballs tell us that, because it leaves the possibility of :b1:. (In a ko position, Wa might be right, to leave :w1: as a ko threat.)

Our solver might briefly consider "b" as an alternative to :b2:, because it also has a potential swing of 3 points. However, that swing takes 3 moves instead of 2 moves for :b2:, so it must be a smaller play.

Note to numsgil: It is possible to prove that :w1: is correct, and then :b2:, by direct comparison of plays. :)

Click Here To Show Diagram Code
[go]$$W White 3?
$$ ---------------------------
$$ | . . X O . . . O O O O X . |
$$ | X . X O O O O . O . X X . |
$$ | . X O O . c . O O O X O . |
$$ | X X X X X d X . X O . X . |
$$ | O O O O X . X . X X . X . |
$$ | . . O X . X . X O X . X . |
$$ | . . O X 1 X . X O X X X . |
$$ | . O O . O X . X O 2 O X . |
$$ | . X O O X X X O O X . X . |
$$ | . O . O O X X . O X X X O |
$$ | . . O . a . X O O O O X X |
$$ | . . O O X X X O . O O X . |
$$ | . . O b O O X O O . O X . |
$$ ---------------------------[/go]


The comparison of "a" and "b" is not so obvious. Their game trees look like this:

Code: Select all


           A
          / \
         1  -1

           B
          / \
         B1  0
        / \
       3   1



"a" is a gote, and each play gains 1 point. "b" looks kind of like a 1 point sente, but each play gains only 1 point. (It is ambiguous.)

Unless the solver has read the relevant material, such as that on SL, this is not particularly helpful.

What about "c"? Could it be a 1 point sente? No, because after Wc, "d" is worth only 0.5 point.

It's a guess, so let's try the one that sort of looks bigger, "b".

Click Here To Show Diagram Code
[go]$$W Variation 1
$$ ---------------------------
$$ | . . X O . . . O O O O X . |
$$ | X . X O O O O . O . X X . |
$$ | . X O O . 5 . O O O X O . |
$$ | X X X X X 7 X 8 X O 6 X . |
$$ | O O O O X . X . X X . X . |
$$ | . . O X . X . X O X . X . |
$$ | . . O X 1 X . X O X X X . |
$$ | . O O . O X . X O 2 O X . |
$$ | . X O O X X X O O X . X . |
$$ | . O . O O X X . O X X X O |
$$ | . . O . 4 C X O O O O X X |
$$ | . . O O X X X O . O O X . |
$$ | . . O 3 O O X O O . O X . |
$$ ---------------------------[/go]


When we play the rest of the variation, we discover something interesting. :w5: and :b6: are miai, as are :w7: and :b8:. We can assume that they are played the same way in the other variation. :) (BTW, we can tell that :w5: is bigger than :b8: by eyeballing. :))

Click Here To Show Diagram Code
[go]$$W Variation 2
$$ ---------------------------
$$ | . . X O . . . O O O O X . |
$$ | X . X O O O O . O . X X . |
$$ | . X O O . O . O O O X O . |
$$ | X X X X X O X X X O X X . |
$$ | O O O O X . X . X X . X . |
$$ | . . O X . X . X O X . X . |
$$ | . . O X 1 X . X O X X X . |
$$ | . O O . O X . X O 2 O X . |
$$ | . X O O X X X O O X . X . |
$$ | . O . O O X X . O X X X O |
$$ | . . O C 3 . X O O O O X X |
$$ | . . O O X X X O . O O X . |
$$ | . . O 4 W O X O O . O X . |
$$ ---------------------------[/go]


:w5: at :wc:

The score differences between the two variations lie around :w3: and :b4:. In variation 1 Black gets 1 point (marked). In variation 2 Black gets 2 points for the White prisoners, while White gets 1 point (marked) plus a Black prisoner, for a net local score of 0. Plainly, variation 2 is better for White. :)

Click Here To Show Diagram Code
[go]$$W Solution
$$ ---------------------------
$$ | . . X O . . . O O . . . . |
$$ | X . X O O O O . O . X X . |
$$ | . X O 7 . 9 . O O O X O . |
$$ | . X X X X . X . X O 8 X . |
$$ | O O O O X . X . X X . X . |
$$ | . . O X . X . X O X . X . |
$$ | . . O X 1 X . X O X X X . |
$$ | . O O . O X . X O 2 O X . |
$$ | . X O O X X X O O X . X . |
$$ | . O . O O X X . O X X X O |
$$ | . . O . 3 . X O O O O X X |
$$ | . . O O X X X O . O O X . |
$$ | . . O 4 W O X O O . O 6 . |
$$ ---------------------------[/go]


:w5: at :wc:

Click Here To Show Diagram Code
[go]$$Bm10 Solution {continued}
$$ ---------------------------
$$ | . . X O . . . O O 6 4 5 . |
$$ | X . X O O O O . O . X X . |
$$ | . X O O . O . O O O X O . |
$$ | 3 X X X X 2 X 1 X O X X . |
$$ | O O O O X . X . X X . X . |
$$ | . . O X . X . X O X . X . |
$$ | . . O X O X . X O X X X . |
$$ | . O O . O X . X O X O X . |
$$ | . X O O X X X O O X . X . |
$$ | . O . O O X X . O X X X O |
$$ | . . O . O . X O O O O X X |
$$ | . . O O X X X O . O O X . |
$$ | . . O . O . X O O . O X . |
$$ ---------------------------[/go]


A count reveals that White does indeed win. :)

Re: Basic endgame problem -- feedback requested

Posted: Tue Feb 01, 2011 8:24 pm
by Bill Spight
A couple of notes on the problem:

First, most of the plays are miai. :)

That is no surprise, because. . . .

This is a last play problem. Namely, :w5:. (We call :w5: the last play, even though there are more to come, because the rest of the plays are miai. :))

Third, eyeballing can help. :) :w1: and :b2: are obvious.

Fourth, this is a reading problem. Whether to play 3 or 4 cannot be told just by comparing the two plays.

Re: Basic endgame problem -- feedback requested

Posted: Wed Feb 02, 2011 12:12 pm
by unkx80
A bit late into the game... but here are my opinions anyway.

The original problem is not likely to be solvable by a 10k on the first try. The simplified problem may be solvable by a 10k, if given sufficient time. In a 19x19 game where each player has 30 minutes to 1 hour, I suspect there may be insufficient time even for dans to read an endgame problem of the size of the original problem.

I never attempted either problem. It is very likely I get something wrong.

Re: Basic endgame problem -- feedback requested

Posted: Wed Feb 02, 2011 12:59 pm
by Bill Spight
unkx80 wrote:In a 19x19 game where each player has 30 minutes to 1 hour, I suspect there may be insufficient time even for dans to read an endgame problem of the size of the original problem.


Reading a typical endgame with 10 different small positions while in 30 sec. byoyomi might be daunting. ;) Still, with a little knowledge, even a 10 kyu would find this endgame trivial to play correctly. :)

Re: Basic endgame problem -- feedback requested

Posted: Wed Feb 02, 2011 4:27 pm
by lightvector
Bill Spight wrote:
unkx80 wrote:In a 19x19 game where each player has 30 minutes to 1 hour, I suspect there may be insufficient time even for dans to read an endgame problem of the size of the original problem.


Reading a typical endgame with 10 different small positions while in 30 sec. byoyomi might be daunting. ;) Still, with a little knowledge, even a 10 kyu would find this endgame trivial to play correctly. :)


I'm a weak dan player, and reading and exactly counting the result in 30 sec byoyomi is not possible for me. Mainly the counting part, while holding in my head the result of the best play. But I can find the sequence of best play very easily.

I bet I could teach a 10k to easily do this sort of thing too. For me, it was enough just to know the endgame values of some of the common moves by sight. I saw the center move had a 4 vs 0 swing, so 2 points, and saving the white stone was 3 vs 0, so 1.5 points, and the rest was a bunch of 1 pt moves, two 0.75 pt pushes, and some 0.5 pt pushes, which I simply knew the values of. It takes me only a few more seconds to convince myself that nothing tricky will happen, like white cutting at K10 doesn't force black to eventually have to capture the stone.
All times are UTC-07:00
Page 4 of 5

Powered by phpBB® Forum Software © phpBB Limited