It is currently Thu Apr 25, 2024 12:33 am

View unanswered posts | View active topics View unread posts | View new posts | View your posts


Board index » Improve Your Game » Study Group

All times are UTC - 8 hours [ DST ]



Post new topic Reply to topic  Page 1 of 1
  [ 10 posts ] 
  Print view Previous topic | Next topic 
Author Message
Offline Bill Spight
 Post subject: Opening problems for AI: Problem 41
Post #1 Posted: Wed Aug 05, 2020 10:27 am 
Honinbo

Posts: 10905
Liked others: 3651
Was liked: 3374
Black to play.

Click Here To Show Diagram Code
[go]$$Bcm23 Black to play
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . a O . . . |
$$ | . . . . . . . . . . . . . O . X O . . |
$$ | . . . O . . . . . , . . . . . X b . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . X . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . X O . . . . . , . . . . . X . . . |
$$ | . . X O O X . O . . O . . . . . . . . |
$$ | . . X X O . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

Should Black play the block at a or the turn at b?

According to Elf, one of them loses 12½% by comparison with the other.

Enjoy!

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.
Top
 Profile  
 
Offline Gomoto
 Post subject: Re: Opening problems for AI: Problem 41
Post #2 Posted: Wed Aug 05, 2020 10:44 am 
Gosei

Posts: 1733
Location: Earth
Liked others: 621
Was liked: 310
I seem to never get this decision right in relation to the board. At least my choice looses this time only half a point according to Kata.

Putting a white or black stone around K16 and asking the oracle again is also interesting.


Last edited by Gomoto on Wed Aug 05, 2020 10:51 am, edited 5 times in total.
Top
 Profile  
 
Offline Knotwilg
 Post subject: Re: Opening problems for AI: Problem 41
Post #3 Posted: Wed Aug 05, 2020 10:44 am 
Oza
User avatar

Posts: 2411
Location: Ghent, Belgium
Liked others: 359
Was liked: 1019
Rank: KGS 2d OGS 1d Fox 4d
KGS: Artevelde
OGS: Knotwilg
Online playing schedule: UTC 18:00 - 22:00
Edited: now I'm in doubt.

Click Here To Show Diagram Code
[go]$$Bcm23 Black to play
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . 1 O 2 . . |
$$ | . . . . . . . . . . . . . O 3 X O . . |
$$ | . . . O . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . 5 4 . |
$$ | . . . . . . . . . . . . . . . . . 6 . |
$$ | . . . . . . . . . . . . . . . . B . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . X . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . X O . . . . . , . . . . . X . . . |
$$ | . . X O O X . O . . O . . . . . . . . |
$$ | . . X X O . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]


After this sequence, Blak has to take gote or allow the marked stone to be cut.

Click Here To Show Diagram Code
[go]$$Bcm23 Black to play
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . 3 1 O 2 . . |
$$ | . . . . . . . . . . . a . O . X O . . |
$$ | . . . O . . . . . , . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . 5 4 . |
$$ | . . . . . . . . . . . . . . . . 7 6 . |
$$ | . . . . . . . . . . . . . . . . B . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . X . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . X O . . . . . , . . . . . X . . . |
$$ | . . X O O X . O . . O . . . . . . . . |
$$ | . . X X O . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]


If Black is going to cover, then probably it's best to play :b25: this way.

Click Here To Show Diagram Code
[go]$$Bcm23 Black to play
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . O 4 2 . |
$$ | . . . . . . . . . . . . . O . X O 3 . |
$$ | . . . O . . . . . , . . . . . X 1 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . X . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . X O . . . . . , . . . . . X . . . |
$$ | . . X O O X . O . . O . . . . . . . . |
$$ | . . X X O . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]


Black takes sente. White stabilizes easily but so does Black. Maybe this is better after all.


Last edited by Knotwilg on Thu Aug 06, 2020 2:07 am, edited 1 time in total.
Top
 Profile  
 
Offline Bill Spight
 Post subject: Re: Opening problems for AI: Problem 41
Post #4 Posted: Wed Aug 05, 2020 10:49 am 
Honinbo

Posts: 10905
Liked others: 3651
Was liked: 3374
Gomoto wrote:
I seem to never get this decision right. At least my choice looses only half a point according to Kata.


If you lose ½ pt. per move on average over 120 moves, that comes to 60 pts. ;)

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

This post by Bill Spight was liked by: Gomoto
Top
 Profile  
 
Offline Uberdude
 Post subject: Re: Opening problems for AI: Problem 41
Post #5 Posted: Wed Aug 05, 2020 3:17 pm 
Judan

Posts: 6725
Location: Cambridge, UK
Liked others: 436
Was liked: 3719
Rank: UK 4 dan
KGS: Uberdude 4d
OGS: Uberdude 7d
Bill Spight wrote:
Gomoto wrote:
I seem to never get this decision right. At least my choice looses only half a point according to Kata.


If you lose ½ pt. per move on average over 120 moves, that comes to 60 pts. ;)


Good enough to beat someone who loses 0.6 points on average by 12 points!

This post by Uberdude was liked by 3 people: Bill Spight, Gomoto, Joaz Banbeck
Top
 Profile  
 
Offline ez4u
 Post subject: Re: Opening problems for AI: Problem 41
Post #6 Posted: Wed Aug 05, 2020 10:56 pm 
Oza
User avatar

Posts: 2401
Location: Tokyo, Japan
Liked others: 2339
Was liked: 1332
Rank: Jp 6 dan
KGS: ez4u
Of course I was wrong in my assessment. :blackeye:

However, I was interested because the large knight's move is an unusual response to the approach these days. The two-space extension at the bottom whispered Lee Changho's name, so this was a relatively recent example. I was interested in what difference the large knight's move made. I cycled through various alternatives for Black's reply to White's approach (four extensions and six pincers) and produced the table below.
This table summarizes the results of using katago and analyzing four extensions: large and small knight's move and 1 and 2-space high, plus six pincers: 1, 2, and 3-space in both high and low alternatives.

First we see whether or not katago calculates White should push once at R16 or hane immediately at Q18. In the game, for example, White played the immediate hane but this was 6.5% (35.1% vs 41.6%) and 0.8 points (-1.7 vs -0.9) worse than pushing at R16 by katago's calculation.

Second we see the situation after an immediate Q18 (regardless of whether this would be White's best choice). We compare the merits of Black turning at R18 or blocking at P16.

The better alternative is highlighted in all cases (bold and colors).

I do not think there are any grand revelations here. The most interesting element for me was how the top becomes more valuable as the pincers get wider until the 3-space high switches from the turn to the block.
Attachment:
Table of Alternatives for AI 41.jpg
Table of Alternatives for AI 41.jpg [ 82.52 KiB | Viewed 6943 times ]

All the calculations were made with the 40-block but without a lot of depth - normally 6K-10K playouts. I ran this from the original game file in GoGoD (1998-02-05c.sgf), so the komi = 5.5 points.

_________________
Dave Sigaty
"Short-lived are both the praiser and the praised, and rememberer and the remembered..."
- Marcus Aurelius; Meditations, VIII 21

This post by ez4u was liked by 2 people: Bill Spight, Uberdude
Top
 Profile  
 
Offline Uberdude
 Post subject: Re: Opening problems for AI: Problem 41
Post #7 Posted: Wed Aug 05, 2020 11:45 pm 
Judan

Posts: 6725
Location: Cambridge, UK
Liked others: 436
Was liked: 3719
Rank: UK 4 dan
KGS: Uberdude 4d
OGS: Uberdude 7d
Nice analysis Dave, confirms my understanding that White should push r16 if black can't block with good shape, and black punishes White not pushing where White should have with the turn, with 3 space high pincer only exception. We can see how as the pincer gets looser p18 intercept becomes less bad and eventually good, presumably a matter of efficiency.

This post by Uberdude was liked by 2 people: Bill Spight, ez4u
Top
 Profile  
 
Offline Bill Spight
 Post subject: Re: Opening problems for AI: Problem 41
Post #8 Posted: Sat Aug 08, 2020 9:21 am 
Honinbo

Posts: 10905
Liked others: 3651
Was liked: 3374
The game was GoGoD 1998-02-05c, between Lee Changho (W) and Yu Ch'ang-hyeok.

As Dave indicates, the 2d line hane for :w22: is a minor error, losing 7% to the 3d line crawl, according to Elf. The mainline variation for :w22: will give us something to compare against the game continuation and the mainline variation for :b23:.

Click Here To Show Diagram Code
[go]$$Wcm22 Mainline for :w22:
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . 9 4 3 5 . . |
$$ | . . . . . 6 . . . . . . 8 O 7 X O . . |
$$ | . . . O . . . . . , . . . . . X 1 . . |
$$ | . . . . . . . . . . . . . . . 2 . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . X . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . X O . . . . . , . . . . . X . . . |
$$ | . . X O O X . O . . O . . . . . . . . |
$$ | . . X X O . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

:b23: could also be at 24. :b27: switches to the top left corner, but :b27: at 30 is also playable. :w28: captures :b25:.

Click Here To Show Diagram Code
[go]$$Bcm31 Mainline for :w22:, continued
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . 5 . . . . . . O X O O . . |
$$ | . a 1 3 . X . . . . . . X O O X O . . |
$$ | . . 2 O 4 . . . . , . . . . . X O . . |
$$ | . . . . . . . . . . . . . . . X . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . . 6 . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . X . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . X O . . . . . , . . . . . X . . . |
$$ | . . X O O X . O . . O . . 7 . . . . . |
$$ | . . X X O . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

:b31: invades the top left corner. :w32: blocks on the left side, but, as usual, blocking at 33 is also playable. :w34: makes the shape play, preventing a Black bulge with sente. :b35: could also descend at a. Then :w36: makes a 3 space high extension. Next, :b37: encloses the bottom right corner.

Game continuation

Click Here To Show Diagram Code
[go]$$Bcm23 Mistake
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . 1 O 2 . . |
$$ | . . . . . . . . . . . . . O 3 X O . . |
$$ | . . . O . . . . . , . . . 5 . X . . . |
$$ | . . . . . . . . . . . . . . . . . 4 . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . B . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . X . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . X O . . . . . , . . . . . X . . . |
$$ | . . X O O X . O . . O . . . . . . . . |
$$ | . . X X O . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

The block on the top side was a mistake. :w24: - :b27: were also Elf's top choices. White has taken the corner and Black has made a solid shape. To me, :bc: seems a bit misplaced.

Top choice: Block on the right side.

Click Here To Show Diagram Code
[go]$$Bcm23 Mainline for :b23:
$$ ---------------------------------------
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . O 4 2 . |
$$ | . . . . . . . . . . . . . W . X O 3 . |
$$ | . . . O . . . . . , . . . . . X 1 . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . B . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . , . . . . . , . . . . . , . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . X . . |
$$ | . . X . . . . . . . . . . . . . . . . |
$$ | . . . O . . . . . . . . . . . . . . . |
$$ | . . X O . . . . . , . . . . . X . . . |
$$ | . . X O O X . O . . O . . . . . . . . |
$$ | . . X X O . . . . . . . . . . . . . . |
$$ | . . . . . . . . . . . . . . . . . . . |
$$ ---------------------------------------[/go]

Blocking on the right side allows White to connect out to :wc:, but Black takes sente, and :bc: looks nicely placed. :)

Not too much to add, I think. With the right hand block the local position looks better to me, and the sente/gote difference is significant.

The mainline for :w22: looks more interesting, but the hane was only a minor error. ;)

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.
Top
 Profile  
 
Offline Uberdude
 Post subject: Re: Opening problems for AI: Problem 41
Post #9 Posted: Sat Aug 08, 2020 2:12 pm 
Judan

Posts: 6725
Location: Cambridge, UK
Liked others: 436
Was liked: 3719
Rank: UK 4 dan
KGS: Uberdude 4d
OGS: Uberdude 7d
Shoulda guessed it was Yu Ch'ang-hyeok with the large knight moves.
Top
 Profile  
 
Offline dhu163
 Post subject: Re: Opening problems for AI: Problem 41
Post #10 Posted: Mon Dec 26, 2022 8:47 am 
Lives in gote

Posts: 470
Liked others: 62
Was liked: 278
Rank: UK 2d Dec15
KGS: mathmo 4d
IGS: mathmo 4d
Speculatively:
Current position:
1. Slightly surprising. a is the standard joseki, though b seems playable. However, B doesn't have support on the upper side and has territory on the right. So perhaps b is sensible to get sente for the upper left 3-3? Rather than gote? However, I still prefer the look of a to move potential to the upper side and reduce W's influence rather than continuing trying to take a small side territory when the lower right remains open. So I prefer a.

Opp to play
1. R16 obviously.

Past:
1. B's large knight's aren't liked by AI unless there is good reason. In this shape, there seems to be no reason as there is no expansion potential when W is strong around. The lower side is the Lee Chang-ho joseki before AI popularised F2.

Check:
1. Hmm, some things to learn here. Perhaps in a game I would have thought more.
Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  Page 1 of 1
  [ 10 posts ] 

Board index » Improve Your Game » Study Group

All times are UTC - 8 hours [ DST ]

Who is online

Users browsing this forum: No registered users and 1 guest

You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group