home *** CD-ROM | disk | FTP | other *** search
- From: wft@math.canterbury.ac.nz (Bill Taylor)
- Subject: Hypercontrol of the board.
-
- This is actually a followup to "how to kill all enemy groups with non-play?"
- by John Tromp (tromp@cwi.nl), which appeared recently in rec.games.go .
-
- Some time ago I looked at this matter of "hypercontrol" of a board.
- ~~~~~~~~~~~~
- That is, on an (n x n) board, what is the smallest k such that...
-
- 1) Black can play k stones in some appropriate pattern;
- 2) White may play any number of legal moves in succession
- (no suicide positions allowed of course);
- 3) Black can then kill all the white stones, with normal play.
-
- The situation described by (1-3) gives Black "hypercontrol" of the
- board at the end of phase 1.
- For board sizes of n from 2 through to 9 I got the following minima for k;
-
- n : 2 3 4 5 6 7 8 9
- k : 2 3 6 8 12 17 22 28 This last one agrees with John Tromp's,
- though I have not seen his solution.
- I'm reasonably confident of the solutions up to n = 7, but not beyond.
- The patterns for n = 2,3 and 5 are unique, but there are many variations
- otherwise. Here are some of the minimal patterns. I'd welcome improvements.
-
-
- n=2 . X n=3 . X . n=4 . X . . . X . .
- X . . X . X . X . . X . .
- . X . . X . X X X X X
- . . X . . . . .
-
-
- n=5 . X . . . n=6 . X . . . . n=7 . X . . . X .
- X . X . . X . X . . . . X . . . X .
- . X . X . . X . X X . . X . . . X .
- . . X . X . . X . X . . X X X X X .
- . . . X . . . X X . X . X . . . X .
- . . . . X . . X . . . X .
- . X . . . X .
-
- n=8 n=9
- . . X . . X . . . X . . . . . . .
- . X X . . X X . . X X X X X X X .
- . . X . . X . . . X . . . . . . .
- . . X . . X . . . X . . . . . . .
- . . X . . X . . X X X X X X X X .
- . . X X X X . . . X . . . . . . .
- . X X . . X X . . X . . . . . . .
- . . X . . X . . . X X X X X X X .
- . X . . . . . . .
-
-
- 2
- It's interesting to note that minimum(k) appears to be n /3 (or just over).
-
- Perhaps this shouldn't be surprising, as apparently black can gain
- hypercontrol of an infinite (or just enormous) board, with just over one third
- of the points occupied, with the pattern
-
- . . . . . . . . . . . . . . . .
- X X X X X X X X X X X X X X X X
- . . . . . X . . . . . . . . . .
- . . . . . X . . . . . . . . . .
- X X X X X X X X X X X X X X X X
- . . . . . . . . . . . . . . . .
- . . . . . . . . . . . . . . . .
- X X X X X X X X X X X X X X X X
- . . . . . X . . . . . . . . . .
- . . . . . X . . . . . . . . . .
- X X X X X X X X X X X X X X X X
- . . . . . . . . . . . . . . . .
-
- with the lines extended indefinitely left and right, and repeated above
- and below. This pattern appears to be 'minimal' for the infinite board.
-
- Further comments would be most appreciated.
-
-
-
- From: hansm@cs.kun.nl (Hans Mulder)
- Subject: Re: Hypercontrol of the board.
-
- How about:
-
- n=7 . . . . . . . n=8 . . . . . . . . n=9 . . . . . . . . .
- . X . . . X . . X X . . X X . . X X X . . X X .
- . X . . . X . . . X . X X . . . . . X X . X . .
- . X X X X X . . . X X . X X . . . . X . X X . .
- . X . X . X . . . X . X X . . . X X X X . X X .
- . X X . X X . . . X X . X . . . . . X . X X . .
- . . . . . . . . X X . . X X . . . . X . . X . .
- . . . . . . . . . X X X . . X X .
- . . . . . . . . .
-
- Using 16, 21 and 27 stones, respectively.
-
- The Dutch national record for n=19 is 120, by Ger Hungerink (3-dan) in
- Go magazine 15/4, page 42:
-
- n=19 . . . . . . . . . X . X . . . . . . .
- . X X X X X X X X X X . X . . X . . .
- . X . . X . . X . . X X X X X X X X .
- . X . . X . . X . . X . . X . . X . .
- . X . . X . . X . . X . . X . . X . .
- . X . . X . . X . . X . . X . . X X .
- . X . . X . . X . . X . . X . . X . .
- . X . . X . . X . . X . . X . . X . .
- . X . . X . . X . . X . . X . . X X .
- . X . . X . . X . . X . . X . . X . .
- . X . . X . . X . . X . . X . . X . .
- . X . . X . . X . . X . . X . . X X .
- . X . . X . . X . . X . . X . . X . .
- . X . . X . . X . . X . . X . . X . .
- . X . . X . . X . . X . . X . . X X .
- . X . . X . . X . . X . . X . . X . .
- . X . . X . . X . . X . . X . . X . .
- . X . . X . . X . . X . . X . . X X .
- . . . . . . . . . . . . . . . . . . .
-
-
-
- From: t68@nikhefh.nikhef.nl (Jos Vermaseren)
-
- I am sorry to say this, but none of the above is hypercontrol:
-
- n=7 . O O O O O .
- O X . . . X O
- O X . . . X O
- O X X X X X O
- O X . X . X O
- O X X . X X O
- . O O O O O .
-
- and white lives. You will have to add a stone on the edge, like in the
- 19*19 example
-
-
- n=7
- . X . . . . .
- X . X . . X .
- . X . X . X .
- . . X . X X .
- . . . X . X .
- . X X X X . .
- . . . . . . .
-
- This is a solution of 16. Haven't found a better one for 8 yet though.
-
- From: tromp@cwi.nl (John Tromp)
- Subject: Re: Hypercontrol of the board.
-
- The following position with only 26 stones almost solves the problem
- (white can still form a `double-headed dragon', as it seems to be called):
-
- . . . . . . . . .
- . X X X X X X X .
- . . . . . . . X .
- . . . . . . . X .
- . X X X X X X X .
- . . . . . X . X .
- . . . . X . X . .
- . X X X X X X . .
- . . . . . . . . .
-
- But adding another stone appropriately on the edge prevents this,
- which brings the 9x9 record down to 27.
-
- Another formation achieving this bound, inspired by Ger Hungerink:
-
- . . . . . . . . .
- . X X X X X X X .
- . . . . . . . X .
- . . . . . . . X X
- . X X X X X X X .
- . . . . . . X . X
- . . . . . . X X .
- . X X X X X X . .
- . . . . . . . . .
-
-
-
- From: wft@math.canterbury.ac.nz (Bill Taylor)
- Subject: HYPERCONTROL : summary and extension.
-
- FIRSTLY, thanks to all who responded to the initial posting on
- this topic. I'm glad so many were interested. I saw replies from...
-
- tromp@cwi.nl (John Tromp)
- hansm@cs.kun.nl (Hans Mulder)
- Klaus Reinhardt <reinhard@orchidee.informatik.uni-stuttgart.de>
- t68@nikhefh.nikhef.nl (Jos Vermaseren)
-
- ..and others (that don't seem to be in my file); apologies for omissions.
-
- SECONDLY, a correction to my original posting. Although no-one picked
- it up, I incorrectly stated...
-
- >The patterns for n = 2,3 and 5 are unique,
- >
- >n=2 . X n=3 . X .
- > X . . X .
- > . X .
- >
- >
- >n=5 . X . . .
- > X . X . .
- > . X . X .
- > . . X . X
- > . . . X .
-
- This was correct for 2 & 3, but not for 5; where there are at least two
- other patterns with 8 stones. Those (like me) interested in such things
- might like to try to find them.
- -------------------------------
-
- SUMMARY: the minimal numbers for hypercontrol of square boards seem to be
- ~~~~~~~
- width: 2 3 4 5 6 7 8 9 13 19
- number: 2 3 6 8 12 16 21 27 56 120
-
- ------
-
- EXTENSION. The above table is just the main diagonal of a table for minimal
- ~~~~~~~~~ hypercontrol numbers for rectangular boards.
-
- RECTANGULAR BOARDS: these do not get much attention from go players, but
- they are of considerable interest for those who delve into optimal
- small-board play, hypercontrol, and similar matters.
-
- For (m x n) boards, the minimal numbers for hypercontrol seem to be...
-
- n = 1 2 3 4 5 6 7 8 9 10 11 ... general term
- -------------------------------------------------------------------------------
- m = 1 | * * 1 2 2 3 3 4 4 5 5 ... [n/2] n >= 3
- 2 | 2 2 4 4 5 6 6 7 8 9 ... [6n/7] n >= 5
- 3 | 3 4 5 6 7 8 9 10 11 ... n
- 4 | 6 7 8 9 11 12 13 15 ... [4n/3] n >= 5
- 5 | 8 10 12 13 15 17 18 ... \
- 6 | 12 14 16 18 20 22 ... |
- 7 | 16 19 21 23 26 ... |
- 8 | 21 24 27 29 ... > [mn/3] m,n >= 5
- 9 | 27 30 33 ... |
- 10 | 33 37 ... |
- 11 | 40 ... /
- 12 |
-
- So; for m,n >= 3 , the general term is [mn/3], { using [x] for floor(x) }.
- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- (The sole exception being at 4x4).
-
- The rows for n=1 and n=2 are quite different from the rest. ( Indeed `go' on
- 1xn and 2xn boards is quite unlike regular go, as small-board analyzers will know;
- especially (1xn) go, where it is almost suicidal EVER to play on an end-point ! )
-
- -------------
- m=1 has unique patterns like . X . X . X . and . X . X X . X .
- ------------
- m=2 has patterns with limiting density 3/7 ;
-
- a typical case being 2x24 : . . X X . X . . . X X . X . . . X X . X X X . .
- . . X . X X . . . X . X X . . . X . X X . . . .
- which type serves down to n=7.
-
- and X . X . . . X . X . . X . X .
- n=4,5,6 have . X X . . . X X . . . X X . X
- --------------
- The minimal patterns for m=3 are trivial, like . . . . .
- X X X X X
- ------------- . . . . .
-
- The minimal patterns for m=4 are of 3 types, depending on (n mod 3);
-
- 4x11 . X X . . . . . . . .
- X . X . X . . X . . .
- . X X X X X X X X X .
- . . . . . . . . . . .
-
- 4x12 . X X . . . . . . . . .
- X . X . . X . . X . . .
- . X X X X X X X X X X .
- . . . . . . . . . . . .
-
- 4x13 . . X . X . . . . . . . .
- . X X X . X . . X . . X .
- . . . X X X X X X X X X .
- . . . . . . . . . . . . .
-
- These patterns serve right down to n=5.
- -------------
-
- For m,n >= 5 , there are 4 typical patterns (which cover all cases).
-
- A: m=3k
- B: m=3k+1 n=3j+1
- C: m=3k+1 n=3j+2
- D: m=3k+2 n=3j+2
-
- The patterns for these are based on those by Klaus Reinhardt and Ger Hungerink.
-
- A: 9x11 . . . . . . . . . . . B: 10x10 . . . . . . . . . .
- . . X X X X X X X X . . . X . . X . . X .
- . . X . . . . . . . . . . X X X X X X X .
- X X X . . . . . . . . . X X . . . . . . .
- X . X X X X X X X X . X . X . . . . . . .
- . X . . . . . . . . . . X X X X X X X X .
- X X . . . . . . . . . X X . . . . . . . .
- . X X X X X X X X X . . X . . . . . . . .
- . . . . . . . . . . . . X X X X X X X X .
- . . . . . . . . . .
-
-
- C: 10x11 . . . . . . . . . . . D: 11x11
- . . . X . . X . . X . . X . . . . . . . . .
- . X X X X X X X X X . X . X X . . X . . X .
- X . X . . . . . . . . . X . X X X X X X X .
- . X X . . . . . . . . . . X X . . . . . . .
- X X X X X X X X X X . . . X . . . . . . . .
- . X . . . . . . . . . . X X X X X X X X X .
- . X . . . . . . . . . . . X . . . . . . . .
- . X X X X X X X X X . . . X . . . . . . . .
- . . . . . . . . . . . . X X X X X X X X X .
- . . . X . . X . . X .
- All of these can clearly be simply extended by . . . . . . . . . . .
- threes, in either dimension; and can be similarly
- reduced down as far as the case 5x5 .
- ------------------------------------------
-
- Well; there it all is. Probably no-one else is interested.
-
- The above diagrams show that the numbers given in the table are at least
- correct upper bounds. They do not, of course, provide a proof that the
- table is exactly right.
-
- If anyone can find patterns of hypercontrol that use fewer stones than
- given in the table, I will be very delighted.
-
- Have fun.
- ------------------
- Bill Taylor. wft@math.canterbury.ac.nz
-
-