The New Sudoku Players' Forum

[JPF]

...in a list papy sent to ruud there were 2 not in gordon's catalog

Code: Select all

120000080400900000000007003097000500000084000000000000800020010000500900000000000
120006000400000010000700500005010000006000900000004000008900000600000000000000042

these are the first new entries in over a year

Félicitations Papy!
You should tell us how you do it !

In the meantime, I made this study :

The strange familiar grid (SF) has been carefully studied in many threads of this forum.
This grid includes 29 non isomorphic 17 clues puzzles.

Code: Select all

SF
123456789456789132789213645275341968314968257968572413537194826642837591891625374


In the actual Gordon's list of the known 17s, there are other grids containing more than one essentially-different 17 clues puzzle.

Here's the list of the non isomorphic grids with more than 5 non isomorphic 17s :

Code: Select all

29  123456789456789132789213645275341968314968257968572413537194826642837591891625374
20  123456789456789123789132564264918357875324691931675248392861475547293816618547932
14  123457689456189273789236451265713948348965712971842536534698127612374895897521364
12  123456789456789132789213546264597813597138264831624957375862491648971325912345678
11  123456789456798132789231546237615894815924673964873251342567918591382467678149325
9   123456789456789132789213546267594813594138267831627954375862491648971325912345678
8   123456789456789123789132546231864975864597231975321468347918652592643817618275394
8   123456789456789231789231564214895673395627418867143952531974826648512397972368145
8   123457689456189327789236451271364895368591742594728136642875913837912564915643278
7   123456789456789132789213456271594863638127945945638217397845621514962378862371594
7   123456789456789132789231546247193658368574291591862374635928417812347965974615823
7   123456789456798132789231546264875391831629475975143268348912657592367814617584923
7   123457689456189237789632154245798316371265948968341572594816723637524891812973465
7   123457689456189273789263451267815394395674128814392567538746912641928735972531846
7   123457689456189372789362415261743958375918246948625731532876194617594823894231567
7   123457689456189732789362154238795461517846923964231578371524896642978315895613247
6   123456789456789123789132465215967348347518296698324517561843972874295631932671854
6   123456789456789123789132465217895346394627851568341297642513978835974612971268534
6   123456789456789123789132564267314958531968247894527316342675891675891432918243675
6   123456789456789132789123546231867954675941328948532671367294815592318467814675293
6   123456789456789132789123546275348961638291475941567328392675814567814293814932657
6   123456789456789132789132546267593418391847625845621973532974861678215394914368257
6   123456789456789132789132564234875691598361427671294358362917845847523916915648273
6   123456789456789132789132564261894357537621498948375216394567821675218943812943675
6   123456789456789132789213456214895673538627941967341825392174568671538294845962317
6   123456789456789132789213654248197563615348297937562418364925871592871346871634925
6   123456789456789132789231546237915468591864273648327915372548691814692357965173824
6   123456789456789132789231546278314695395678214641592378534967821862143957917825463
6   123456789456789132789231546295814673674392815831675924367928451548163297912547368
6   123456789456789231789231564215347896378695412964128357537914628692873145841562973
6   123456789456798132789213546247589613538167924961324857395641278672835491814972365
6   123456789456798132789213546267345891395681274814927365572869413648132957931574628
6   123456789456798132789213546274935861391682457865147293537829614618374925942561378
6   123457689456189237789263154268935741394716528517842963671324895842591376935678412
6   123457689456189237789632514274963851618524973935718426341295768567841392892376145
6   123457689456189327789632541237945168864713295915826473342598716578261934691374852



Here's the number N of non isomorphic grids with n non isomorphic 17 puzzles.

Code: Select all

N      n

1      29
0      28
0      27
0      26
0      25
0      24
0      23
0      22
0      21
1      20
0      19
0      18
0      17
0      16
0      15
1      14
0      13
1      12
1      11
0      10
1      9
3      8
7      7
20     6
18     5
80     4
226    3
1504   2
32244  1
-----
36628
     


JPF

[gsf]

here are the box-normal canonical grids for the two new papy 17's

Code: Select all

123456789456789132789231645237915864591648273648327951315872496864193527972564318
123456789456789231789132546215648973364917825978325614547291368691873452832564197


the first one is in gordon's list but the second is not

[tarek]

Finally, something materialised out of this.........

Papy's trials were eventually friutful......I'm glad that the language/communication barriers were overcome by Papy's tenacity & the Players' help, well done.

tarek

[RW]

here are the box-normal canonical grids for the two new papy 17's
...
the first one is in gordon's list but the second is not

So it's still possible to find new 17s, even from grids already on the list... Would be nice to see the previous 17(s) from that first one, to see how much it differs from Papy's new puzzle.

RW

[gsf]

Would be nice to see the previous 17(s) from that first one, to see how much it differs from Papy's new puzzle.

there was only one for that grid in gordons catalog
here are the submitted forms for gordon and papy

Code: Select all

000000081700050000020000000601200000080000040000090300300004000200000500000800000
120006000400000010000700500005010000006000900000004000008900000600000000000000042


and the box-normal canonical puzzles (based on the box-normal canonical grid) in the same order

Code: Select all

000400000000700030009000600000000004000008070608020001300000000000090500070000008
000400000000700030009000600000000004000008073608020000300000000000090500070000008


[RW]

and the box-normal canonical puzzles (based on the box-normal canonical grid) in the same order

Code: Select all

000400000000700030009000600000000004000008070608020001300000000000090500070000008
000400000000700030009000600000000004000008073608020000300000000000090500070000008

Only one clue different, I thought Gordon had checked all those already... I wonder if there's more puzzles around the existing 17s that he has missed.

RW

[m_b_metcalf]

In a totally vain attempt to win at JPF's Puzzle Game I incidentally, while testing some new code, came up with a 17-clue that is not in Gordon's list, but which I suppose is isomorphic to one of them. Could I ask someone to kindly check?

Code: Select all

000000010600000000040000000000050602008000300001070000300600200050100000000809000


Thanks,

Mike Metcalf

[JPF]

Could I ask someone to kindly check?

Code: Select all

000000010600000000040000000000050602008000300001070000300600200050100000000809000


#4185 in the Gordon's list

Code: Select all

000041000009000007050000300000000541000708000000000060000530200100000000000200000


JPF

[coloin]

Well done Papy.
Apart from gfroyle, some unreferenced Japenese source and Wolfgang, to my knowledge no one else has found a 17.

Even before Checker give us any uniques.......

I will not take any credit for the third 17 in the first of Papy's grids

Code: Select all

...4........7...3...9...6..........4.....8.736.8.2....3............9.5...7......8
...4........7...3...9...6..........4.....8.7.6.832....3............9.5...7......8
...4........7...3...9...6..........4.....8.7.6.8.2...13............9.5...7......8

C

[RW]

A third one! Looks as though I picked the wrong grid, just ran checker on the other of Papys grids, only one there. Did you use checker, or did you just try adding and removing clues? Could there be even more in this grid?

Btw, setting r8c7=1 gives three new ones. Are those three puzzles already in the list?
...4........7...3...9...6..........4.....8.736.8.2....3............9.1...7......8
...4........7...3...9...6..........4.....8.7.6.8.2...53............9.1...7......8
...4........7...3...9...6..........4.....8.7.6.832....3............9.1...7......8

Actually, play around with r8c7 and r6c5 and the values 1,2 and 5 and there's six combinations that yield valid 17s with r5c9=3. For all of these grids you get three valid 17s with the same masks as the three for Papy's puzzle.

RW

[coloin]

No checker is still running - no uniques found as yet
No I ran a "suexmult" on the 16 common clues.

There were no more found at the 15 clue level
.................................

I wonder if there's more puzzles around the existing 17s that he has missed.

Yes, I suspect not many, but there is only one way to find out !

The reason why no one has found a 17 is that it is so difficult - starting from a likely grid less than 1 in 100000 will have a 17. Even when you know a grid has a 17 it is difficult to tease it out by analysing the slightly increased frequency of clues in randomly generated 21 and 22 clue sudoku puzzles in the grid.

With the improve version of checker this can be done quite quickly in many grids.

Since Gordon's program was so productive, no one really bothered, and since the generation of new 17s withered we lost interest, Gordon was intent on finding a 16, but when this became statistically unlikly we all gave up.....

With our improved puzzle generation techniques I would think it would not be too difficult to get a rough idea as to how complete Gordons collection is.

We could easily search fo all the puzzles in one 17-class [not yet classified]

From earlier on in this thread, the distribution of clues and box occupancy was.
Clues

Code: Select all

      7 0 1 1 1 2 2 3 3 4
    175 0 1 1 1 2 3 3 3 3
    197 0 1 1 2 2 2 2 3 4
   5908 0 1 1 2 2 2 3 3 3
     12 0 1 2 2 2 2 2 2 4
   8377 0 1 2 2 2 2 2 3 3
    730 0 2 2 2 2 2 2 2 3
     10 1 1 1 1 1 3 3 3 3
     81 1 1 1 1 2 2 2 3 4
   2220 1 1 1 1 2 2 3 3 3
     13 1 1 1 2 2 2 2 2 4

  11091 1 1 1 2 2 2 2 3 3
   4104 1 1 2 2 2 2 2 2 3
      5 1 2 2 2 2 2 2 2 2

Boxes

Code: Select all

    2 001122344:B
    2 001222334:B
    2 001223333:B
    2 002222234:B
   12 002222333:B
    7 011112344:B
    4 011113334:B
    2 011122235:B
   17 011122244:B
  177 011122334:B
  108 011123333:B
    2 011222225:B
  381 011222234:B
  946 011222333:B
  100 012222224:B
  945 012222233:B
  157 022222223:B
    1 111112235:B
   22 111112244:B
  486 111112334:B
  135 111113333:B
 1341 111122234:B
 2904 111122333:B
  659 111222224:B
11342 111222233:B
11769 112222223:B
 1405 122222222:B

Is it beyond our means to pick a common combination of each of these charactaristics and enumerate all the non-isomorphic 17s ?

I suspect that all the 17 puzzles have these 6,7 or even 8 equivalent clues....

Code: Select all

+---+---+---+
|1..|...|...|
|...|2..|...|
|...|...|3..|
+---+---+---+
|.4.|...|...|
|...|.5.|...|
|...|...|...|
+---+---+---+
|..6|...|...|
|...|..7|...|
|...|...|...|
+---+---+---+

Its difficult to find a puzzle without the first 7 clues [equivalent] as charted.

maybe this observation doesnt contribute much.......

C

[gsf]

I will not take any credit for the third 17 in the first of Papy's grids

Code: Select all

...4........7...3...9...6..........4.....8.736.8.2....3............9.5...7......8
...4........7...3...9...6..........4.....8.7.6.832....3............9.5...7......8
...4........7...3...9...6..........4.....8.7.6.8.2...13............9.5...7......8


the 3d one is already there

Code: Select all

.......817...5.....2.......6.12......8.....4.....9.3..3....4...2.....5.....8.....,684,gfroyle

[gsf]

Are those three puzzles already in the list?
...4........7...3...9...6..........4.....8.736.8.2....3............9.1...7......8
...4........7...3...9...6..........4.....8.7.6.8.2...53............9.1...7......8
...4........7...3...9...6..........4.....8.7.6.832....3............9.1...7......8

these are already in the list (gordon(36627) + wolfgang(1) + papy(3))

[edit: corrected count discrepancies noted by JPF]

[tarek]

From what is there I have already assumed that someone should have already done this.......

1. take each invidual puzzle of the 17-clue catalog
2. 17 different 16 clue templates.
3. plug in an extra clue (81-17)=64 times per template

that shouldn't be that hard work & I assume it was done

tarek

[RW]

the 3d one is already there

I believe coloin had them in the wrong order. The second one was the new puzzle:

Code: Select all

...4........7...3...9...6..........4.....8.7.6.832....3............9.5...7......8

Are those three puzzles already in the list?
...4........7...3...9...6..........4.....8.736.8.2....3............9.1...7......8
...4........7...3...9...6..........4.....8.7.6.8.2...53............9.1...7......8
...4........7...3...9...6..........4.....8.7.6.832....3............9.1...7......8

these are already in the list (gordon(36621) + wolfgang(1) + papy(2))

So papy's second puzzle was also found by just adding one clue and removing another from an existing 17... then there might definitely be more of these. Someone with an apropriate program should do the test tarek suggested.

RW