The SUPERIOR thread

Everything about Sudoku that doesn't fit in one of the other sections

Postby nathanmcmaster » Wed Apr 19, 2006 6:09 pm

i no i agree with you
nathanmcmaster
 
Posts: 5
Joined: 19 April 2006

re: JPF/tarek #1

Postby Pat » Thu Apr 20, 2006 7:06 am

re: JPF/tarek #1

i needed just one duo,
plus box-line interaction.

how can this qualify as a "superior"
under tarek's definition??

~ Pat


tarek wrote:here are two modifications of JPF's

They are superiors



JPF/tarek #1
Code: Select all
 6 . . | 3 1 9 | . . 7 
 . 1 . | 2 . 4 | . 5 . 
 . . . | . . . | . . . 
-------+-------+------
 1 8 . | . . . | . 6 9 
 7 . . | . 2 . | . . 1 
 2 4 . | . . . | . 7 5 
-------+-------+------
 . . . | . . . | . . . 
 . 6 . | 1 . 8 | . 3 . 
 9 . . | 7 5 3 | . . 6 



JPF/tarek #2
Code: Select all
 6 . . | 3 1 9 | . . 7 
 . . . | 2 . 4 | . . . 
 . 7 . | . . . | . . . 
-------+-------+------
 1 8 . | . . . | . 6 9 
 7 . . | . 6 . | . . 1 
 2 4 . | . . . | . 7 5 
-------+-------+------
 . . . | . . . | . 9 . 
 . . . | 4 . 8 | . . . 
 9 . . | 7 5 3 | . . 6 


tarek
User avatar
Pat
 
Posts: 4056
Joined: 18 July 2005

Postby tarek » Thu Apr 20, 2006 7:22 am

I rechecked them Pat..........

The first puzzle needs FOR SURE 3 Naked doubles & (one hidden double or a naked triple) AT LEAST to solve.......depending on how you solve it.... there also a chance for a an x-wing

The second one needs FOR SURE 1 Naked double & 1 X-wing AT LEAST to solve........

Both would score more than 3 & therefore were superiors........

Could you point out that 1 double that solved the puzzle....

tarek
User avatar
tarek
 
Posts: 3762
Joined: 05 January 2006

re(2): JPF/tarek #1

Postby Pat » Thu Apr 20, 2006 7:30 am

tarek wrote:I rechecked them Pat..........

The first puzzle needs FOR SURE 3 Naked doubles & (one hidden double or a naked triple) AT LEAST to solve.......depending on how you solve it.... there also a chance for a an x-wing

Could you point out that 1 double that solved the puzzle....

tarek


well i saw the 5,7 duo in box 9

thus i got the 9 for r8 and for box 9,
and the 1 for r7

then the r8 4 must be in box 7,
so r9c3 = 8

then the 8 for r7



~ Pat
User avatar
Pat
 
Posts: 4056
Joined: 18 July 2005

Postby tarek » Thu Apr 20, 2006 7:51 am

hmmmm,,,

The hidden double in box 9, or the other in column 7 does solve the puzzle.......

Now both of these hidden doubles have a naked triple counterpart....

In the analysis of all puzzles, I (& I think other participants too)...considered that spotting a hidden double or a naked triple a more taxing task (correct me if wrong)..... & therefore you shouldn't spot the hidden double before the naked doubles........

what is your view on this Pat....did you spot the double in row 2 & decided to go for the hidden one in 9, which is easier ???

tarek
User avatar
tarek
 
Posts: 3762
Joined: 05 January 2006

"hidden" vs. "naked"

Postby Pat » Thu Apr 20, 2006 7:53 am

tarek wrote:The hidden double in box 9, or the other in column 7 does solve the puzzle.......

Now both of these hidden doubles have a naked triple counterpart....

In the analysis of all puzzles, I (& I think other participants too)...considered that spotting a hidden double or a naked triple a more taxing task (correct me if wrong)..... & therefore you shouldn't spot the hidden double before the naked doubles........

what is your view on this Pat....did you spot the double in row 2 & decided to go for the hidden one in 9, which is easier ???

tarek


sorry but i usually find the "hidden" stuff easier

~ Pat
User avatar
Pat
 
Posts: 4056
Joined: 18 July 2005

Postby tarek » Thu Apr 20, 2006 8:08 am

hmmmm (for the second time:D ):

I think you have a point Pat......

All techniques used should be absolute..........

The only absolute techniques needed (whenever they occur) are the hidden triples & the x-wings....... the mentiened puzzle was one of the few

The rest are not abosolute........say for instance you spot a hidden triple before a naked double....then the naked double which I thought was essential is now redundant.......

This would just mean reviewing all puzzle in the superior list again:(

but as the number of techniques are not that many.....it should be doable....

I'll check them today....& report the results....

I don't think we should change puzzles now....but we can re-rate them & report the WEAKER ones.........

thanx Pat... the mentioned puzzle now scores only a 2 (which is less than the minimum of 3 found on the list)....

tarek
User avatar
tarek
 
Posts: 3762
Joined: 05 January 2006

Postby ronk » Thu Apr 20, 2006 10:23 am

tarek wrote:All techniques used should be absolute..........

The only absolute techniques needed (whenever they occur) are the hidden triples & the x-wings....... the mentiened puzzle was one of the few

The rest are not abosolute........say for instance you spot a hidden triple before a naked double....then the naked double which I thought was essential is now redundant.......

So "absolute" means necessary? Or "essential"?

Confused, Ron
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Postby tarek » Thu Apr 20, 2006 11:08 am

Pardon me for not choosing the correct words...

Essential is what I was after..........

So if a naked double becomes unneceessary after using a hidden triple for example......then that naked double is not essential......

I'm going to make minor adjustments to the solver to see how different the rankings would be if the above was noted...I'll highlight the ones which score below 2 & probably post a new ranking list.....

It may be harsh on some puzzles & it will definitely downscore some good puzzles...... but consistancy is what we're after

The following paragraph is probably only useful for programmer (skip it if it's not your cup o' tea):

The solver now runs 5 times, (you could make it 6 for box-line interactions)... the technique placed in the outermost loop that registers an eliminations Is essential (hence 5 times)....

tarek
User avatar
tarek
 
Posts: 3762
Joined: 05 January 2006

Postby ronk » Thu Apr 20, 2006 11:45 am

tarek wrote:So if a naked double becomes unneceessary after using a hidden triple for example......then that naked double is not essential.....

Whether or not the naked double is essential depends upon the chosen objective, either
  • to find the sequence of simplest steps despite their number, or
  • to find the shortest sequence of steps despite their difficulty.
And when scoring the steps, I guess there could be two more objectives, either
  • to find the lowest possible score, or
  • to find the highest possible score.
It's your thread ... and you're running the tests, so you have the right to define the objective.:D
ronk
2012 Supporter
 
Posts: 4764
Joined: 02 November 2005
Location: Southeastern USA

Postby tarek » Thu Apr 20, 2006 12:32 pm

well I actually started the thread, I'm extremely flexible on how to rank these puzzles.......

Simple, Now what is simple.......Pat clearly spots the hidden double before the naked double.....

Shortest.......
Some people go searching for the difficult first to cut corners.......

Therfore Essential techniques can guarantee MINIMUM quality & MINIMUM difficulty level.

I've just run the puzzles through the modified solver.....

The puzzle that Pat mentiened #084........can also be solved using one hidden triple without the need for a hidden double........
So in terms of consistancy & insuring a minimum quality & minimum difficulty level it scores a 0.

The only way as a programmer to see what techniques occur every time is to have 5x4x3x2 runs for the solver per puzzle....each run has the order of techniques changed........analysis then would not just tell of what is essential or not, it can then give an idea of relative essentiality of technique.

returning to the 100 we have now.......... The highest scoring consistant puzzle is Ruud's #021, However & whatever your approach is, you WILL encounter the following (at least):1 Hidden triple,1 Naked triple/Hidden triple.......,1 X-wing, 1 Hidden double/Naked triple......

You see now, I made sure with only 5 runs of the solver what techniques would definitely be used in this superior.

tarek

[EDIT: small correction... the naked triple is a Hidden triple with a naked triple counterpart]
User avatar
tarek
 
Posts: 3762
Joined: 05 January 2006

Postby tarek » Fri Apr 21, 2006 10:37 am

After revising my program to try insuring minimum quality predict what techniques to be used...

I am posting yet another version of the rankings, I'm going to number versions, which I would advise other members to do if the want to post their rankings......

Code: Select all
Tarek's rankings of the 100 superior puzzles  Ver  4.0:
Rank   Compiler                Score Var  Dif
001   Puzzle#021 (Ruud)          13   4   639
002   Puzzle#090 (vidarino)      12   3   954
003   Puzzle#075 (vidarino)      12   3   952
004   Puzzle#074 (vidarino)      11   3   1046
005   Puzzle#067 (vidarino/ab)   11   3   777
006   Puzzle#078 (Ocean)         10   4   647
007   Puzzle#050 (gsf)           10   4   514
008   Puzzle#088 (vidarino)       9   2   803
009   Puzzle#066 (vidarino)       9   2   707
010   Puzzle#098 (Ocean)          9   2   663
011   Puzzle#013 (henk)           9   2   644
012   Puzzle#077 (Ocean/ab)       9   2   593
013   Puzzle#096 (Ocean)          9   2   516
014   Puzzle#097 (Ocean)          9   2   515
015   Puzzle#095 (Ocean)          9   2   496
016   Puzzle#091 (Ocean)          9   2   474
017   Puzzle#093 (Ocean)          9   2   463
018   Puzzle#020 (Ocean)          9   2   449
019   Puzzle#083 (Ocean)          9   2   445
020   Puzzle#099 (Ocean)          9   2   436
021   Puzzle#085 (Ocean/ab)       9   2   426
022   Puzzle#092 (Ocean)          9   2   425
023   Puzzle#082 (Ocean)          9   2   396
024   Puzzle#043 (vidarino)       9   2   371
025   Puzzle#058 (vidarino)       8   3   972
026   Puzzle#031 (henk)           8   3   869
027   Puzzle#032 (henk)           8   3   696
028   Puzzle#080 (Ocean)          8   3   458
029   Puzzle#070 (Red Ed)         8   2   543
030   Puzzle#007 (vidarino)       8   2   472
031   Puzzle#005 (ab)             8   2   469
032   Puzzle#045 (Red Ed)         8   2   439
033   Puzzle#051 (Ruud)           7   3   536
034   Puzzle#033 (vidarino)       6   2   849
035   Puzzle#057 (vidarino)       6   2   779
036   Puzzle#047 (Ocean)          6   2   627
037   Puzzle#079 (Ocean)          6   2   461
038   Puzzle#049 (gsf)            6   2   428
039   Puzzle#087 (Ocean)          6   2   388
040   Puzzle#081 (Ocean)          6   2   353
041   Puzzle#048 (gsf)            6   2   329
042   Puzzle#038 (ab)             6   2   318
043   Puzzle#076 (tarek)          6   1   347
044   Puzzle#089 (vidarino)       6   1   267
045   Puzzle#100 (Ruud)           5   2   733
046   Puzzle#061 (tarek)          5   2   665
047   Puzzle#073 (Ruud)           5   2   640
048   Puzzle#055 (Ruud)           5   2   537
049   Puzzle#064 (ab)             5   2   517
050   Puzzle#094 (Ocean)          5   2   514
051   Puzzle#016 (henk)           5   2   449
052   Puzzle#010 (Ruud)           5   2   449
053   Puzzle#060 (tarek)          5   2   435
054   Puzzle#071 (Ruud)           5   2   430
055   Puzzle#009 (Ruud)           5   2   428
056   Puzzle#002 (Ruud)           4   2   565
057   Puzzle#026 (henk)           4   2   565
058   Puzzle#008 (vidarino)       4   2   419
059   Puzzle#018 (ab)             4   2   328
060   Puzzle#065 (henk)           3   1   853
061   Puzzle#052 (vidarino)       3   1   767
062   Puzzle#056 (vidarino)       3   1   687
063   Puzzle#027 (henk)           3   1   669
064   Puzzle#053 (henk)           3   1   639
065   Puzzle#039 (Ruud)           3   1   632
066   Puzzle#072 (Ruud)           3   1   541
067   Puzzle#025 (henk/ab)        3   1   538
068   Puzzle#014 (henk)           3   1   487
069   Puzzle#054 (Ruud)           3   1   480
070   Puzzle#042 (vidarino)       3   1   454
071   Puzzle#044 (vidarino)       3   1   451
072   Puzzle#017 (ab)             3   1   445
073   Puzzle#028 (henk)           3   1   444
074   Puzzle#037 (JPF/tarek)      3   1   438
075   Puzzle#034 (vidarino)       3   1   433
076   Puzzle#068 (Red Ed)         3   1   430
077   Puzzle#062 (Ruud)           3   1   423
078   Puzzle#086 (ab)             3   1   348
079   Puzzle#015 (henk)           3   1   326
080   Puzzle#003 (tarek/ab)       3   1   325
081   Puzzle#011 (gsf)            3   1   310
082   Puzzle#059 (henk)           3   1   281
083   Puzzle#029 (vidarino)       3   1   264
084   Puzzle#023 (Ruud)           3   1   236
085   Puzzle#041 (tarek)          3   1   233
086   Puzzle#001 (tarek)          3   1   217
087   Puzzle#069 (Red Ed)         3   1   214
088   Puzzle#024 (Mage)           2   1   604
089   Puzzle#035 (tarek)          2   1   279
090   Puzzle#040 (ab)             2   1   261
091   Puzzle#012 (gsf)            2   1   228
092   Puzzle#046 (ab)             2   1   188
093   Puzzle#063 (ab)             0   0   547
094   Puzzle#019 (vidarino)       0   0   386
095   Puzzle#006 (vidarino)       0   0   326
096   Puzzle#036 (Ruud)           0   0   320
097   Puzzle#022 (Ruud)           0   0   315
098   Puzzle#004 (ab/tarek)       0   0   310
099   Puzzle#030 (vidarino)       0   0   254
100   Puzzle#084 (JPF/tarek)      0   0   107

Score: Score according to the formula mentioned before
Var:    Number of consistant batches of similar techniques
Dif:    Difficulty according to my normal solver



What it means ?

1.As you go up the list, you are guranteed to get more hidden triples & x-wings no matter how your technique in solving puzzles is
2. The puzzles scoring low are not bad puzzles, but are puzzles that do not display consistancy of techniques on changing solving approach.
3. Low difficulty scores combined with a low overall score means that for some puzzles....solving them can be easy if you were lucky to choose the correct technique (ex. Puzzle #084)

tarek
User avatar
tarek
 
Posts: 3762
Joined: 05 January 2006

Postby tarek » Mon Apr 24, 2006 9:01 am

Anyone interested in another thread, with more difficult puzzles:?:

tarek
User avatar
tarek
 
Posts: 3762
Joined: 05 January 2006

Postby Red Ed » Mon Apr 24, 2006 7:57 pm

Are you going to tell us what's happening with this lot first? Will they be published?
Red Ed
 
Posts: 633
Joined: 06 June 2005

Postby tarek » Mon Apr 24, 2006 8:28 pm

I've PM'd Papocom soon after we've reached the 100, he PM'd me back confirming receiving the 1st PM & that he would look into it soon...

That's really it:(

tarek
User avatar
tarek
 
Posts: 3762
Joined: 05 January 2006

PreviousNext

Return to General

cron